Oracle Corporation (ORCL) Earnings Call Transcript & Summary

Good morning, everyone. Thank you so much for being here today. And for those of you that were here earlier, thank you for coming as well. When you come to Oracle CloudWorld, you get a wonderful opportunity to not only see the technology, but you see customers. And I was really struck by Safra's keynote on Tuesday, where we had 7 customers come up and each of them told their story about how they were having success with Oracle in the cloud. And we do best when our customers do best. So it was a fantastic thing. The show has been great. Jason, I think, if I remember right, attendance up over 20%. So it's really big. You can feel the momentum when you go into sessions. And very happy to be here and have you all here. So let's talk about what we're going to do today. Let's see. We're going to start -- Jason is going to come up and just kind of give you a little bit of an overview of what's going on, things that are top of mind as it relates to CloudWorld and, more importantly, company and go-to market. Following that, Clay will come up and talk with you about Oracle cloud infrastructure. Juan will come up and talk about database. Steve will talk about cloud applications, including Fusion. Mike Sicilia will come up talking about Oracle industry clouds. Doug will then go to a financial update. At that point, that will take us right to about lunch. We'll take a break. And we'll be -- have an opportunity to get food, it will be over here. And we'll bring it back, and then we'll go into Q&A, with Safra first, and then Larry. And I'm going to bring Mike back up on stage to talk about Oracle Health. There's a lot of interest around this. It's a newer for us. We closed, what, it's been a little over a year now. And so Mike will come out and talk to you about that. My job though is really to talk to you about this. Yes. And so -- and seriously, though, we will be making statements and comments today. And so in the interest of time and efficiency and expediency, what I'm going to do is I'm going to go through that now, and then each presenter, as they come up, will make reference to, yes, what Ken was talking about. This is the slide that Ken was talking about. So there's 2 slides in some of the decks but most all -- all will have this safe harbor statement. So as a reminder, we will be making forward-looking statements today, and these statements are subject to risks and uncertainties. Many factors could affect these forward-looking statements, and they can cause actual results to differ materially from the forward-looking statements being made today. A detailed discussion of these factors and other risks is contained in our filings with the SEC, including Forms 10-K and 10-Q. All the information in this presentation is current as of today, September 21. And lastly, we don't undertake any obligation to update the information in light of future events. Okay? That's our safe harbor statement. The other one is, and this will be Doug's deck, in that we do make use of non-GAAP information. And so you'll hear some non-GAAP numbers being used in our presentations today. If you didn't get a chance to be here Tuesday and Wednesday, there were some fantastic keynotes that a lot of things you're going to hear today, you can hear in more detail. And actually, I would, in particular, point to Safra and Larry, but you'll be able to find those, if you go to oracle.com/cloudworld, you'll find a link and you can sign in and view all the keynotes. In addition to Safra and Larry, all these key presenters that will be presenting here today, you can see their keynotes and their presentations from the actual customer event. So with that, I'm going to turn this over to Jason Maynard, our Executive Vice President of Revenue Operations.

As we've transitioned to the cloud. We have made incredible progress as we've consolidated customer success. We've brought out new capabilities. If you visited the hub or the trade show floor over the last couple of days, you'd see the customer success circuit. And one of the great things that we're doing is, when we talk to our customers, they want to know how do we do it? Tell us more about not just the technology, but what's the change that you have to make from a process standpoint? What's the change that you have to make with your people? How do you do it? So we launched a thing called the Oracle Playbook, which is, think of it as a how-to guide on how we implemented Oracle technology to run our business, and then sharing that with our customers. So take a look at that. Some great advice, really good learnings. And I'm going to have a homework assignment for all of you at the end of this as well about the Oracle Playbook. So customer success has been a big part of our transformation. You saw it on stage in many of the keynotes with great stories about how our customers are transitioning to the cloud and, ultimately, how we're on that journey with them, not just from the first implementation, but as they go live, optimize and continue to expand their footprint with Oracle. So great work there. The second thing is we have increasingly been expanding our ecosystem across a number of different areas. First off, on the technology side. You're going to hear a lot more about it in -- with some of the other keynotes, but the relationship with Microsoft around multi-cloud opens up a tremendous amount of new opportunities for Oracle to bring the Oracle database services on OCI into the Azure Cloud. The commercial opportunity there is significant. We'll get into that later in terms of where it goes. But we're at the early stages of this relationship with Microsoft that's expanded from our initial agreement a couple of years ago. Second thing, you might have heard a little bit about AI over the last couple of weeks. And there's been a little bit of AI talk the last few days. Our work with NVIDIA and Cohere continues. I'm going to give you a little bit of clues later about how some of the things we're doing to go to market with them and really take advantage of the AI opportunity. But again, a lot of openness in terms of working with multiple partners. The other group that I want to call out, because it is really important to customer success, has been our work with the strategic integrators. We work with both GSI and regional partners. More and more the integrators are participating within our implementations. It's a big part of how we go to market. And across the board, we're getting engaged earlier, we're bringing partners in, and partners are also starting to bring us in. And so we're seeing a greater focus on how we not just have them implement but actually how we go to market and we align in the field. So a lot of work there. The third area is this concept of end-to-end industry automation. Right? Our customers, in many ways, are also our partners. So if you were here for Safra's keynote, you might have seen Uber's CEO on stage with Safra. Yes, Uber is moving their workloads to the Oracle Cloud, but we're also working with Uber Freight, Uber Direct, integrating Uber into our retail solutions, into all sorts of different capabilities to deliver a whole product solution for specific industries. So great work there. In Steve Miranda's keynote, and you'll hear from Steve, we're doing incredibly interesting work in B2B. So partnering with FedEx and JPMorgan a year ago, expanding it to Mastercard, HSBC, and really bringing together a whole ecosystem around the ERP market. Again, a really interesting area. The last one I want to mention is in communications. And Clay, I'm sure will talk about Alloy and the capabilities there. But we're helping our communications customers transform from not just being, if you will, basic providers, but leveraging OCI and actually delivering capabilities on top of Oracle Cloud themselves. So across the board in almost every industry, customers are partners, and we're enabling new ways to go to market that solve complete problems across the industry. I think it's a really powerful thing. Last piece I want to talk about is actually how we do this. How do we actually translate all this innovation, the partnerships, into our go-to-market? Three things to focus on. One is sales and marketing is unified and aligned around these core sales plays. Whether it's applications, infrastructure or industries, we have separate capabilities for each industry, we leverage all of our product line, our product suite, and we go to market universally. What that means is we have a playbook that we run, and I think of it as an Oracle playbook for go-to-market, and how we scale this out. So whether you're in North America or Europe or whatever region, there's a set of commonality around how we work together, and frankly, how the teams in sales collaborate to take advantage. And so some of the changes we've been making have streamlined our sales and marketing, but also help us drive greater growth. And so you'll hear some of those product capabilities and that dovetails exactly into how we translate that into value for customers. The last piece of this is operationalizing it and scaling it, right? I mentioned NVIDIA earlier. So NVIDIA, we have a great technology relationship. What's happening now is we're turning it into a unified and consistent go-to-market with NVIDIA. So over the next few months, you're going to see Oracle, NVIDIA go-to-market around data and AI. You're going to see a broad range of things we're doing in the field together, aligning with their teams to really go in and help solve those customer problems and talk to them about how they move those workloads to Oracle. Where this shows up is really kind of 4 areas. First one is we want to win that workload. That workload can be AI, can be ML. There were lines out the door for once, database, Oracle 23c sessions, a lot of interest there about how we win those workloads. It's not just enterprise workloads, but it's increasingly with ISVs as well. So that's one motion that we have, operationalizing out in the field. The second one is the line of business, selling to the functional buyer, whether it's CFO, HR, CIO, CTO. Those are discrete sales plays that are scaled across the globe. And increasingly, what we're seeing is those customers start then to think about the multiproduct buy. And that's the third thing, is buying the suite. We have the capability not only to deliver a suite for applications, but how do we bring our various groups together. One area that you'll see more of is how we bring health care and ERP together, right? So some of the power of the things we're doing with those 2 product lines. That's starting to show up in the field in our go-to market. And then last but not least is this notion of industry. We have the capability to deliver end-to-end industry automation solutions. Health care is obviously one we're going to talk a lot about today, but it's other industries as well: financial services, communications, retail. All across the board, we're taking a look by industry and how we take our discrete motions, bring the entirety of the Oracle product portfolio together. Our teams are collaborating, delivering value to customers, getting them live, and delivering ongoing customer success services. So across the board, tons of product innovation this week. I hope you feel the commitment to customer success because it's something that we do every day with our partners, and it matters. The openness of the ecosystem, the alliances, technology, integration, and really our customer partners, all bringing together to solve problems. And then ultimately, we're translating this into revenue, through this unified go-to-market, bringing the portfolio together, the right product, the right time and the right region for our customers. Now I did say there was one more thing before I get off stage that I had to ask you, which is, in this room, many of you work in financial services. You can all benefit from the power of the Oracle Cloud here. So I know some of you are very influential in your organization. So take the homework back, the message back, bring it to your CIOs, your line of business. Oracle Cloud, we're open for business. We have plenty of inventory. Come to Oracle for your financial services' needs. You can call me. I know many of you, you know how to get a hold of me. So we'd love to do business with you, we'd love to partner with you. And with that, I'm going to hand it off to our EVP of Oracle Cloud, Mr. Clay Magouyrk. Clay?

This is the slide that Ken told you about earlier? Did I do okay job, Ken? Okay. Great. So rather than just talk today about a bunch of individual products or improvements or different services that we're adding, what I'm going to do is kind of make sure we're all aligned on the base of where OCI is at today, cover it as a whole, and then break the business down in individual sections. So to start with, all of these numbers are -- they're not new because I wouldn't be allowed to do that. These are numbers that we talked about in our Q1 earnings call. So to just kind of baseline, people, $1.5 billion of infrastructure revenue, consumption up 91% year-over-year. And if you look at it in the aggregate, right, if I had only 1 minute to explain to people why we're seeing that growth, it's because we're continuing to add more services. We're continuing to add more regions around the world. And for certain parts of our business, we're getting pre-commits based on future delivery, right? That's what these big AI contracts are doing. So as a whole, the business is growing quite strongly. Let's kind of dive deeper into what's causing that individual growth. And the way I've structured this presentation is, I talk about our business in several sections. I'm doing them from the order in which we've actually been addressing them at OCI, not necessarily based on growth rate or size or any other characteristic, and you'll see that as we build a time line. So the part of the business that I think is obvious to most people, that's coming to OCI, is our traditional enterprise customers. We've been a long-standing applications and infrastructure provider on-premise. We have an incredible applications and industries business in our cloud. And those customers end up choosing our platform because we're the best place to run Oracle workloads, we're the best place to run your mission-critical workloads. We have an extreme focus on security, reliability, availability, disaster recovery capabilities. We have a global footprint, as you just saw, up to now 64 customer-facing regions around the world. So we could talk a lot about this segment of customers. But in reality, this is kind of the thing that we've been working on the longest. It's clear to people why, right? You'll hear from Juan in a second about all the great innovations that we're making across the Oracle database. Those things work best in our cloud. And that's the reason that many of these customers choose our platform. But to give you a bit more color, one of the things that I think isn't always clear is if you take -- as I show each segment, when I talk about, say, 37% year-over-year growth for these -- for this set of customers, it's this segment, right? It's not across the whole business. For these types of customers, right, if we didn't add a single customer year-over-year, if I look at last year and I say, how much did that -- those customers grow on our cloud in the past 12 months? They grew by 37% without us adding a single new customer. So the way that we continue to grow this segment of our business is well, we've got a lot of great salespeople, we've got a lot of great partnerships, we're out there every day talking to new customers, that have been longtime Oracle customers but haven't moved to our cloud yet, and we're getting them onboarded. But even the customers that are already on our cloud continue to show significant year-over-year growth. And that's through a combination of our field engineers, helping them move more workloads. It's about the fact that their businesses are growing and scaling. And when they do that, it grows on our cloud. As well as our partnership ecosystem that Jason talked about as we work with some of our biggest and best partners like Deloitte, Accenture, et cetera. They help those customers move more and more workloads to our platform. So the way in which we continue to expand this is obviously go after more of those customers, help them when they're on the platform to expand. But also it's continuing to expand our ecosystem. Right? So just over the past couple of months, you've seen announcements from us around things like our Amdocs support, OpenShift being certified on OCI, our enhanced collaboration with VMware and much more. And so as we continue to build this ecosystem, more and more of those workloads move to OCI. And just for reference here is a time line at the bottom, right? OCI launched in 2016. This was our primary focus. This is our -- when we're going out with our MVP, these were the customers we were going after. And so we've been working on this for quite some years now. And we're continuing to see this ongoing growth. A piece that Jason touched on is our multi-cloud customers. So last week, Larry and Satya shared an extension to our existing relationship. For those of you who aren't well educated in our multi-cloud relationship with Microsoft, we started in 2019 with our first interconnect region. As of today, we have 12 interconnect regions around the world. We have almost 500 customers, and the growth of those customers is continuing to accelerate. These customers choose us because we have highly differentiated services. When you listen to Juan talk about all of the amazing things that are happening in Oracle Database, autonomous database, what we're doing with Exadata, right? When you listen to what Edward said yesterday in my keynote about all of the great things that we're doing with MySQL HeatWave, that's technology that is highly differentiated, and customers were telling us, we love that technology but we want to be able to use it in other places. And so with our newest announcement of Oracle Database at Azure, what we've done is we've made that even easier to use across Microsoft. And so these customers who are longtime Oracle customers, some of them are new MySQL HeatWave customers, as an example, they pick our cloud and they want us to work together with other platforms, and we're seeing a lot of growth in that area. Yes. This is just kind of a quick overview of the different types of services that we've offered across our multi-cloud relationships over the years. The big thing that I think it's important for people to understand, beyond just the technical, right, with our most recent Oracle Database at Azure announcement, more than just the fact that, yes, we have lower latency, we have a more integrated experience, the thing that is also, I think, opening up the biggest opportunity is that there's a collaboration now because, by Azure having Oracle database services in their marketplace, by us being able to enable private offers, customers who have existing commitments to Azure can use those commitment dollars to consume Oracle Database services. That's the big thing that unlocks. So suddenly, you have a massively larger number of people that are selling to an enhanced number of customers with existing commitments to consume these services. So how do we continue to accelerate this segment of our business? Well, part of it is that we have to first make this successful, right? We have a certain number of interconnect regions. We're building this new technology. We roll it out, we get reference customers, we continue to expand. It's somewhat early days. And in addition to that, over time, you'll continue to see us do more and more with different cloud providers in different ways. It's an important part of our overall strategy. So another big segment of our business that's growing very quickly is what we call our cloud-native business. These customers choose us because of extremely high performance, our availability and our highly secure overall architecture. The thing that makes these customers special is that they don't just show up and take a little bit of capacity. They tend to be very large scale. They have high demands when it comes to elasticity. And the price that they're paying for their infrastructure has a significant impact on their bottom line. When I talk to people and they ask, why are they choosing you versus someone else? In addition to the fact that we're incredible at actually operating and running these businesses for them, you can take a simple example workload like this, and you can go and you can compare across the different cloud providers, what does this cost you over a total cost of ownership for several years? And what you see is that, list price to list price, there's a massive price difference between what we offer at Oracle and what our competitors offer. And even still, if you go to our competitors and you say, "I'll pay you upfront for the next 3 years," committed completely, we're still 20% to 30% to 40% cheaper than our competitors. Now for some customers who have a small footprint, maybe that doesn't matter, right? As Safra has told me is that 20% of $1 is $0.20, it's not 20%, right? But 20% of $100 million is $20 million. And so the reason that this segment of customers is choosing our platform is because the value of these economics and the fact that we run their infrastructure so incredibly well is very valuable to them. Now how do we continue to grow this? Part of it is, when we engage with these customers, we have a great relationship, and we work together to really move these most demanding workloads to our cloud, and we -- as you heard from some of our other -- [ Maddy ] was on stage yesterday in my keynote talking about how the lines between his engineering organization and ours blur. That's a big part of our strategy. We also offer, because of our history with -- through our Sun acquisition, our great Oracle hardware design team and our optimized supply chain, we actually can go in and offer the flexibility of customizing different hardware shapes and properties for these types of customers, which makes a massive difference. Rather than just saying, "Here's what's on the truck, take it," we can go in because the volume is enough, optimize it for them and then still offer that in our unified cloud experience with the same integrated security and performance, which makes another big difference to those types of customers. And then, of course, for these types of customers, it's not enough that it's cheap, it has to work and it has to be available when they need it. And so we've done a lot of work across the years to really optimize our supply chain, from the initial design all the way through it being available in the data center for the customer, and that's really important for this type of customer base. As Jason mentioned, AI is the talk of the day, talk of the month, the quarter, the year, really. I have 2 sections about AI. For this segment of our customer base, I'm really talking about customers that are in the business of training and doing large-scale inferencing on their own models or retraining of other models. These customers, what they need is extremely large amounts of extremely high performance compute capacity and networking. It sounds a little bit similar to our last section, but even more specialized. They're choosing our platform because of the investments that we've made across the stack to be the best platform for training their AI workloads. Well, how do we do that? One of those pieces of technology is our OCI Super Clusters. We've been in the clustering business for a long time. right? The reality is, if you go and ask -- if someone were asking me even a decade ago, what's the biggest difference between Oracle database and other databases? Fundamentally, Oracle database is a cluster database. Because of the investments that Oracle made for decades in the rack, we've been working on clustering technologies. And so when we built OCI, right, because I'll tell you one thing about Juan. He's not an easy customer, he's a demanding man, he knows what he wants, and he's going to get it. And he made sure that we built a great clustering technology into OCI's fabric. Well, as we kind of expanded our business, we adopted -- brought up HPC customers over time. But when this AI inflection point hit, we already had the investments into our RDMA network, we already had the investments into our clustering abilities. So we just had to scale that up and obviously optimize it for AI workloads. But we spent a lot of energy tuning this, but we were starting from a higher position because of our previous work in the clustering technology. So the key here why customers are choosing us is because of the performance and the overall availability and reliability of this platform. But we're not content to stop there. As you can see, this segment of the business is evolving very rapidly. So for us to be able to be a credible platform, not just now but in the future, what we have to do is make sure that we're staying on the cutting edge. So we're constantly evolving our network design, building bigger networks, faster clusters. We're working with critical suppliers like NVIDIA and AMD, getting early access to their hardware, putting that available into our cloud so that customers can test out these clusters very early, doing both training focused type new hardware with things like the H100, H100 Next, H200. Also working on inferencing solutions with things like Grace Hopper, or working with AMD on some of their GPU offerings. Because what we see is the demand for this only continues to increase. And the efficiency and performance of this matter so much, we have to stay on the cutting edge. Next, with Alloy and our dedicated region business, it's been something we've worked on for quite some time. I'll take a moment to just explain to you how we got into this business and why it's so important to us. From the very beginning at Oracle, we knew we had to build off a lot of regions, and we had to build them fast. And we also knew that we wanted to have different types of regions. We wanted to be able to scale up to a very large size. But based on what our customers told us, especially those traditional enterprise customers I talked about first, they needed in-country disaster recovery and they needed it before they could even move their first workload. So what we did is we have an architectural difference such that we can actually start very small with our cloud regions and then scale them up. Over time, that enabled us to go, well, we could actually deliver this for individual customers. And so you have customers like Vodafone or NRI that have purchased multiple dedicated regions from us. And they use that to really optimize their cloud workloads right next to their on-premise platform. And then as we continued in this business, we had different relationships with some of our partners, where what they wanted to do was not just to consume these regions for their own workloads, they wanted to be able to combine that great technology that we offer at Oracle, they wanted to combine that with customizations they have. Different platforms they've built, managed services, integrations. And offer that to end customers. And in the beginning, we didn't have that business. We didn't have that business model. We didn't know how to do it. But over the years, we worked with them, we developed it. And that's what we call Oracle Alloy. And fundamentally, it's about the complete set of functionality needed for someone else to be a cloud provider. Thankfully, Steve Miranda and Mike Sicilia have done an amazing job of giving us an applications platform. So as an example, Alloy is not just OCI, it's also fundamentally about Fusion. Because think about it, if you're going to offer a cloud to somebody, you don't just need to have some infrastructure. You need to have a billing system. You need to have a service system so you can do service requests. And so the fact that we're both an infrastructure and an application provider allows us to offer a pre-integrated set of functionality. So someone can purchase Alloy, they get Fusion, they can do billings to their customers, handle their service requests, and more. And so we have a lot of excitement and interest in this type of our business. Just while I was here, I met with a customer, I won't name them, but we are now in the process of how can we finish the contracts as quickly as possible in the next few weeks. Those are the kinds of conversations I'm having with our partners about Oracle Alloy. So how do we go about growing this business? Fundamentally, it's about continuing to take that integration of our cloud, make it easier, and then working with our partners to actually enabling grow that business. We're still early days. But as we're building the references, right, because there's a certain amount of propagation delay for this type of a business. As we grow it, we're seeing more and more excitement. And I think you'll see a lot of traction from this segment of our business going forward. The last thing I'll talk about is our generative AI and how we see this as a new emerging segment. Obviously, you heard from us talk about our new generative AI -- our new generative AI service. We have an amazing partnership with Cohere, which is doing great things with their large language models. We're working together to take things like retrieval augmented generation to be able to make this really easy for our customers to use. Given the newness of this segment, obviously, the growth is still nascent, but there's an immense amount of demand. And so the fact that we have this great partnership, the fact that we operated all the layers of the stack, we have our infrastructure, our great GPU and networking infrastructure, the fact that we now have this generative AI service, the fact that Mike and Juan and Steve are all pre-integrating this generative AI functionality into their services, what we see happening over the next few years is we launched the service, it's available in all of our distributed cloud regions, and suddenly, customers are doing fun new things with this technology. The other thing to understand is this technology also is an accelerant for the adoption to the cloud, right? So one of the things that this technology does, above and beyond the individual success in revenue and consumption you get from those services, is that if someone comes to you and says, "Hey, I have an application, I have a workload. I really need to get access to this technology," that technology is only available on the cloud. And that becomes a conversation that goes, great, let's help you move this to the cloud, right, and get access to this new technology. Now I'm not the greatest at drawing boxes, as you can see. But the thing that I'm trying to show you with the stair step, because while the size of the boxes is not relevant to anything, what does matter is you can see when these segments of the business has started for us. And I don't know if you've ever seen stairs, but if you've ever walked up stairs, this would not be a good staircase, right? You'd be like these stairs are uneven, and I'm tripping as I -- it's like, why do they start off very far apart and they get closer together? The reality is, the reason you're continuing to see this acceleration in our business, and I think you don't quite understand where it's going to go, is because it's not one business, right? Our traditional business is nowhere near saturation. Our multi-cloud business is just now starting in many ways with our most recent partnerships. Our cloud native, we only started this 2, 2.5 years ago with many of these customers. And they're continuing to grow, while at the same time we're adding new segments of our business to OCI, right? That's what's driving these numbers. And I think part of the reason why I wanted to highlight this to you today is because you can -- without thinking about it, you can say, well, what is Oracle Cloud? What is OCI, right? It's FedEx. Yes, it's FedEx, or it's DHL. But it's also ByteDance and Zoom and Uber. Right? It's also NVIDIA and Cohere and RECA and Mosaic. It's also NRI and Telecom Italia. It's all of those things together. And the reason that our business continues to be so strong is because we are continuing to add more segments that are growing faster and faster across the portfolio, and they're enabled and continue to be strengthened by the great technology of the other parts of our business, right? Alloy's made stronger by our apps and industries business, right? Our multi-cloud would not be nearly as good if we didn't have the most amazing database technology in the world. So it's that combination together that then ends up resulting in this massive OCI growth. So hopefully, that was helpful. I've enjoyed talking to you about the world's worst staircase. And up next is Juan Loaiza.

All right. Do I get the safe harbor? Yes. Don't sue me, please. So yes. I think you all speed-read this. It's interesting, just listening to Clay, I've been in Oracle for a long time. I think completing my 35th year here working on database. So I'm kind of an old guard technology guy. But what I feel like is OCI is on the map. I mean the last really couple of years, before that, people didn't really believe we could do it. And now in all the customer meetings I have, they're like, this is for real. I don't know if -- 2 years ago, I wasn't sure it was for real. But this OCI thing, it's for real and it's happening. And my staff is interested in it, the AI stuff, he said. So this is really happening. It's interesting. It's fun for me to see. And actually, the other thing I see is every one of the customers that I see, and you all are a lot of this, they're also running Fusion apps. So again, if you go back 5, 6, 7 years, it was mostly Business Suite, SAP. Now everybody is on Fusion app. So it's really picked up. It's interesting. So what I'm going to do is I'm going to take you through a quick journey of what we do in Oracle Database and where we're going and some of the big innovations that we're driving. So this is kind of what we're about. At a big picture level, we have 2 things that we focus on, right? This is the very big picture level. One is our -- what we call converged database, which is our Oracle database that runs everything. So our goal is we're going to run all kinds of data types, all kinds of workloads, everything you want. We do it seamlessly. We're best of breed at all of it. And we scale, we're available. There is, I don't want to say no financial institution. I think I can't say that. I think I can say there's nobody in the world that doesn't run our stuff. Because there used to be a few mainframe holdouts. And I think they're gone now. So your entire business runs on our -- on that technology. And it's because it's really, really good. The other thing that we focused on is our cloud products. We have our autonomous database. So we've taken all that technology, which used to be complicated, it used to be like flying an airplane. You have to get trained, there's a million switches and you got to know what to do. And we've made it autonomous. So we made it very easy to deploy. And one of the things that is underappreciated about this is the same technology that runs the most critical stock exchanges, banks, medical systems, telecoms, is available to anybody now. So even a small little database can get stock exchange level performance, availability, security. So that's kind of a sea change. It used to be little guys couldn't get this kind of stuff. It was too complicated. It's like you had to fly a jumbo jet to get it. Well, you don't have to do that anymore. It comes straight out of the box with autonomous database. All right. So that's kind of a big picture. If you compare what we do, we say, "Hey, we'll run everything in one." So you can focus on innovation. You can just use the product and focus on innovation. Other people say, no, you need 8 different databases to support an application, and then I'm going to move the data around, I'm going to deal with all this stuff, and it's a mess, and it's very bad for productivity. Okay. Another thing that we're very far ahead of the game on is our Exadata, which is our database platform. Clay talked about clusters and RDMA. We re-architected for this about 15 years ago. And the other part of this is smart storage. This is one of the big reasons why we run all the mission-critical systems of the world. Nobody else has done this. Nobody else has done the RDMA clusters. Nobody else has done it. Nobody else has done the smart storage. 15 years later, it's very unusual to have a 15-year lead in technology. I don't know. I don't know if it's more than usual. It's unheard of. Have you heard of a 15-year lead? And you see the success. It's not like it's some niche thing. It's like it runs everything. Well, nobody else has the RDMA, the clustering, the smart storage. And again, that's why we're very successful. And you can see it in the numbers. Most of the world's large businesses, lots of others run for everything, for warehouses, for the most mission-critical OTP, all the packaged applications, the big SAP systems run on Oracle Exadata. And we're getting a lot of them in the cloud now, consolidation, everything. Okay. Another big thing that's underappreciated is our mission-critical technology runs everywhere. So you can get it on premises. You can get it in -- you can get our cloud. We will run our Exadata stuff in your data center. And many banks are doing this already, especially in Europe, it's interesting. And Europe is way ahead of the United States. Normally, United States is way ahead of Europe. Europe is actually ahead of the United States on some of this. So we will take our cloud and put it in the customer's data center, either as cloud customer, as Exadata cloud customer, or as a full region inside the customer's data center. No one else does that. So we go to where the customer is. And what's not obvious is people like Amazon have something that they'll run in your customer. But it doesn't run the important stuff. Like all their database workloads don't run. So you really can't run anything interesting on it. So we are the only people that can run mission-critical workloads in the cloud but in the customer's data center. No one else has that. And it all works the same. And then Clay mentioned this. This Azure thing is a big deal. Again, we run everywhere. We go to where you are. There's a lot of customers that have committed tens, hundreds of millions dollars to Azure, and they're struggling because they can't get their database to work. They're saying, "Oh, yes, maybe we'll switch databases." Nobody's ever successfully switched the mission-critical database, so, because our stuff is so far ahead. So this opens up a big opportunity for us. The ability to use the funds that have been committed to Azure to purchase our cloud and our database is a huge deal. I think this is going to make a big difference in financials. Okay. So what are we doing? So that's kind of where we are. So that's where we are. I'm going to talk a little bit about what we're doing. In particular, there's 2 things. So we're just releasing our next-generation Oracle Database. So this is something we've been working on for about 4 years. It does -- it enhances everything we do, availability, security. We have 300 features. There's a lot of work that we've done. But there's really 2 key things that we focused on. One is developers and the other one is AI. And what we've done here is we've tried to make really leapfrog improvements in these areas, to leap over what the competition has. So I'm going to talk a little bit about those 2 things: The developers and the AI. All right. So there's a lot of technology in database that we've really -- rethought, and I'm going to walk through a little bit of this. But JSON, operational property graphs, micro services, caching, getting all the right languages, SQL simplification. So I'm going to walk through a little bit of this. But the key message is we've really focused on developers. So Oracle has historically been very mission-critical, and actually, we've worked a lot with developers. But developers are increasingly important in the market. So we said we got to jump ahead of everyone. So like what we did with Exadata, we're like, we're going to jump way ahead of the rest of the industry. And that's how we're going to win the market. And so we're doing the same thing on developers and AI. So let me walk you a little bit through this. This is very quick. If you watch the keynote, it's recorded, it goes through this in actually a lot more detail, but I don't have enough time to walk through all that. So I'm going to kind of tell you at a conceptual level what we're doing. Okay. So the relational model is where Oracle started, right? We won the relational game and relational became the leader in the database market. And that's where we are -- where we are today. And relational has a huge amount of advantages. Particularly, it keeps your data consistent, right? It keeps data integrity. If you're a bank, if you're a retailer, if you're an airline, you care about data integrity. You can't like lose data or have data get messy. And what's happened over the years, particularly, I think, over the last 5, 6, 7 years, developers have become more prominent. So data professionals love relational databases, because they're extremely powerful. They have the SQL, this declarative language, they keep the data secure and integrated. However, there's a new breed of developers that comes around and says, "Hey, this relational thing, great for you guys in the data world. It's not so much fun to develop against. It's easier for me to develop against the different model," right? And so they're using what's called the document model with JSON. So they're writing apps for that, because they're like, "Hey, this is easier for me as a developer." They're writing these things called Graphs because it's easier for them, for developers. Now these -- that creates problems in the data management layer, but they don't care. They're like, hey, somebody else's problem. It gets messy. That's why, this is one of my rants, I'm not going to go into the whole rant, but so many times I try to get an airline ticket, and there's like crazy ship, like, oh, what happened in my seat assignment. It's gone. Oh, I can't check in because I get some weird error about some data inconsistency. I mean this stuff is happening all over the place, and it drives me crazy. As a guy who's been working on, data is important, like get all these weird errors from all these commercial systems as an end user and I have to call the airline and be on the phone for 2 hours with customer support. I'm like this is crazy ****. But anyway, so developers are doing this stuff because it makes their life easier. So this is kind of what's going on. So we're having this battle between, hey, make it easy for developers and keep the data integrity whole. And so this is what's been going on, and it actually dates back. So what's happened is the world of data has split into pieces. It's kind of like the world has kind of been broken up into pieces. So there's JSON stuff, which developers like with JSON databases and JSON storage, and there's Graph stuff with great Graph databases and Graph storage. So that's what's kind of caused the schism in the data world. So this is something which is fundamental to data technology, which is we figured out how to reunify these worlds. Okay? And it's something we call JSON Relational Duality. But the idea is you store your data in normalized format where the data is consistent, but you let developers access it as if it was all JSON or Graph. That's the fundamental idea. And this is a big deal. This is kind of the grand unification of all these things. And by the way, this schism dates back 30 or 40 years. It used to be called hierarchical databases and network databases are now it's Document databases and Graph databases. So the big picture is here, we're jumping ahead of everyone. We're saying, hey, these worlds are splitting apart. And when you split these worlds apart, it creates a giant myth. And we can put this back together by supporting all -- everything that the developers want with all the data integrity that the data professionals like. And this is a very big deal. This is going to be adopted throughout the industry. This is going to become the new model for data in the world. And this is something that we've embedded and we're releasing in Database 23c. All right. So that's one thing. And this is just -- if you read some of this stuff, if you talk to people that have been in the data management world, you can go talk to any of these guys, and they'll tell you this is the biggest change that's happened in the data -- the world of data in decades. It really is. It's a big deal. Okay. So we're jumping ahead of everybody else. All right. There's another big thing that's on the horizon, right, which is AI. Okay? So I'm going to walk you briefly through what this does and how it works with databases and what we're doing in the world of database and AI. So first of all, there's a concept of an AI vector. So this is kind of a new thing that's come up on the world in the last couple of years. And what an AI vector does is it stores the semantics of some unstructured data. So I'm going to give you an -- I think I went ahead. No, no, I'm looking at the wrong screen, sorry. So here's an example. So you can take an image or a document and the AI basically embeds that, converts that into a sequence of numbers called the vector. So you get something like 1,000 numbers. So an image is represented by 1,000 numbers. And the important part there is it's not the pixels, it's what's in the image. So for example, if I take a picture of you, your vector will encode you. And normally, if you look at pixels, if you had your glasses on and off, not the same guy; when you're closer, different, not the same guy; when you're looking to the side, not the same guy. So the beauty of the AI vectors is they encode the actual content. So if I take a picture of you now and I take a picture of you tomorrow looking to the side with different clothes, he says, it's the same guy, even though no pixels are the same. It figures it out. It figures out the content. And the same thing with documents, you take a document, it encodes the meaning of the document, the semantic content of the document. So it's not the words, like it would know this document is about AI vector search. It's not about Fusion apps, something else, right? So when you compare these vectors, it encodes the content. So this is, I think, the real magic of AI these days. You can take content like a picture or a document and turn it into a series of numbers, like 1,000 numbers. And then you can put those numbers in a database. It's called a vector. So yes. So the way these vectors work is things that are similar, the vectors are close. The numbers are similar to each other. And things that aren't similar, if I take a picture of you, take a picture of you, your vectors are going to be farther apart. That's basically the fundamental concept. Okay. Now that AI technology has been invented. And what we're doing is we're combining that with business data. So we're taking the semantic vectors, and we're combining it with business data to generate business value, okay? That's what we're doing at Oracle. And if you want to get business value, it's not enough to have some semantic content. You have to combine it with the business data. And business data, it's things like end-user data, their buying history, their interest, their balance, the product data. What is the product? What are the features? What are the limitations? How much inventory do you have? Doesn't really do you any good to have a semantic search without this other data. So it's the combination of these 2 that's really going to enable AI in to bring value to businesses, okay? And that's what we're really working on. So I'm going to show you a simple example of something you can build. So imagine you have a house hunting, you're looking for a house. And you have an app. So you drive and you're like, "Hey, I like to look at that house. I'm going to take a picture of it." And what the app is going to do is it's going to find a house that looks like that, that's for sale. So I say, I like that house, find me one that looks like this that's for sale, right? So there's 3 parts to this, right? There's find me a house that looks like this. That's the AI vector stuff. That's that similarity, the content search. There's another part to that, which is I need to know houses for sale and I only care about houses in my city. I don't care if there's a house for sale 5,000 miles away that looks like this house. And it has to be in my budget. So there's product data, what house, is it for sale? What are their costs? There's customer data, which is where am I looking? What's my budget? And then there's the similarity. So you have to combine all 3 of those things in order to actually get something useful for a business. And so there's kind of 2 things you can do. If you want to combine this data, you can either take all your business data and move it into the vector database. But that obviously has a lot of problems. It wasn't really designed for business data. You have to ship all your business data over there all the time. Or you can basically do what we're doing, which is you can add vectors to the Oracle database. So that's what we've done. We've added vector directly into the Oracle database. So a customer can search their business data and their AI vector data all very simply in one solution. So that's kind of the idea. So we've announced this here -- here at CloudWorld. There's a demo at our tech hub, where we're demoing all this technology together, and we're signing up customers for the preview version. All right. So this is actually a SQL statement, so I'm going to walk you through this. So here's what we're trying to do. Somebody said, find houses that looks like this house, but they have to be -- match the customer's budget and be in the customer city. So that's the 3 things: the product data, the customer data and the vector data that have to come together. Now this is, again, we're all getting our mind around this whole AI thing. If you go back a year, AI was really complicated. We have AI in our database. Actually, we've had AI in our database for 20 years. But you have to be a data scientist. You have to have a degree in data science. You have to know all the different algorithms. You have to do -- know how to do AI training. You have to do -- know how to do validation. You have to know how to set hyper parameters. It was -- it's kind of a rocket science little area, that was a niche. I mean, we've had it for 20 years. People have used it to optimize their systems. The new AI, as of last year, is super simple. So this SQL statement, I can walk you through it. It says, select the house from the houses for sale where the price of the house is less than my budget. And the city that the house is in the city I'm searching for. And in this last part says, order them by the similarity to this picture. Okay? So that's 5 lines of SQL that any SQL developer can learn how to write in 10 minutes. So it used to be you had to have a PhD in data science to do AI. Now we've integrated this whole thing into the database. All the data is completely consistent, the business data, the vector data, and it's incredibly simple. I mean you can read that thing. It's incredibly simple. You don't need -- you need to know -- I mean literally in 15 minutes, you can be doing this thing. So anyone who's a developer who's a DVA, the only new thing here is this, compare the vectors and find me the one that's most similar. That's the only thing. So this is a big sea change in the whole AI world. And I compare this a little bit to ChatGPT also because -- how many -- do you have to be a PhD to use ChatGPT? Anybody ever tried it? Are any of you PhDs in data science? So the world of AI has kind of radically changed, and it's radically changed on the data side also. And you can see -- you can see that this has never happened before. It's easier to search, this AI search, than to do a text search, to do -- search for text and that, because you have to know about synonyms and this and that, and there's all sorts of different terms. It's actually much simpler. So this is a big deal, and we've built the whole thing into our database. Okay. What else do I have here? And yes, so I'm going to quickly go through a lot of the technology that we've built for these AI vectors. We're building these specialized vector indexes. We can partition up by city to automatically make it much faster. You only search the city that you're interested in. You can run this in our scale-out clusters, the stuff that you all are running your banks on today. It automatically works, it scales out across everything. You can isolate your vector workload from your other workloads. You can use our intelligent storage that we -- again, the next data we built to accelerate all these AI searches. And by the way, all this stuff, I'm saying, this -- nobody else does this stuff. You can't run intelligent searches. They don't have the clusters. And there's lots of other stuff. The stuff that we've been working on for 40 years at Oracle, Parallel SQL, transactions, analytics, disaster recovery, security, you get it all, it just works. It just works. So all the stuff that you're used to for -- from Oracle that keeps your data safe and secure integrity, it all just works. So this is going to be a big deal, and it's unbelievably easy to use, unbelievably easy to use. All right. And then this AI vector surge, you combine it with generative AI with something called retrieval-augmented generation. And very quickly, again, I don't have a lot of time to talk about this thing. R-A-G, RAG, retrieval-augmented generation, how many of you heard of RAG? Probably a lot of you. Yes, you've heard of it. This is another big deal. This is how you combine the generative AI with the database. So the idea is, the user asked a question. That question is turned into a vector. That vector contains the semantics of the question. What is it that they're asking? We use that vector in the Oracle database to look up documents that match. So if a customer is asking, "Hey, tell me about this product and this version and what the limitations are with this." We find that question, we map it into a vector. We use that vector to find all sorts of documents or pictures that are related to that. And then we combine that with the Gen AI to answer the question in English language or natural language. So that's kind of how generative AI and databases work together goes into vector, the vector -- the database searches for relevant data, this relevant data hand over to the generative AI, it then answers that -- it provides an English language answer to the end user. Okay. Yes, and it's called retrieval-augmented generation, which many of you heard, which is -- if this is a big deal. Again, it didn't exist a year ago. So anyway, this is the big picture of what we're doing in Oracle Database. This is our goals, best database, best database, Oracle Database. We've been doing this for a lot of years. The best platform -- we are the best platform. There's nobody else that's even close. Autonomous database, if you look at the industry analysts, Gartner, IDC, they're now rating at the top for cloud database, cloud of customized -- there's nobody else that even has an offering that even competes in this space. Document data, this JSON stuff that I talked about is going to change the whole game on documents and graph and everything. We're unifying all these worlds. It's very different same thing with Graph. And then this vector stuff, the AI stuff, we're going to be ahead of everybody on that. So we're definitely going to win on that one also. So anyway, so that's the big picture. And I think I'm handing it off to Steve now. Thank you.

Great. Okay. So there's Ken slides. Ken's given me 20 minutes for apps. I need about an hour, but the good news is the first 40 minutes play in 1 I've already done. As you know, we're based on our stack, every feature that Clay talked about, and every feature that Juan talked about its inherited by 100% of our SaaS customers. So that's something that's very unique and very differentiated. Now the first few slides I'm going to go through are a set of logo slides. And I'm going to not really talk to the individual logos, but I'm going to talk to the importance of the logos and what's really happening from your perspective of the dynamic. You hear Larry talk about often, our lead, particularly in Cloud ERP. But I actually think he's underselling that lead for a couple of different dimensions. Here's one. So first off, the customers you see here in health care, what's starting to happen is, as we got a who's who of health care, we're learning from those health care customers. They're giving us great feedback. And as we're combining it with Mike's apps and Cerner, we're building out more and more features to complete the end-to-end business flow. So not only do we have an impressive set of logos and this is true across all the industries I'll show you in the coming slides, but that lead is increasing because in working with what those customers need in driving value, we're adding features that our competitors just aren't adding. This next slide is a who's who of basically everyone in the room, I think it literally would be everyone in the room, except there's a few of you that don't allow us to use the logos, but you actually are customers of ours. And financial services, what I want to stress here is, I'm going to talk to them for a moment in B2B. And a key driver in what enables us to do B2B is the fact that we are a cloud SaaS provider, not a cloud-hosting provider. You'll hear a lot of our legacy ERP competitor talking about how they're in the cloud. There's a few simple tests that you can apply. First, every single logo I'm going to show you, any customer you talk to at this conference or anywhere else, the Fusion customers, they are all on the exact same release. They all know when they're getting their next update. They all know where they're going to get their update in the third quarter of 2024 and 2026. It is a constant quarterly cadence of over 100 new features in every single area. Second, you can just look around on the public Internet. There is lots of documentation about Oracle updates. There is lots of documentation about other cloud updates. The legacy ERP competitor does not have anything out there publicly and you can't find any evidence that their customers are getting frequent updates. Why is that important? First, cost-wise, I don't know how you would possibly -- I don't know how we would possibly contain costs if we have thousands of customers all on disparate versions. Second, new features, and I'm going to talk about 2 of them today, AI and B2B. AI, as you all know, we've talked, and Juan talked and Clay talked. I've heard you guys talk about the data gravity. We have telemetry and data on 100% of our customers to base our AI on. When we roll out AI, we rolled out to all 14,000 Fusion customers. AI gets better over training. All 14,000 get it immediately, all get trained, all get the revisions and it gets better and better. So another example, we're extending our lead. Next slide is a set of communications and professional services, again, a who's who. Another thing that I don't think we stress a lot, but just so you guys know, when all of these customers, not only are on the same version of Fusion, when you get Fusion, you get all of Fusion. And one fundamental tenet of Fusion is that when you set something up, it is set up once. So for example, if you start with Fusion Financials, you're setting up your company structure. You set up legal entities and chart of accounts and business units, and the legal structure. You enter your employees for approvals. You do all of that set up and then you run your financials. If you then want to set up HR or HCM, you've already set up your legal entities, you've already set up your business units, you've already entered your employees. You've probably set up half of HCM to add incrementally. The same thing works in reverse. If you start with HCM, you've done all that set up. You've probably set up most of your financials going forward. So when you hear customers talking about how they're adding things to Fusion it is much, much easier to set it up because of that fundamental tenet. It's not different systems. You've done most of the set up already. Next, hospitality and gaming and quick-serve, another key set of drivers. And that's the final part that I'll talk about the importance of it. The third feature I'm going to highlight today is a new feature around data and analytics. And that's really combining all the parts that I've talked about before. One is, it's an end-to-end business process, not only our ERP and HCM, but also the industry verticals. And once you've set something up, it's set up completely and it's only set up once. So there's no data confusion, there's no data mismatch. Why is that important? Well, when you try to make sense of that data and extract that data and run reports, you have one source of truth, and it gives you much faster, much more accurate data. You guys all know our suite of [indiscernible], again, fundamental-driving tenet, single instance across the suite, single source of truth, makes the processing -- single process across the organization, makes it much, much faster. Okay. So those are the key points behind the tenets. We have more customers and more industries than I even have points, so there's still some more industrial manufacturing as well as logistics. Okay. We -- when we talk to our customers, we provide everything they need complete suite of applications. Innovation that matters, I'm going to highlight 3, and then our customer commitment to success, which is really our focus on driving that customer success. What do we mean by everything they need? In the Fusion applications, it's full suite, from sales, service, marketing, on towards supply chain, order management, procurement, inventory costing, warehouse management, on through the financials, a full robust set of financials, including EPM consolidations, reporting and HCM, full HCM, benefits, payroll, talent management, recruiting the whole lot. And then what -- so that in itself is unique for a cloud vendor. When you then add to that Cerner and our utilities applications and our hospitality applications and our financial services applications, and I'm going to leave a few out that Mike's team has and that integration completely and totally unique. And then you take the investment that Clay and Juan talked about, our development teams work together to design and improve these features. So Clay referenced how Juan is a pretty tough customer. And I'd like to thank our team is a tough customer on Juan and then a tough customer on Clay. We ask for features, they build features, and they build them for 14,000 customers. And then we get to deploy them to all the customers, and then we improve them rapidly. Some of that again, if you're not a true cloud vendor and if you're not both in the technology business and in the application business, very, very difficult, if not impossible to do, and we're the only ones that do this. And of course, the example of the day happens to be AI. Fusion applications, we are exclusively focused on use cases. Why? Because the underlying technology, the underlying improvements of the [ LLM ], the underlying security, the speed of processing that Clay and Juan both referenced, we had just inherit by our design point. So we are 100% taking advantage and really a customer of Clay and the services that they provide. And as that improves, automatically flows through and improves through the Fusion applications. And that translates into speed, because we didn't have to write technology and because we did not have the right uptake, we focus on use cases. And over the next 2 releases, we'll have 50 use cases across the suite, which I'll detail in just a moment of generative AI. Now before I go through the generative AI, I just want to remind people, we've talked a number of times this conference, we've been doing AI for a long time. And in fact, I think 2 or 3 years ago at this meeting, I talked about our new AI apps. So just as a refresher, we've had AI and supply chain planning forever. It's essentially an AI app. We've had an IoT app in supply chain. That's essentially an AI or a machine learning app. From the introduction of Fusion, we had features around document image and image scanning that does character recognition that improves over time based on feedback that you get. That's been an AI app from the beginning. We've had audit capability that does things like look for anomalous transactions and trying to detect fraud, that's been AI app forever. And the whole of our marketing suite is essentially an AI app. When we send out marketing e-mails, we use AI to determine what's the best subject line to put so that you will open it, what's the best time of day or day of the week that you would put to open it, what's the best offer that we should give to each individual customers, all of that's tuned in AI. So what have we introduced here? So what's here, we've introduced generative AI. And before I get into the use cases, a couple of big parts of feedback we have from our customers that we follow. So first, we do not take our customers' data and use it to train AI. So all the service levels, all the protections that we have with our customers before AI, we maintain with the AI capabilities that we offer today. Second, we never pass PII to the AI engine. So there's no personal identical information that gets sent in any AI use cases. And finally, at the moment, with generative AI, it is all humanly reviewable. So that means we don't automatically pass things through transactions, and this protects us and protects our customers against what's today, commonly known as hallucinations from the AI engines. Now as those things improve and we get security on the AI and we get improvement to AI, again, on the Fusion apps, we don't have to do anything from the technology. We just changed the use cases. So the use cases we've announced at this conference. In CX, it's a lot of authoring use cases, both in service, so to be able to reply to the service requests, and it's a little bit like using generative AI to do chat bots that you might see on a common website when you need help, but it's just much, much better, it's much improved technology than what you see in most places today for service. Knowledge articles. It's very common for organizations to try to be proactive with their support so that instead of taking an inbound question, they produce, how to or knowledge articles, that can all be now generated with generative AI. And again, if you use ChatGPT, you see examples of that because it does an excellent job of writing and writing complicated text. Also, in field service, we use it for field service recommendations. And in sales, we use it for sales reps to generate proposals to their customers. So no longer do you have to have a sales rep or that marketing engine writing the e-mails or contacts, you have generative AI generate that on their behalf. And again, it improves going forward. In ERP, this is a use case, which is very interesting because it's not necessarily authoring in the same sense, but it's a set of functionality we call financial narrative reporting. And what that means is you can pass any financial report to the generative AI and ask it to summarize that report for you. So it's a balance sheet and income statement, AP trial balance, a financial report and AI can both summarize and alert you of what may be the trend or an anomaly in the report. And again, this is something that you can test or see very easily on the public LLM, you can take some reports that you have or an e-mail or a PowerPoint, send it to that LLM, ask it to give you a summary of what it's about, does an excellent job, and this is a great use case for the ERP users. In supply chain, we use it for item description. So especially large manufacturers, item descriptions have a long text associated with that. Now the LLM could generate that on our customers we have. And also supplier negotiation suggestions is another use case. And then finally, in HR, a host of areas where the AI or the LLM replaces tedious work that humans do today, and in some cases, a lot better than what humans could do today, especially in generating written. My favorite example, it's my favorite because you could test it very easily yourselves, is to help generate job posts and job qualification. So with every job post, if you're going to post a job, you need to include, does this need 3 years of experience and a certain degree, certain background, et cetera, now that generative AI can generate it on your behalf, the human reviews it and they post it for your job posts. Great efficiencies and, frankly, most times better written job descriptions and job posts. And again, you want to see what that's like. Take your own company name, go to a public LLM and say, write a job description for your own job or a job you're looking for or one of your peers, and it will write a very, very good job description. Now it's embedded into our HCM and our recruiting applications. And all of our customers get this as just part of their normal update. No changes on their behalf, no extra subscription, making our apps better and better and better and extending our lead. So AI is one set of the big announcements we had this week. The second set was really expanding on our B2B announcements of last year. So last year, we talked about a partnership with JPMorgan and with FedEx to improve payments and to improve logistics provider. And the basic notion, very similar to our strategy with industries where we want to get much more of a full suite of transactions extending that to what's commonplace ERP because today, in ERP, you're in the ERP, but if you have to do a payment, you go out to a bank or you have to go to a shipment, you go out to the FedEx or UPS or the DHL shipment portal. So we've released all of those components we talked about last year. We're working with a set of early adopter customers, both on expenses, payments and logistics. But this year, we announced a further partnership, now with Microsoft and HSBC, and we're going to continue in this mode to expand out partners and in particular, with pardon me, I think I may have said Microsoft, with Mastercard, we did partner with Microsoft for a different issue. This was with Mastercard and HSBC, which has the same payment capabilities and expense capabilities we talked about last with JPMorgan. But in particular, we're very excited about 1 more use case that Mastercard and HSBC brings to the fore. And this is with virtual cards. And the key here is this really unlocks a brand-new market in terms of how B2B payments are done. So let me just take a moment to spend on this part because I think it's critical, it's important. Today, you as a consumer, most of your payments are done via credit card. Some of us may still have a checkbook and may write checks for a few things, but for the most part, that is done. Today, B2B payments, business-to-business payments are still very much done in an EFT or essentially check on a net 30, net 60 type of arrangement for payments. Very little B2B transactions are done via automated transaction via a credit card like we do consumers. Why is that? Well, one of the reasons why is as consumers, you apply for a credit card, you have a credit limit and you spend it on whatever you choose to spend it on. That's not quite how B2B or how businesses work. In businesses, you tend to work with a purchase order. So there are certain purchases that are approved. Now that purchase order is for a given supplier for a given goods or service at a given price or price range for a given point in time. What a virtual card is, it basically encodes that into a card. So it's not a card like a consumer card that you can spend on anything. It's a virtual card, which is designed for a specific purchase from a specific supplier, specific goods and service, if you will, very similar analogous to a purchase order. So what is the problem today? The problem today is businesses have hundreds, if not thousands of purchase orders that you make. And to get a virtual card today, you need to get that information into the bank or the Mastercard portal, apply for that virtual card, get the virtual card issued, then go forward. And what you've done is, in that process, it's really quite -- there's a lot of friction in that process, a lot of time and money, and it offsets all the benefits that you would achieve. With this integration, we're announcing with virtual card, essentially all of that data that you need to encode the virtual card is in Fusion ERP. So when the vendor -- when our customers want the virtual card, they basically call the API, goes out to Mastercard, creates the virtual card much more frictionless and allows a B2B payment. Huge benefits for our customer, the buyer, huge benefits for the seller and a lot of economies of scale going forward. Now this feature is only possible because we have point-to-point partnerships with banks, financial services and in some cases, logistics providers, meaning our cloud calls their APIs. If we were not a single cloud with all of our customers on the same release on the same infrastructure that Clay provides in all those data centers around the world, we would have to code this API every single time. So it's another example. I talked about the benefits of AI, but also to do things like B2B transactions that extend, if you're not an actual cloud customer, if you're a legacy ERP vendor, almost impossible to have end-to-end business transactions at any kind of cost because we've done this partnership once we've coded it once, and it works on behalf of 100% of our customers. And we have -- we talked a little bit earlier about the sponsorship with Uber Freight. We're going to continue to extend these partnerships to have end-to-end business transactions across the board. Okay. And then the last topic I'll talk about is our Fusion Data Intelligence Platform. And at the beginning, I opened and talk about that single source of truth and the importance for every single customer. And this is sort of extending that and really bringing that and the AI together. So with the Financial Data Intelligence Platform, today, we have what's called the Fusion Analytics Warehouse. And it's basically a data warehouse that integrates with Fusion applications. So if you use finance or HR or supply chain or CRM or all of the above, you can extract it into the financial data warehouse and it could do reporting for you, just like a classic data warehouse will do. It gets automatic integration, has a lot of speed, a lot of benefits. But what we're now extending is change that into a full data platform. So it has a couple of significant adds to the analytic warehouse. The first add is applying data science, AI and ML to the data warehouse. So instead of a static reporting view, is we've now taken advantage of AI to give recommendations. And those recommendations will vary, whether you're talking about an HR report, a finance report, a CRM report, supply chain line of business, again, detect anomalies, detect problem areas and push recommendations. And because it is not a stand-alone data warehouse disconnected from a transactional app because it sits right on top of Fusion, it takes those recommendations, and if you choose, can issue the appropriate transactions to actually take advantage of it and issue it within Fusion. The second big enhancement over the analytic warehouse is now we've added capability to further extend your data model to handle third-party data. So again, we have it to focus on industry, we have it to focus on B2B and now taking your analytic warehouse and allow you easier and flexible ways to add third-party data, so you have full end-to-end visibility into your business and reporting with AI on top of that and integrate it to transactions. Okay. And then the last thing I'll point out is our ongoing evolution from -- really, to become a service company that lasts, S in SaaS. And hopefully, you've heard that a lot from our customers this week, talking about how we've been involved with their implementations, the success of their implementation. I want to call out one example, which is our customer connect. So I mentioned at the beginning, we averaged a little bit over 100 new features every product area, every quarter. About 80% of those new features come from our customer suggestions. So we have 300,000 people on our customer connect, it's basically an online forum, a Reddit-style forum, if you will, though Reddit has some negative connotations, but a Reddit-style forum. And one of the areas is an Ideas tab, and allows any end user or any of our SI partners to log ideas or enhancements and then has a voting mechanism. So that raises those ideas to our product managers who could then ask questions to the actual people who logged it, get feedback. Oftentimes, we have partners, not only giving feedback, but also suggest, here's how they actually solve that problem without the enhancement at other areas. But it drives 80% of the ideas that we have. Now I'm going to come full circle back to the first thing I announced with the AI. While we're very happy with the speed of those 50 use cases that I talked about, what I'm much more excited about is now -- just like ChatGPT spawned a bunch of ideas for how AI is going to change multiple areas, and now there's a lot of investment, a lot of companies trying to take advantage of those ideas. As we get these 50 use cases out to those 14,000 Fusion apps customers, the importance of it going to all of them immediately is that they now will get ideas of their own. And you tie that with this community, they'll push those ideas into -- directly into our product managers. So our AI, A, we're going to go faster than our competitors because my team doesn't have to worry about the technology, Juan and Clay, improve that; and B, we're not alone in coming up with the ideas we're going to instill to improve the product going forward. We have those 14,000 customers in the customer base which are going to give us not only ideas in horizontal applications, but also have their industry flavors of all those industry sections that I talked about at the beginning, okay? So with that, that's Fusion applications. I will hand it over to Mike Sicilia.

Thanks, Steve. I'll say a few words here today about our industry applications. And what's, I think, allowed us to have very differentiated conversations with customers, and actually partnership opportunities with these customers that we otherwise wouldn't have because of the industry applications. Of course, I'll do so under safe harbor. Okay. So for many years at Oracle, we have been providing industry-specific applications. If you think about the biggest difference between the Oracle industry clouds and everybody else's industry clouds, is that we actually have industry applications and industry IP. And we're in the business of providing last mile, consumer-facing, patient-facing, customer-facing technologies to our customers, so that, that technology can better enable their interactions with either a customer or a patient. So we're focused on complex operations, multi-country operations. We're looking at regulatory compliance as a service. I mean a lot of our industries, banking, utilities, health care, heavy regulations, and we spend a lot of time making sure that the cloud services that we provide are compliant, not only in the United States, but globally as well. You've heard quite a bit about the end-to-end applications, and I'm going to walk through a few examples is to why this puts us, we think, in a really interesting and strategic conversation space with our customers of communications, financial services, health care, retail and so many more, I'll walk through them. Integrated with, sitting on top of our Fusion applications, ERP, supply chain, human capital management, and of course, sitting on top of OCI. So OCI has become an incredible differentiator for us in so many of these markets. If you think about health care, for example, and you think about the global opportunities in health care, where the rest of world outside the U.S., outside the U.K., Australia, Canada, is largely greenfield for electronic medical records and health care digitization. There are 2 huge barriers to implementing digital health care solutions in the rest of the world. The first is, there aren't enough doctors. That's a big problem. The second is, data sovereignty. Data sovereignty is the #1 topic that comes up every time we talk with countries around the world about adding in the health care system. We are committed to not only delivering health care applications, but we're also committing to deliver in sovereign infrastructure in countries. It doesn't matter how big or how small the country is. We will deliver fully-sovereign country -- some of fully-sovereign solutions. Some competitors will come in with a little bit of edge compute, but the [ DR ] goes somewhere else. That's not enough for health care. It's not enough for personal information. The data has to stay in the country. What we're delivering is a full stack, front to back with full sovereignty, full disaster recovery, full active -- full active cloud infrastructure inside the borders of that country. Again, no matter how big it is, no matter how small it is. How can we do that? Well, this infrastructure here continues to scale and continues to get more and more efficient as we move from 12 racks to 10 racks and you'll see that number continue to move down. That's a huge differentiator. Without that, we'd have a difficult time delivering at the top because we couldn't do so in a sovereign infrastructure. Now of course, the other clouds don't have all of it and don't have the ability to deliver the sovereign infrastructure. So we think it puts us in a very unique space. So we're focused on automating core operations for many industries. I'll go through them in a second. Generating new industry-specific revenue opportunities. This is a byproduct of being close to our customers' customers. And I'll give you a few examples in a second. Standardizing operations and again, complying with industry requirements. So let me walk through our industries and what we're up to. In communications, we've primarily been a supplier of billing solutions for our communications customers as well as core network infrastructure. So the infrastructure that makes your phone calls and text messages work, if you will, 4G, LTE, 5G and so on. Interesting things happening in the communications business today on the wireless side. So many of the major telcos have announced 5-year price caps, so that they're not going to raise prices for the next 3 or 5 years. That puts downward revenue pressure on everybody else. At the same time, the communications industry has to spend a tremendous amount of money moving from legacy 4G LTE solutions to 5G. So a lot of times you see it on your phone, you see 5GE. That means you're connected to a 5G radio. It doesn't necessarily mean you're connected to full 5G infrastructure. The build-out of the infrastructure for communications for 5G is a multiyear, probably decades long process that will happen over time. So very expensive. It's a very expensive process. And when you've got also headwinds with revenue because you've got consumer pricing that's capped by your competitors, what are you going to do about it? Well, where is the money coming from today in communications? Where is the money coming from at telco? It's coming from streaming and it's coming from over-the-top content, at least at the top line. So we're not only now delivering as a cloud service, our billing solutions. We're not only continuing to deliver 5G infrastructure, but we're delivering using APEX as the low code, no code generator that you heard Larry speak about the other day, an application development platform right on top of the 4G, LG and 5GE, 5G network infrastructure. What that means is that our telco customers can create their own over-the-top applications, their own vertical applications right on top of the Oracle stack. And we're virtualizing that so that as they move from 4G in some places in the world, still 3G, 4G, LTE to 5G, they don't have to change a line of code. They don't have to change any of the code. APEX, low code, no code generation, will take care of all the anomalies in the virtual layer. In engineering and construction, we supply the largest general contractor subcontractor payment system in the world. So if you look by SIC code in the engineering and construction industry, approximately 88% of all businesses in SIC are small businesses in engineering, construction are small businesses. The primary reason for major delays and major problems on large-scale construction projects is cash flow. Small businesses don't get paid on time. They stop showing up and stuff stops getting done. It's pretty much that simple. We've automated the process for small businesses to get paid on time with a couple of conditions. The first is, they have to have lean waivers on file. They have to have insurance certificates on file. And then they have to have -- in the U.S., it's OSHA and other countries, safety regulations are different acronyms, but they have to have their safety regulations on file. If all that's on file, we guarantee that they'll get paid on time. Last month alone, we processed over $14 billion in payments in that construction cloud. So it is the largest single construction cloud. And it's a really interesting dynamic because, as I said, cash flow is the single biggest inhibitor of progress in construction cloud. That continues now to expand globally. In Food and Beverage, we recently -- we've long been in the food and beverage restaurant automation business with point-of-sale. We recently launched our online menu management and our payment system as well. So we are now competing directly with Toast, Square, Shift4, Clover, everybody else in the payment space. We announced Waffle House, as our signature customer sign on. And we're very excited about this because we're able to deliver a full suite of applications, no matter of the size of the business. It's point of sale. It's menu management, it's payments, it's integration with Uber Eats and it's also depending on the size of your customer, NetSuite plus Fusion applications. Similar thing in hospitality. Property management system. We've been rated by IDC as the #1 property management cloud in the world. And that is -- we're seeing -- excuse me, a very interesting dynamic here in hospitality as customers are striving to move to the cloud, particularly with cyber activities happening, some of them you saw here in this town very recently. The average age of a lot of the big property management systems or the average versions of these -- a lot of these big property management systems is over 5 years old, 5 to 7 years old and some of the hotels, which in consumer-facing technologies is pretty old. So what we've done in hospitality is something very interesting. It's exactly the same thing, lesson learned that we're doing in Cerner. We've taken our older versions, our on-premises versions, and we've not changed the schema at all. So we've not changed the schema. So for customers who want to move from on-prem to cloud, there's 0 data migration. You push a button, you've heard this before. You push a button, and you move it to the cloud. So we're literally taking major hoteliers, major brands to the cloud in a matter of hours without changing any of the data. Local government and public safety, you saw in the keynote the other day in Larry's keynote. They announced the go-live in Santos Los County of our police automation, command center, body-worn devices in local government. In retail, we talked quite a bit about what we're doing with Uber, and I'm going to walk through an example as to why I think there are a lot more opportunities to think about what we're doing with retail from a consumer-facing standpoint with companies like Uber for partnerships. In life sciences, we're working together with companies like NVIDIA to do molecular simulation for clinical trials. We've long been in the clinical trials business. We have the largest pharmacovigilance or drug safety business in the world. Last year, last calendar year, 91% of all adverse events that were reported to the U.S. FDA for vaccines came through our drug safety system. Health care. I'm going to talk more about in a second. Financial Services, as I mentioned to some of you the other day, we've been in the core banking business, the demand deposit accounts, banking originations both on the retail side and on the investment banking side for many years, banks are now -- many of you, you really, I would say, in the last 6 months, really clamoring to move those core banking applications to the cloud, where they were resident first, are now starting to move their core banking applications to the cloud. Our Energy & Water business continues to be very strong, both for meter data management, power controls, grid management, as well as looking at consumer demands as well. So in our Opower Cloud today, we can tell when a consumer is pulled in their Tesla, what time they plugged it in. We know whether it's a Tesla, we know whether it's a Chevy Bolt. We know what the model is. We know which models of appliances you have, and we help our utility customers give back to our customers and tell their customers some intelligence around when may be optimal times to charge their cars and how we can save the other over 32 terawatts of electricity today. So another way to think about our industry applications is that they sit between our back office applications and our front office applications or CX applications. Sometimes they're a little bit more towards this edge because these applications actually touch our customers' customers, and I'll show you some of them here in just a second. I talked a little bit about the importance of OCI and the importance of OCI as the enabler, thinking about how we deliver these applications. Our Fusion applications as well as our industry applications now are certified to run on any one of these delivery models. And that's incredibly important to our customers because a lot of our banking customers or [ comms ] customers are looking at dedicated regions. A lot of our government customers, of course, are looking for government-facing customers for health care are looking for public cloud, but of course, with government certifications, with government controls in them as well. And then, of course, we've got many customers who are running in the multi-cloud environment, no matter where the industry applications are working. No matter what the deployment model is we're able to certify them and run on them on top of OCI. So let me move ahead to a couple of examples here in the interest of time. In health care, as we acquired Cerner, and we said last year, our theory was that others have tried in health care. Others have thought about it. But the reason they haven't been successful was because they weren't willing to take on enough of the problem. They weren't willing to take it on. They took it on, maybe one piece of this, and then the EMR pieces got broken up into so many different pieces. And it just was -- it was kind of a mess. And if you look at the integration cost, the regression costs, and if you go into these major health care organizations, and you look at the average age of the IT assets they're running, it's incredibly old, incredibly insecure and incredibly rigid. So as we put together with our health care applications, we've now integrated, as Steve said, our Cerner applications with our Fusion applications for HCM, for example, so that we know, based upon the number of patients that have checked into the ER, how many nurses you may need in the ICU 2 days from now based upon their case progression? We can predict based upon how many people may predict with acute illnesses that your ICU may get overloaded and you may be understaffed. Supply chain management in health care is another example where we are looking for clinical -- surgeries that are scheduled. So we've got the whole list of the surgery schedule, the operating room schedule. We can check with the supplies, the supply chain management solution, not only to see do we have the right number of supplies? Are they in inventory? But one of the biggest problems in hospitals is nobody can find them. They don't know where they are. They don't know where they are in their rooms. So we can give everybody a heads up to say, you don't have enough supplies for the next 2 weeks' worth of orthopedics -- knee replacement surgeries or hip replacement surgeries. That said, the stuff you do have is in closet X or closet Y or somewhere along the way. So bringing this whole thing together allows us to have a conversation with our health care customers. And I can say having been in many of these conversations, our conversation with our health care customers are always with the CEO, if not the CEO plus the Board. This is an all-in solution. And what we're looking at is a differentiated conversation to say, we can obviously give you better outcomes. We can give you far better security. We can deploy this in the model that you'd like to deploy this. But the conversation really is this. How much are you spending on IT today? And we can do it for what you're spending minus X faced the conversation. That's very different than anybody in any one of these pillar spaces conversation is having with our customers today because they're not able to look at the holistic piece. We're here to solve the whole problem. Now what often happens is that the customers will say, love that strategy. And we've got many examples. I'll talk through some of them in the health care session later today. I've got to start somewhere. Where should I start? So we can -- it doesn't really matter where they start because the APIs that we put together, and the way we've organized the development teams inside Oracle, I think, is very unique in that we've got health care functional teams in every one of our horizontal applications as well as our vertical applications. The APIs that we're using to put this together are all public. So that we use the very same set of APIs. We document all those APIs publicly that any of our customers can use. We don't have any private APIs. There's nothing in the way. And if customers wish to sub out any of these components or other components, they're free to do so. And we guarantee that it's not going to break anything in the cloud upgrade. It's a huge difference from what's happening in health care today, keeping in mind that in the United States, there are many customers who are running a competitor, the [indiscernible] product that runs on a [ mom's ] database that was created in 1965, right? That was when it was created. A little different than the Oracle Autonomous Database that this now runs on. Okay. Looking at banking. As I said, a lot of our banking customers now are really getting interested in moving what we call the core banking assets, the retail banking and corporate investment banking, as well as our analytics applications to the cloud. I mean, of course, we saw a lot of traction, as you've seen over the years, Steve showed us the logos, for HCM and ERP, I would say there were some reticence on the critical core assets to move to the cloud. Now the demand is very high to move these applications to the cloud. But it brings me to an interesting concept, what does our banking cloud have to do with our construction cloud? And this is something that we're having real-time conversations on -- that I find very interesting. Going back to this construction example with all these small businesses, not only do small businesses tend not to get paid on time, which causes problems in construction, but they also have a difficult time getting capital from banks because largely, their invoices are labor invoices. And banks that typically don't lend or collateralize labor invoices, they won't factor labor invoices. They only factor in invoices that could be collateralized. So -- but in our construction cloud, in our payments construction cloud, we have all of the data that shows the default rates. It shows how many times an invoice was paid late, short paid or not paid at all. So we can -- and we also have KYC compliance in our construction cloud as well since it's a payment app. So we can provide to our banking customers, information to them that could be very interesting for them to enter new verticals with over a decade of data that shows a lot about that customer, their payment patterns, and again, the number of times that they may not have paid or paid at all. Without OCI at the helm, and without having a huge central database for all these payment applications, we wouldn't be able to present this to the bank. So now we have an opportunity to say, not only bank as a customer, but bank as a partner as well. Same thing happens in our food and beverage payment systems as well. We can bring merchant services and credit card payment and all these other things bundled with our food and beverage systems right to our customers. Okay. And then finally, retail. You heard Safra talk with Dara on Tuesday on retail. So we started the conversations with Uber here at the OCI level. Of course, it makes a lot of sense, lift to shift your critical assets, run it on the best cloud, security, all the things security, performance, reliability, all the things you know about OCI. As we started talking with Uber at the business level, we said, "Hey, by the way, we have our food and beverage business, our restaurant business that's integrated with Uber Eats, where we can dispatch Uber Eats drivers from -- directly from our point-of-sale terminals in micros." Uber came back to us and say, "Hey, by the way, talked about it to a lot of our retail customers. And they really like your planning inventory operations and merchandising solutions. Here's a real problem that we're having. It's not just delivery, it's actually returns." So when you buy something from -- particularly if you buy directly from a manufacturer, like I got roller board luggage for travel, which -- and I got the thing and I'm dragging it, and I think this isn't one of these things with the 4 wheels that you can just push it easily. I've been traveling for a long period of time. I don't know why I made that mistake, but it was fortuitous that I did because I had the experience of having to return it. And I learned -- and this is one of the inspirations we had on our retail team for this, they said, we'd much rather return it to the local big-box retailer than ship it back to us because they can put it right in inventory. When I started having discussions, I don't think it was just my luggage thing, but we started having discussions with Uber about the complications of returns and last mile logistics. So Uber as a customer, Uber as a partner, now Uber as a feature in our retail merchandising and point-of-sale systems. That's the differentiator. I hope that gives you a little bit of a flavor of all the different levels of the stack that I think allow us to have more strategic discussions with our customers, more strategic discussions with our partners, and frankly, the line between customer and partner continues to blur. And I think that's a great thing for us and for their customers as well. I'll talk more about health care, more specifically about health care and Cerner in the later session. With that, I'll turn it over to Doug for the financial update.

Great. So that is a tremendous amount of technology. I hope everybody got a good idea of what's going on at Oracle. What I really wanted to do today was bring it all together and reflect on some of the goals and aspirations that we laid out here at the Financial Analyst Meeting last year and how we're doing against it. As Ken has said, there's a safe harbor statement. And the second up here, which he also alluded to is we'll go through non-GAAP measures today. So with that, let's get into some of the details. So first, let me go back a little bit and start down our cloud transformation journey because I think that helps give you a little more context before we get into where we're at in the -- as of today and compared to last year. So it's -- as everyone who's been an investor with us knows it's been a bit of a journey. As you can see up here from 2010 to 2020, you can see our growth was tepid as we went through the difficult process of really moving 2 major things. First, we had to go out and build all these amazing cloud technologies that we've talked about through the course of today, not only to get them built, but actually to get them to parity when compared to our on-premise products, as well as then to build in differentiated capabilities compared to our competitors. And then second, we had to go through the difficult process of trading all of that upfront license revenue that we used to get in return for ratable subscription revenue in the cloud world, which really made seeing our progress difficult to do from the outside. Now on the positive side, the combination of these 2 things have really given us confidence in our future, and they actually gave us a really solid line of sight into the revenue acceleration that we hope to achieve. And if you look at the mix of our businesses, the revenue acceleration really becomes apparent because, going back to FY '15, cloud was about 9% of our growth businesses, which is what largely the cloud, was about 9% of revenue. As we went through that transformational process, by the time we reached FY '20, that mix had now reached 26% and it was now larger than our declining businesses, which was largely license and hardware. And as a result of that shift and to the growth business as being a bigger percentage, we knew that revenue was going to grow. It was just math. I mean we could see it from the financial models. So we went ahead and wanted to make sure everyone else understood this as well. And if you recall back in, at the end of FY '20, in our Q4 earnings call, Safra highlighted it back then, about this mix change and how because of what we are seeing, we knew we were going to return to revenue growth. And not only we're going to return to revenue growth but that we expected our revenue to accelerate as we proceeded through the -- from there on forward. And lo and behold, that's what happened. So it started ticking up. In FY '21, we saw a 2% growth in revenue, and that increased further to 7% in FY '22. And again, as you can see, those -- that mix of growing businesses as a percentage of our overall revenue continue to go up, now from 26% to 32% by FY '22. So as we exited that year, FY '22, with those trends in mind, we came out last year and put out some -- what we thought were achievable long-term goals to help you understand where we saw the business going. And if you recall, there were 3 measures. One was that, by FY '26, we believe that we'd reach $65 billion in organic revenue, including the Cerner business. The 2, that we increase our operating margin back up to 45% by then and that we would return to a greater than 10% annual EPS growth. Now as we add another year against that plan, you can see that indeed, that continues to be the case. So as everyone recalls, FY '23 had the impact of the Cerner acquisition, excluding it, it was another 7% of organic growth. Now it looks like 7% and 7% are the same, but they actually aren't. It's almost 1 percentage point of additional growth in FY '23, it's just the rounding works out this way. Now in order to achieve that $65 billion revenue target, basically, the implication is that we have to grow 9% a year. So I want to do is, go through today and give you a little bit of a sense as to why we're confident that we're going to be able to achieve that. Okay. So in order to do so, I think there's kind of 3 things that we're really focused on in order to attain those FY '26 targets. First and foremost is, we've got to retain our competitive differentiation. Innovation is, I think everyone can agree is one of the most critical factors in being successful in the technology industry. You heard a lot from our development executives about what they're doing in order to achieve that. We've really been paranoid on this point, since we've been around for more than 40 years. If you're not paranoid, you can't last that long. So we think we've got a lot of unique differentiation that's going to help us, again, accelerate the growth. Second is, once we've got all that technology put together, we've got to obviously go out and sell it. We've got to continue to execute on the go-to-market side of our house. And as a result, when we continue to do that, cloud is going to continue to grow as a percentage of our overall revenue, which again helps naturally accelerate the revenue growth rate given its much higher growth rate. And finally, we've got to convert that bigger revenue base into higher margins as we scale. So now on the good side at Oracle, that's always been our bread and butter. We've always had one of the highest operating margins in the industry. I think that's the part where we know, we know how to execute. So as I mentioned, innovation is job 1 at Oracle. So all of the speakers before us talked a lot about the differentiation that's here. What's interesting is that having that full stack of applications and infrastructure really served us well in the on-premise business. It's actually even more important in the cloud because as we talked about, being able to take all those applications and build against a common layer gives us a lot of consistency, manageability, security that no one else can offer since they're executing on someone else's cloud. Now on the application side, both Steve and Mike talked about the fact that not only having front office and back office or ERP and supply chain and HR, but also the CX side of things, then married to all these industry applications gives us a very unique differentiation. Mike talked about and I did the same last year about how Oracle Health is so unique, that we're able to take these different components and put them together into end-to-end business processes in a way that, again, our competitors can't do. Now on the infrastructure side, we've made some pretty amazing strides as Clay went through, not only to help move our customer base to the cloud, but also to capture new types of workloads like cloud native and AI companies. So all in, I think the really interesting thing is to see how cool it is to be at Oracle again. How we've all of a sudden got our mojo back, and I think customers, as you saw throughout this conference this week of -- can help explain that. Okay. So when you look at the overall opportunity for us, we think, in terms of available market, we've had about $0.75 trillion of spend that we can go pursue. Now within those 2 areas, we start with the cloud applications where we actually got started a lot sooner in the cloud. And now we've built a business that's over $11 billion in revenue. That's growing organically at 17%, using FY '23 as the number. On top of that, though, our strategic back-office cloud apps are growing at 23 -- 27%, so even faster than that. On the infrastructure side, the market opportunity is about double the size. And in that segment, we've only recently hit our stride. As you said -- as we've talked about, we're now at Gen2 versus our competitors in Gen1. Took us a little longer to get there, but as we saw here, in FY '23, we grew that business 63%, and it's actually 74% when you exclude legacy hosting services. So we're catching up to our competitors in this area. We're growing meaningfully higher than them. And so as a result, we really are seeing hyper growth mode at this point. Okay. So what I want to do is dive into each of the areas of applications and infrastructure, give you a little more clarity in terms of what we're going after and why we think that available market and the opportunities in front of us give us confidence in reaching the long-range targets we've discussed. So no presentation. I can't imagine that the developers couldn't do this before I did, but no presentation. Oracle doesn't have the Oracle Red Bull Racing team in some way as a great analogy. So what I'd like to think about in applications is just like Oracle Red Bull Racing has won the Constructors' Championship last year and has a lot to do it again this year. That's kind of how we think about the cloud applications business. We've been on a tear, and we're in a leadership position. I don't think our current overall portfolio is -- can be met by anyone else in there. So we feel really good about what's in front of us on this side to continue to be a leader. But let me just start by recapping kind of what we think are the main points of differentiation and how that translates from a financial standpoint. So we start with the innovation story. So again, comprehensive end-to-end applications across mission-critical business process requirements in the enterprise, that's our -- that's what we focus on, and that's where we think we have a huge amount of differentiation. Now having all those -- these comprehensive set of applications actually gives us a lot of at-bats. Getting into these strategic conversations with, as Mike alluded to, the CEOs and the Boards of Directors, back in the on-premise stage, that really wasn't what Oracle did. We are very much focused on point areas in which we sold. Now with this portfolio and the transformation that these industries are going through, we're not only selling higher, but we're selling broader in terms of the size of the deals that we're getting. And the other big area that's been critical here is that we've really increased our retention rates. So a big part of the cloud now is that not only do you have to sign up customers, but you got to keep them around. And so we've launched, if anyone had seen, the customer success services in the last year, and that's a real great differentiation. So anyone that, as Jason talked about, that went to the expo hub, that was a very busy area. Customers love to go talk to other customers, talk to our experts, really learn how to maximize what they're using from Oracle, and that helps us out with -- from a retention standpoint. Third is to move our on-premise customers to the cloud. I'm going to give you a little more information about this in a second, but really, that's the low-hanging fruit. We just need to have the features they want and meet the expectations from a service standpoint. But it's -- there's a lot of opportunity still in front of us to migrate our on-premise installed base into the cloud. It's hard for others to match it. And so we feel good about what that opportunity presents. And the other big area is to actually move our competitors' on-premise installed base to the cloud as well. Again, we think, as we've talked about, we're really the only true cloud-native company, as Steve discussed. And I think that gives a huge -- again, a huge point of differentiation in talking to customers about what they should do next in order to really maximize innovation and utilize AI. Okay. So against that cloud strategy, you can see here that the percentage contribution in our applications business from cloud has now grown from 17% in FY '15 to now 63% of our applications revenue. As I said, it's about an $11 billion business. Now you dive into the cloud portion of the applications business, you can see the strategic gaps. The higher growth areas now represent 74% of our overall cloud revenue. So again, that's a big deal in this effort because we had to go through a pretty painful process. As everyone knows, we made several acquisitions early of cloud companies. All those cloud companies are running on a multitude of technologies and underlying data center platforms. We have gone through the process of all converging them on to OCI and oftentimes actually rebuilding that. So like Taleo got rewritten as Fusion recruiting. So we've gone through that process. Now we add to those strategic apps the NetSuite acquisition, which went after a different set of the market. And between those 2, we now have a greater and greater percentage of our cloud revenue that's now coming from these strategic high-growth areas. And turning to the incremental opportunities that are existing with our on-premise installed base, there's not only our horizontal application opportunities, so between PeopleSoft and JD Edwards and the Oracle E-Business Suite, but there's the industry application customers that are available as we've now been busy building all of our industry applications as cloud technologies. So as you can see here, only about 10% of our installed base of support customers and applications have fully moved to the cloud, still is a lot of support revenue that's available to us. And again, I think we talked about this last year that we see about $3 to $4 of cloud revenue in return for $1 of support that's migrated. So when you take that against that support installed base, there's about $15 billion to $20 billion of incremental revenue available to us as we migrate our installed base of customers to the cloud. Similarly, if we look at our competitors' support installed bases, again, we have another great opportunity to migrate them. We've actually had more success upgrading competitors' installed bases than our own because E-Business Suite, PeopleSoft, I mean these are rock-solid application platforms in comparison to folks like Infor and Epicor where there really is no good cloud option for them. So we've moved a lot of those folks. And the other big elephant in the room is the opportunity to move the SAP installed base. Now on the good thing about SAP is they're trying like heck to help us out. They've announced their desupport of many versions of their software by 2027. That's actually forcing a lot of conversations where customers realize this is only lift and shift at SAP. I really want true cloud. I want true AI. I want to go after what the power of the cloud offers, and that's giving us great opportunities to move folks like Grupo Bimbo, Skanska AB, the Kroger Company, many other SAP installed base customers who have now moved to Oracle Fusion. So when we look at the same opportunity with about $25 billion of installed support base based upon the research that we did, multiply that by that conversion rate, we got about $75 billion to $100 billion of opportunity to move our competitors' installed bases to our cloud. So when you roll this all together, starting with our run rate business, about $11 billion of Oracle Cloud, plus the opportunity to convert our support installed base plus the ability to land our competitors' installed base, it's already over $115 billion of available market to us. And that doesn't even count things like new opportunities for greenfield things where customers like start-ups are just getting going or people have built custom-built software that aren't even in these packaged software categories or actually switching other cloud providers like Workday, who just aren't working, who want to move to Oracle Fusion. All right. So now I'll move to the infrastructure side of the story. So again, let's do a little racing reference. So versus the applications where they're a clear leader, obviously, we'll be coming from behind. We weren't in the pole position. But as your rearview mirror often says, objects in the mirror are much closer than they appear. That's sort of how we feel, which is we've got amazing momentum, and we're sneaking up on these other hyperscale competitors as we continue to land workloads based upon what Andy -- sorry, what Juan and Clay went through. So let me go and recap again the strategy on the infrastructure side in terms of our differentiation, so that gives us the confidence to keep lending these workloads and hopefully pass our competitors. We've got Google clearly in our sights at the moment. So the first thing it starts off with is we've got to build that global footprint of data centers in order to make sure that we can meet customer needs anywhere in the world of any size and type. Clay went through a lot of that differentiation already, so I won't belabor it. But we think, again, that we've got -- we're in a unique position now in the marketplace. Second, which hasn't necessarily always been an Oracle strong suit, is we've got to build a rich ecosystem of partners. There's been a real transformation internally at the company, really led by Larry to go out and embrace these things. The Oracle-Microsoft relationship would never have been something we would have seen in the on-premise world, right? It's a very different environment. So we've really tried to embrace partners and make sure that it's not only partners for ISVs to make sure we can create holistic solutions on OCI or in our cloud but also actually port our technologies and make sure that we can run in a multi-cloud environment. Third is we've got to rapidly expand the work we're doing with native and AI-focused companies because those are the folks that are spending an inordinate amount of money on the cloud. And as you've seen, we've been able to land many of those. So there's a great opportunity to expand with them. We've already signed $4 billion of contracts with AI companies. So lots of growth available there. The other area is to -- and final point is to move our Oracle Database customers to OCI. They've patiently been waiting for us to catch up in this market, mainly because when they've tried to move these workloads to other clouds, they just don't have the experience and know-how that Oracle does. And so now that we're where we're at, we expect that many of these workload migrations to actually accelerate going forward. So we feel great about where we're at from a strategy standpoint. On the financials front, again, I mentioned, we're a little further behind. So as you see, we moved from 3% of revenue being in the cloud and the infrastructure space in FY '15, and now it's about 20% as we exit FY '23. So we're really still in the early innings of this opportunity with lots of headroom in front of us. Now getting bigger requires going global. Clay mentioned the 64 data centers that we have. It's sort of a breakdown of it, and I'll just walk you through it. So we have 44 public cloud regions and other 6 that are in the process of being built. 12 of these public cloud regions interconnect with Azure in order to give the multi-cloud capabilities. We also have 9 dedicated regions with 11 more planned, and we have 9 national security regions and 2 EU sovereign regions and again with increasing demand for each of those. So that, in aggregate, gets you the 64 OCI data centers, but that doesn't include all the cloud and customer implementations that we have. It's really that sizing flexibility, that deployment optionality that really give us unique advantages in the marketplace. Okay. And I mentioned the ecosystem. So again, in addition to having a global network of data centers, it's important in the cloud as an infrastructure provider to have a rich and extensive ecosystem. So we talked a lot about the Microsoft announcement already. But as you can see, we have examples of really strategic partnerships across a wide range of areas, including things like Amdocs on an application standpoint or Commvault from a PaaS standpoint. So we'll continue to expand these relationships as we go forward, working with customers to identify as they think about the workloads, what are the critical partners that we need to integrate and make available to you in order to both do on OCI or do through multi-cloud environments. And then in terms of customer traction, I talked a lot about these names, digital native, artificial intelligence-based companies, they're all here at the conference this week. So hopefully, you got a chance to hear and see from them as to why they're making such large investments in the Oracle Cloud. And as Safra said on our last earnings call, the best spokespeople for our products are our customers themselves. So that's what doing these events like Oracle CloudWorld here globally as well as then the regional events that we'll be launching over the next few months are so important to bring and connect our customers together in a way that they can learn from each other and know that going with Oracle is a very safe bet. And then finally, the biggest opportunity that remains in front of us is to capture the move of our on-premise Oracle workloads as they move to the cloud. These customers have been patiently waiting for us. As you can see, less than 2% of our support installed base and infrastructure has fully moved to the cloud. So again, when we look at -- and last year, I talked about a $4 to $5 conversion rate opportunity, that shows at least a $55 billion opportunity in moving our customers. And sometimes, we get even more than that. There's a lot of workload types where they need additional cloud capabilities, and we can get multiples of those dollars on top of it as the customers make their move of these mission-critical applications into the public cloud. So again, if I roll it all together, when we look at the cloud infrastructure potential, it's about twice the size of the applications opportunity. And again, this is near term, identifiable opportunities, not just TAM. And given the hyper rate -- hyper growth rates that we're experiencing in the infrastructure space, we would expect that the infrastructure cloud will surpass the applications cloud within the next few years. Okay. Let me try and wrap this all together and give you a sense of where we're at. So as Safra talked about last week at our earnings call, we are confident in our ability to achieve those FY '26 goals that I went through again at the start of the presentation. Getting there require 3 things. First, we've got to expand our leadership in cloud applications. Second, we've got to move up in that race for leadership in cloud infrastructure. And third, we've got to grow the bottom line as we scale. I think we've got a really good idea of what we're doing. And hopefully, course of today, you've got a sense of how we're going to get there. On the revenue side of the goal, as you can see here, we're trying to get to $65 billion in revenue. That means we need another $15 billion of incremental revenue compared to where we are today. And when you add up all those categories of revenue opportunity that I described that we see line of sight to, it's about $350 billion, which means we have to capture less than 5% of what's really available to Oracle in order to achieve that. It seems quite available to us. And on the profitability side of this goal, you can say it's a little bit different. So on the profitability side, that's the one we feel a lot more confident about because we control this. We know this. We've been doing this for decades. It's why customers ask for the Oracle playbook, as Jason talked about, because we can take all of our learnings in terms of how we run our business processes on our systems in order to both drive revenue growth as well as drive margins with that or profitable growth. So in order to get to that $2 billion, which is really the gap to go from a 42% margin at $65 billion, which is 42% where we're at today or in FY '23. So the $65 billion at the 45% margin is another $2 billion of efficiencies. We're going to get there through 4 things. First, we're going to focus on increasing our cloud margins as we grow and fill out our existing data centers. We've got to plan for that. Second, we're going to continue to identify opportunities across the biggest areas of spend outside of our data center, which is our sales and engineering teams to continue to be efficient and make sure that we get the most out of those organizations. Third, we're going to look at all of the rest of the business in terms of how do our business support organization utilize AI, analytics, and other tools in order to drive efficiencies so we can remove manual labor where possible, move to higher-level activities and save costs. And fourth, it might be the most important because we're going to get to that later, which is how do we convert Cerner from a services-based organization to an IP-based organization. We're well on our way, and we feel confident about where that's going to result. And again, as Safra has talked about in the past, we expect Cerner to move to our overall contribution margins as we do that. Okay. Before I wrap up, let me just give you a little bit more incremental data points to give you confidence in where we're at. So I'm going to build upon what Safra talked about our business in the earnings call last week. So she indicated our commitment to accelerating total revenue growth this fiscal year 2024, excluding Cerner. You guys all heard that. As you can see from our slide, our current expectation on an organic basis, excluding Cerner, is that we'll grow 8% this year, so another uptick in our overall growth. But I also want to comment on what happens with Cerner. So if you include Cerner, we expect our revenue to grow about 7% this fiscal year. In addition, we expect that our non-GAAP operating margin will tick up to 43% this year as well. So again, this is all in the aspect of building upon our direction to reach the FY '26 goals we talked about. So all in, just to reiterate, we remain committed to those long-range targets. I think what we'll continue to do is show you how we progress, how we're progressing towards hitting those goals and, I think, hopefully exceeding those goals as we go and as the next 3 years unfold. So with that, thank you.

Okay. We're going to -- mic on, please. Sorry about that. We're going to go ahead and take a bit of a break here. Lunch will be outside. We've got 25 minutes for that. And then when we come back from break, we're going to have Mike Sicilia present next. Thank you. We'll talk to you in 25 minutes. [Break]

Okay. I'm back to talk about health care. The disclaimer hasn't changed. The rules haven't changed. Yes, [ as we go ]. So let me start here at the beginning what compelled us to make the Cerner acquisition. So during the early days of COVID, we work with the United States government, particularly the CDC and the NIH, in which case, we built and donated many APEX applications to help the government manage COVID. They were national vaccine databases, patient-facing -- patient-reported outcome adverse event reporting things and an extension to the government's SAP system that wasn't able to be put out onto the public cloud to allow providers to actually order supplies from the government, vaccines, PPE, things like that as well as working with CMS, the Centers for Medicare and Medicaid, to look through electronic health records to see if there was any medication, anything that was efficacious or having some luck with COVID-19. There were several problems that we discovered in that. The first thing is, notice I said we built all these applications during COVID because none of them existed. There was nothing that existed to get real-time information for vaccines, for example. So for example, there are 96 different jurisdictions in the United States that administer vaccines, 96 different central hubs, if you will, called central hubs, that literally federate thousands of endpoints, be they pharmacies, hospitals that are giving up vaccines. There was no way for the government be able to see in real time. And from a practical standpoint, we couldn't go switch out all -- thousands of systems all at once to able to get real-time telemetry into what's happening with vaccines in the United States. So we built a system to do that. We're COVID screening for the [indiscernible] called operation [indiscernible] clinical trials. There were a couple of big fears. Number one, we wouldn't get enough people to volunteer for the clinical trial. Number two, the amount of people that volunteered would not be diverse enough to meet the diversity, equity and inclusion requirements for the clinical trial. So we built a system that prescreened 650,000 Americans who volunteered to enter the clinical trials. We placed those into 88 different live sites, and not only were we able to meet the diversity, equity and inclusion requirements, but we doubled the requirement for diversity, equity and inclusion. Again, a system, electronic prescreening for clinical trials that did not exist. So the story continues to go on. There are lots of these different examples of things that we built. And our fundamental thesis was the state of health care and the state of the IT systems that make up health care was completely broken. It was completely broken. And I think COVID explained -- sorry, exposed rather the fact that there is little to no real-time information. When we work with the Centers for Medicare and Medicaid to be able to mine the electronic health records, not that we could see anybody's personal health information but look for patterns, that was a few months into COVID. There was one glaring problem in the data. There was no COVID data -- COVID patient data in the systems because the systems only get refreshed every 6 months. So 3 to 4 months into a pandemic, we had very little central telemetry into what was happening with patients who were hospitalized with COVID-19. We decided there must be a better way. There must be a way to really turn this into a real-time system, just like our systems for property management that I talked about earlier. When you checked into your hotel here on the Vegas Strip, chances are you interface with the Oracle property management system. That system knows a lot about you. It knows the last time you stayed here. It knows what your preferences were. It knows what you ordered for room service, and it processes your credit card payment and knows your payment history, right? When you take that same credit card and use -- and run it over our core banking systems anywhere in the world and you go swipe it, it knows that you have enough credit to be able to buy what you're trying to buy. Our hospital systems don't work that way. When you try to take your electronic health record with you, you try -- it doesn't even have to be electronic. You try to take your health record with you, it's a cumbersome process. So we decided that we had lots of technology that we deployed during COVID that we could augment the current health system with, and we could put in stop-gap solutions like we help the United States government put in and governments throughout the world. But in order to really change health care in order to really get to the core of this problem, we needed to own a core asset. Cerner was that core asset for several reasons. First and foremost, Cerner is the largest electronic health record system in the world. Cerner has always been a global company. Oracle has always been a global company. We like global leaders, and Cerner has a commanding lead on the pack in terms of global market share. Epic is bigger inside the United States by a little bit, but globally, Cerner is the biggest. The other thing that was really interesting is that Cerner is built on top of the Oracle Database. At the time we acquired Cerner, it was running an Oracle 19C database. That was just -- as a reminder, 1.25 years ago. So we decided that what we wanted to do, and I'm going to talk about the investments we're making, was to make sure that as we put in revolutionary -- evolutionary technology and as we roll that out to our customers, we didn't want to put them through a data migration process or data upgrade process. Hospitals today are running on limited margins. I met with one of the marquee hospital systems last week in the Washington, D.C. area, and they were thrilled to eke out 1.1% margin last year. That's not uncommon. Major hospital systems like the Cleveland Clinic lost money last year. There's not a tolerance for major IT transformation projects, even though they know that it will help them get better. Even though we're short almost 1 million nurses in the United States and we can't get there without virtual nursing, there isn't enough money to be able to go through a major reimplementation. So the trick is we have to figure out a way where we can deliver new technology where our customers can spend less on IT and not more on IT. And that's what we're focused on. The fact that Cerner runs on the Oracle Database gives us a tremendous advantage. It means that we're layering in new technology, and I'm going to show you some of this here in a bit. We're layering in new modules, which are now available written in APEX using generative AI, using voice recognition right on top of the existing schema. There is no data migration for the customer. They're not moving anywhere. There's no data migration. There are no patient privacy issues. There are no HIPAA complaints. All of these things that typically happen when you have to move from one system to another go away. In addition to being a popular choice inside hospital systems, Cerner is also a very popular choice for governments, both the United States and rest of the world. We have very large government customers around the world. Clearly, the United States Department of Defense today, it has 165,000 users. They announced this week, and we had the United States Department of Defense and the federal electronic medical record governance here speaking at the Oracle Health Conference. They'll add a mere 10,000 users to that this weekend -- just this weekend. So these are major government implementations for electronic health records. Same thing happened in Sweden; in the U.K. with the trust; in Australia; in the Middle East, both in the UAE, Qatar and Saudi Arabia. So it's not only that we're dealing with hospital systems, but we're also dealing with countries. And when you start dealing with countries, you almost never see Epic. You almost never see the #2 when you start dealing at the country level. The problem is Epic is written and runs on top of the MUMPS database, as I said before, which was built in 1965. Not so scalable, not so great, not so secure. So the trust that we put in and the fact that we can deliver the entire stack, the electronic health record, the human resources, the ERP, the supply chain the patient administration, the patient billing, the revenue cycle management, the health insurance suite, all together as a single cloud solution on top of OCI allows us to have discussions with not just hospitals but with countries. So now as we've digested Cerner, we've taken the first year of ownership of Cerner. Again, we just crossed the first year of ownership of Cerner. And we've started to think about 4 -- not starting to think about. We're actively working on 4 major initiatives. The first is simplifying the health care process, using generative AI and voice navigation and also, as Steve mentioned earlier, adding in health care-specific features and our supply chain management system and our HR system. What we did with the Cerner acquisition from an R&D perspective is very different than what we've done with any other acquisition. We've taken the engineering teams from Cerner and, if you will, spread them across Oracle. So we have a clinical team that works on the core EMR. We've taken expertise from Steve Miranda's HCM team and a supply chain team, and we've taken Cerner engineers. They now work for Steve to create the vertically specific features for supply chain and HR. Again, as an example, I talked about it earlier today, in case you weren't here, we have a system to be able to say, well, based on the number of patients that have checked into the ER, we can predict how many nurses you may need in the ICU 24 hours from now or 48 hours from now because we can also predict, based on statistical analysis, how many of those are likely to progress to intensive care. That's very different than the customer having to take their EHR, having to take their HR system, having to take their time and labor system and stitch all that stuff together. And by the way, we can also tell you whether you've got enough staff to be able to staff that ER. And if you don't, here are all the nursing pools, here are all the contract organizations that are available for you to go get temporary help as you need them. The second area of focus is modernizing the Cerner applications. Another big benefit that Cerner had in addition to running on the Oracle Database was that almost 90% of all Cerner customers [ run ] as a managed cloud service, [ were unhosted ]. So when we're lifting and shifting, of course, all of those Cerner businesses and all of those Cerner applications and all those Cerner customers to OCI, just like we've now done for all of our other vertical businesses that I spoke about earlier today, we're really changing out the computers with much better computers, which are the OCI computers, the Exadata cloud services, the autonomous database and all of those things. All of that's happening right now, and I'll speak about the progress in a second. The other piece is federating disjointed data sets. Larry, in his keynote last year, said that we're going to build national databases. We're going to build applications at scale, and we're going to build a system that can handle and be tailored to the need of the country. We've delivered the technology for the national databases. We've taken the Cerner HealtheIntent application, which was a data analytics platform, ported it to OCI, reengineered all of the components, running this on autonomous data warehouse, and now it is ready to scale for the needs of a country. This is a very important strategy for us and, I think, a huge differentiator. If we think about -- I'll go back to the problem that I presented initially, which is health care is disjointed, tons of systems, some bespoke, some off the shelf, some written in 1965. It's a smattering of things. And things like COVID exposed the need for public health officials to have real-time information. It's impractical to think that even with all the brand new modern technology that we're putting into the Cerner applications that everybody could switch out all the input systems at once. It's just impractical to think that. What is very practical and what we're doing is saying, we'll meet you where you're at. We'll meet you where you're at, and that's exactly what we did to create the national vaccine database for the CDC for the United States government. All 96 of those systems, which are central hubs for their region, that's all 50 states, plus territories, plus special regions like New York City versus the State of New York. All of them today feed, in real time, a single national database running on autonomous serverless, running on the Oracle Cloud without changing out any of the input systems. Now of course, over time, our strategy is to introduce the generative AI, voice navigation systems to, in fact, replace out those systems. But I don't think we're going to be able to fix health care by insisting that everybody rip and replace everything they have to make it better. And until now, that's been the strategy of a lot of the pillar vendors. They're saying, "We've got the next best thing, but you got to rip out everything that you have." At a foundational level for our server customers, we will never say that. We will say your data stays. Your data structure stays, your schema stays, we'll never change it. The other thing that's really become interesting as we think about the full stack is we're delivering Cerner, ERP and OCI across customers. Cerner customers are picking Fusion ERP. Cerner customers are stopping their Workday evaluations for ERP and HCM and picking Fusion, okay? Cerner customers are moving their bespoke workloads to OCI. Cerner customers who, for some reason -- usually, M&A is a driver why people switch out EHR systems because switching out your HR system is more complicated than switching out your ERP system. Cerner customers are pausing their decisions to go to competitors because they like the Fusion SCM story, because they like the HCM story, because we can have discussions, as I said, with the CEO and the Board, which is this, how much do you spend on IT? We're willing to do all of it. And that's a very differentiated story versus our competitors. And of course, as we've added in the Cerner -- as we embedded in, rather, the Oracle engineering talent to the Cerner team, we get, we think, tremendous margin efficiency certainly in the gross margins and moving a hosted services-heavy business to a true cloud business. And obviously, a lot of third-party decisions that Cerner made, we're in the process of replacing with Oracle technology. So let me talk about, if you will, one of the new Oracle technologies. We unveiled this week called the Oracle health clinical digital assistant. Big complaint that we get from providers, doctors, nurses, radiologists and stuff is a lot of clicks in the system, it's -- and this is true of the HRs in general. This is not just a server thing. A lot of clicks in the system takes me a lot of time. I end up, frankly, not doing as much of it at the point of care at the time as I want to with the patient because, otherwise, I can't pay attention to the patient. I have my head in the screen the whole time. So they do what's called the pajama time. They do this at night by themselves, and they're up till 11:00, 12:00 at night, and this has been a fuel for a lot of highly qualified providers to leave traditional medicine and move to concierge medicine, which, I think, potentially creates an equity problem and a health equity problem for the world because the systems are just too cumbersome to use. So the initial feedback we got was if you could just go from 10 clicks to 7 clicks, it would be tremendous. That would be really tremendous. We started to experiment with large language models and our Oracle Digital Assistant, not just for voice recognition, for dictation but voice as a navigation mechanism. And now we just rolled out, we announced on Monday in the health conference that was here, and our customers will start uptaking this in limited availability in the month of October, a system that goes from 10 clicks to 0 clicks. It goes from log on using user name and password to voice identification and biometric identification. If you think about I witnessed this too as I've been able to work with some of our customers, our providers, you go into a patient room, you put your gloves on, you're about to see the patient, we forgot to log into the system. You got to take your gloves back off, you're going to get log into the system, you got to put your gloves back on. If you just say, "Hey, Oracle, wake up," it's a much better paradigm. And that's what we unveiled, and that's what we've delivered. And by the way, it works right on the mobile device of the provider. This is a mobile device supplied by the provider. You can see, as I'm walking to the patient, I say, show me my next appointment, shows me the next appointment. Show me Mary's lab results, tells me her lab results and everything comes up on my phone. It reads them to me if I'd like to. At the end, if the patient can sense we're listening, so-called ambient listener or background listening, to the entire encounter between the patient and the provider, everything is automatically encoded in the electronic health record, and the doctor can give voice commands to put in orders, dosage, how many times they want to do this in the number of refills, then electronically sign the order with their voice. This is incredibly different than the way the systems work today. There is zero interaction with a computer screen, using a mobile device as a listener. There's no typing. There's no log on. There's no user name. There's no password. We know who you are, we know where you are, and we know what you want to do for the patient. So we're talking about a system that's revolutionary, not evolutionary. Instead of going from 10 clicks to 7 clicks, we're going to an entirely different user interface. As you can see here, the doctors ask to see some x-rays, ask to see some images. Of course, if you want to do this on a tablet, if you want to do this on -- wanted to use a large screen, you're more than welcome to do this, the technology completely, but it's completely portable, completely works. And you take the EHR to the patient rather than having to swivel-chair into the EHR every time you're dealing with the patient. So this was met with great fanfare this week for our customers to reduce and eliminate pajama time. I mentioned before the importance of OCI for several reasons. Number one, health care is, as you well know, a big target for bad actors. And you often see that health care systems, particularly small and medium-sized health care systems, are frequently compromised and have no choices but to comply with the demands of the bad actors to be able to get their data back. Not such a good situation. It's tough when it happens to a casino, even tougher than when it happens to a community hospital. Moving to OCI. As we know, and we've said is the most secure cloud in the world, obviously provides us tremendous economies. We will have hundreds, hundreds of Cerner customers move to OCI by the end of this calendar year. We have moved every one of our other applications in our other vertical businesses. There are 43 acquisitions that make up our global businesses line, including Cerner. 100% of that runs on OCI. 100% of that came from somewhere other than OCI. Now it all runs on OCI. Hundreds of customers running this year in OCI. When we move applications to OCI -- other vertical applications to OCI, without changing any of the code, without doing anything else, the worst we've ever seen is a 10% performance improvement. That's the worst we've ever seen. In many cases, for long-running transactions, we see 40% to 50% improvements simply because the computers are better. They're more powerful. There's a lot more that happens in memory. That's without changing any of the code. I predict Cerner will get very big benefit because we're actually moving the Oracle Database, which runs really well in the Oracle Cloud with it. So I'm very excited to see the performance benchmarks that we roll out to our customers here as we make progress on that. That's customers of all sizes. That's individual doctors' offices. That's small community hospitals, very large systems, very large systems, some of them in the Middle East. Again, as a reminder, government is a customer outside the United States, inside the United States, too, but obviously, government is the primary customer outside the United States as health care is a government region. It's impossible to have a conversation with the government without talking about data sovereignty. And we're not talking about [ full ] data sovereignty, where the primary region is in the country and the DR is somewhere else. To give you a full data sovereignty, full active-active, full replication in the country. So we're delivering a full stack for governments throughout the world on top of our dedicated infrastructure for government. So modernizing the health applications, user-centric design, self-driving, self-preparing, customer benefits, easier to use, less human error, no pajama time for governments and for public health officials, even for our existing health systems who have multiple instances of EMRs. It's actually not uncommon because of mergers and acquisitions for a "Cerner customer" to also be running competitive EMRs as well because they've acquired hospital systems and they have them. There's no way today without our single-patient data platform, our health data intelligence platform, as Larry referred to it in his keynote, to be able to see what's happening across their entire system. Very difficult, and they spend a lot of time and money on manual process to do things like have people set appointments across providers. If they happen to be at one hospital or another hospital, it's a really tricky process. We're solving all of that with our APEX-driven registration scheduling across our self-driving patient data platform. AI and ML. Obviously, there are huge benefits here to infusing large language models into the Cerner applications in a couple of ways, and we're doing those in a couple of ways. The first is to automate routine tasks. So if you've ever had the experience, at least in the United States, of going to the emergency room for a procedure, usually, the single longest task is waiting for your discharge notes unless you're very sick and you have to go somewhere else, which hopefully not. But if you go in there and they say, "Oh, you kind of look okay, like I happen to do and like stop complaining, and we'll be with you in a bit," usually, it takes sometimes a couple of hours to get your discharge notes. And your discharge notes when you read them, largely, you think I could have wrote this myself because this is exactly what the doctor told me to do. But what happens is the nurse or somebody else, they have to go to the computer. They have to go type it all in. They have to get it out of the system, and if you had an x-ray, well, that goes into a different system. If you had an MRI, that's in a different system. If you had radiology results, that's in a different system. If they ordered a medication for you to take home with you, that might be in the same system. Unless you're going to pick it up with the pharmacy near your house, then that's in a different system, too. So that's what takes so long to get the paperwork together to allow you to leave. Now you could, of course, just walk up and leave, but most people don't do that. That's what has to happen. This is a tremendous opportunity, and this is what we've rolled out. Again, limited availability for our customers starting next month to be able to automate that entire process. Now what we're not doing though is replacing the doctor. We're not replacing the nurse. We still have to have what we call a provider or a human in the loop. We're creating draft notes to say, we think this is what you should give the patient from a discharge perspective. We think this is what you should give the patient from an after-visit summary, but somebody has to approve that. Somebody has to say, yes, I approve, or no, I don't approve. Based on the listening, based on what we're doing, we will highlight things that are fuzzy, things that we didn't hear, things that may not be as clear as we want to be, and we'll do that. We're using Cohere large language models for health care to do that. And I think that some customers even could bring some of their own large language models as well. Some major big university-led hospitals are creating their own large language models. The next big piece for AI is what I call clinical recommendations and thinking about specialty recommendations. I had this experience with a -- the CEO of a very large health care system in the South. We spoke about AI. We spoke about what his needs were. He said, here's the problem. If you come into our hospitals here between Monday to Friday, between 7 in the morning and 7 at night, you get pretty good care. And the readmission rate is actually very low. If you come in, in the middle of the night, or you come in on the weekend, our readmission rate is much higher. It's much higher. What I need is the system to make clinical recommendations to those doctors that are working the late shifts or the weekend shifts so that they have the same amount of data and the same access to information as the doctors that are on sort of the prime time hours, if you will. That, I think, is the next big piece, and that's exactly what we're working on. Now I think what that requires and where the investment will come will be from specialty large language models. You want to have specially large language models. Steve mentioned hallucinations as a problem. And in health care, if you look at -- people are trying to use ChatGPT and other things to self-diagnosed, a little bit dangerous, I think. You want to make sure that the cohort of data and the cohort of patient data that you're pulling from is as succinct as -- only as succinct as you want it to be. You don't want to have too much data. So for example, if we're dealing with a pediatric infection or a pediatric center, they -- we want to have a specially large language model that only deals with pediatric data for that particular pathogen, that particular disease. That's what we're working on now. I think that's the next big area to be able to infuse AI. Now again, what's really interesting, if you look at the competitive landscape here for this is we are going to embed that AI as a feature right into the application set. If you look at some of the other things in the market today, the doctor would have to go write a query or somebody or the IT informatics team would have to go write a query to go do that, that's not our strategy. Our strategy is that AI, be it generative AI, be it machine learning, is embedded right into the application stack, and it's a feature in the clinical workflow. And we're excited to be able to roll that out on the very same cadence that we now roll our Fusion applications out of that. If you look at what's happened with the average age of EMRs and the average life span of the EHRs today, be it Cerner or any of the competitors, many of them are running very old features, very old functions. So imagine now taking a system of record which largely HRs today are because they were graded by billers and turning them into a system of intelligence by looking at all the data that's in the system and helping medical providers make better predictions and create better outcomes for both patients and providers. Again, I mentioned earlier, we're seeing a lot of traction in our health care customers, looking at the Oracle full stack, whether they're running existing Cerner applications and coming to Fusion ERP and OCI or whether they're running Fusion ERP and OCI and coming to the Oracle health applications. The electronic medical record, national databases, the CareWare real-time IoT services, that's traction, that's real, and it's in the market. Here are a couple of examples. Here we go, a couple of examples. Ascension health care, a very large leading nonprofit health system in the United States, existing Oracle health customer, selected Oracle OCI, SCM and ERP, existing Cerner customer, very large Cerner customer, great customer, great partnership, full-stack solution, having the conversation, as I said, at the CEO level, at the executive level, how can we do all of what you do for less? Great example. Children's National in Washington, D.C., it was an existing Cerner customer, also selected OCI for their bespoke workloads, for lift and shift of all their bespoke workloads and now are applying Oracle's AI and machine learning capabilities to research efforts for pediatrics. And if you think about being able to really change the world and do something impactful, we're also in the clinical trials business. We have -- I mentioned earlier, we have a large clinical trials business, the largest pharmacovigilance business. And over the last year and quarter, I've met with a lot of our pediatric customers. And I didn't realize this, but pediatric medicine today is very underserved by pharma companies. Pharma companies are hesitant to do clinical trials and experimentations on children, which are perceived liability, and frankly, just from a sheer numbers perspective, children happen to be healthier than adults. So there's a lot of research, and there's a lot of stuff that doesn't happen for children because for lots of reasons, some of it good, some of it not so good. But there's a lot of data. There's a lot of data out there. And turning the data into an input for a virtual clinical trial would be amazing, right? Turning the data for what's called -- to be able to do what's called bedside trials. So in pediatric hospitals today -- well, let me say this, for clinical trials today, only 3% of people that are eligible for clinical trials actually sign up. That's it. Only 3% of providers actually know the clinical trials available. In pediatric customers, these are not easy things to talk about, but I think this is a technology that could help. If a child was sick and they're off for a clinical trial, the sign-up rate is sky-high. People will do anything they can do to try to save that child. Parents will participate. The problem is there just aren't enough of them. And they're not run in the same way like a centralized double-blind, placebo clinical trial. It's almost what's called always off-label, which means we're going to try something that we think may work. We can stitch together a network of pediatric hospitals throughout the world. And because in any given pediatric hospitals, just because of the sheer low numbers of children who get sick, you may only have 1 or 2 patients, may only have 1 or 2 people. But if you did that across the world or you did that across the country, you have enough data, and you have enough efficacious study to be able to indicate whether or not there's off-label medications, meaning an existing medication may work for somebody else. I personally think in working with our -- working with Children's National, exactly the type of things we're working on and others throughout our pediatric installed base, this is mind-blowing. And this kind of stuff here has changed Oracle. It's changed us to become a mission-driven organization. I've got more people inside Oracle that want to sign up to work on the AI and ML for pediatric hospitals and probably any other feature set we have going on inside the company, as you might imagine. There's nothing more gratifying than hearing from our customers, and we heard this recently in the Middle East, that some of this technology -- this is posted on LinkedIn from one of our directors, some of this technology actually saved a 7-day old baby. And this is what, I think, can really change hearts and minds in health care. Coming back to a more -- maybe less emotional thing, we're seeing lots of wins of SAP customers as well. In Germany, Sana recently selected Oracle health EHR, the electronic health record Cerner and HCM. They're an existing SAP customer, moving from SAP to our full stack and digitizing all their health records to reduce physician and nurse burnout. So the EHR, the electronic health record, is the mission-critical system for hospital systems or government systems. We're able to, we think, really package and deliver a comprehensive solution between electronic health records, HCM, supply chain and then, of course, the rest of ERP as well. No other vendor on the market is delivering that comprehensive integrated suite. No other vendor is delivering that as a cloud service. No other vendor says, we'll update for that for you automatically, 4 times a year and do all the security patching for you as well in an area where supply chain -- where cybersecurity is so precious. Okay. Another thing that's changed over the last year with system integrators. So [ SER ] philosophy was to prime a lot of their implementations. And that was a philosophy -- it was a cultural philosophy, and that -- not necessarily would have been Oracle's philosophy. So you're seeing tremendous rallying from a lot of them were here this week at the Oracle Health Conference, sponsored the Oracle Health Conference and presented in system integrators, large system integrators, scaling up their Cerner implementation practice. Of course, it makes a lot of sense for them because they're already implementing Fusion HCM. They're already implementing Fusion ERP. So they're adding in skill sets on the clinical side to add capability to deliver that full stack. I think that has a tremendous ability to drag us into markets that we otherwise may not be in. And of course, continue to invest and deliver our full stack solutions, we think, are -- which are unmatched from a cost per money perspective. From a sales and influence perspective, we've had one -- enjoyed a wonderful partnership with the Tony Blair Institute for Global Change in Oxford University. We're working on a pathogen analysis system with Oxford University, which essentially allows pathogen analysis as a service. So if you check into a small community hospital today and you were recently traveling abroad, recently went to some places where there's some tropical diseases, they may take a blood sample. They may take a sputum sample and may not know what to do with it because it's not something they've ever seen before. This happens quite frequently. With Oxford Nanopore which is a portable pathogen sequencer, a portable genomic sequencer. We can take that sample. And within 20 minutes using OCI, we can identify where it came from. The identification today of these things, this is what you see in these community hospitals if you're ever driving. You see those little metal boxes sitting outside full with samples and somebody is going to go ship it overnight and you might get an answer in 2 or 3 days for some rare tropical disease. That should be cloud service, we think. And you should be able to diagnose somebody at the point of care, exactly what we're working on with Oxford and others around the global pathogen analysis system. And of course, as I mentioned, government being a rich opportunity for us outside the United States for health care, we, of course, sell and deliver lots of technology to government. So internally at Oracle, I'd say we have a tremendous teamwork and collaboration happening in health care with our public sector teams outside the United States. We still continue to do strategic advisory implementations and managed services. These are huge differentiators. And this is a change for Oracle. So customer service delivery, interesting. Of Cerner's employees that we have at Oracle now 1,500 of them have a medical degree. They're either a doctor, a nurse, a respiratory therapist. Many of them were here this week at the Oracle -- Oracle Health Conference. We still provide strategic advisory services, change management services, medical encoding services to be able to help our customers understand what we think is best and how to take advantage of those opportunities, where I think the SIs will team with us is on the implementations, the integrations and all those things. But it's very important us -- for us to be a strategic adviser, particularly in a vertical industry that's so far behind on the technology curve, because what they don't quite understand sometimes is the art of the possible. Again, I used the example, the feedback was, can you help us go from 10 clicks to 7 clicks. Not understanding that it was possible to go from 10 clicks to 0 clicks. All right. And then finally, as we think about the health efficiencies -- the efficiencies of the service business, this is where we think we will continue to get. We've gotten -- as Doug said, we're very happy with the progress we've made. We'll continue to get more efficiencies in services and cloud automation, reducing implementation cycles, obviously, accelerating go-lives with true cloud technology and where possible, replacing Oracle technologies -- with Oracle technologies, lots of third-party technologies. And that just makes the stack much less expensive to maintain, much less expensive to upgrade and that's going quite well. I think you saw a good example of that this week with the Oracle Clinical Digital Assistant, which wraps a bunch of third-party technology that we just don't need anymore. We have wound down a lot of the -- few of the low-margin businesses in Oracle -- or maybe another way to think of it is we can operate some of these low-margin businesses at a much better clip and turn them into margin businesses that would be representative of an Oracle business. We're moving away from adjacent noncore services businesses. There were some things, some general services and stuff that Cerner was providing to their customers since they tend to be the biggest operator, but those things. We'd -- much rather focus on growing the software business, gross margins in the cloud and operating this like we operate all of our other vertical businesses. We've consolidated facilities. We're moving through operational integration of the Cerner business so that we move all the existing Cerner back office systems to the Oracle systems and then reducing the duplicates stack. So lots of things going on here, lots of technology investment. We spent a lot of time thinking about modernizing the stack, creating better experiences. But I would say that above all that and all this stuff, for us, it's been a terrific opportunity to impact humanity. And it's probably been the last year and quarter have been certainly the most humbling experience I've had with customers, looking through and looking at how they apply this technology to help people. So we are very fired up. The Oracle Health team is very, very excited. We had a terrific Oracle Health conference here this week, led by many of our customers. And I'm very happy with where we are, and we'll keep giving you all the news and all the updates as we make progress. Thank you very much.

Please welcome to the stage, Safra Catz.

Okay. As Safra gets herself seated. We'll go ahead in the Q&A portion of the meeting. If you have any questions you'd like to ask, the easiest way to do it, send me a text, send me an e-mail and then where you're sitting, and I'll come find you. Okay? .

Just remember, Larry is coming after me. So some of you are going to want to save your questions for him.

Great. Safra, Brad Zelnick, Deutsche Bank right up here.

There you are, right in front of me, Brad.

Safra, I have the utmost confidence that Oracle will achieve the margin targets that Doug laid out for us, knowing your history, I think there's no doubt. But at the same time, with all the innovation and opportunity on display this week and what's been reviewed for us today, and Larry telling us that this is essentially a watershed moment in computing history. How can we be sure that at the same time, you're still able to invest enough to maintain leadership and to really be a leader in this generative AI era in the years to come?

Yes. Well, we have a pretty good spending envelope at this point. And we have very good line of sight on a lot of the business because this isn't like every quarter, we've got to think about the license and all. There is so much business already there. And what we keep seeing in our RPOs, by the way, is that our customers are burning through it faster. And so all of this has been happening really in the past couple of years. And so we feel like you should not worry that we're being cheap and -- which we are kind of cheap, we're known for cheap. But listen, it's your money. So it's very important, and you're advising -- are owners, but simultaneously imagine some of the stuff you saw here, we've been investing within the envelope. If some of you saw the policing and all that local, that's all been happening in that R&D envelope. And so we've got the R&D envelope, we have the sales and marketing envelope. And we've got our eye on where the growth is, and we have a pretty good sense of how the money is coming in. So as we scale, we have massive economies of scale. We have line of sight on savings. You just saw Mike just very briefly cover a bunch of the Cerner savings while rewriting all the products. And we're just expanding everywhere. So I feel very good about how much we're spending. And the truth is between us, many of the technical teams did not even really need to spend to their budgets and probably won't. So they're not starving at all. We're not pulling in because of our concern here, we just have a lot of momentum on the revenue side, and we've got plenty of room on the spending side.

Okay. We'll next go to Karl Keirstead.

Right here, Safra. Safra, I'd love to ask you about the cloud infrastructure performance. 64% pretty remarkable relative to all of your cloud infrastructure peers came right down the middle in terms of your guidance. But frankly, I think the stock reaction the day after your print had a little bit to do with just given the incredible $4 billion of commitments that you announced, Uber ramping, we expected maybe a little bit more umph in that number. I wanted to ask you, what's -- how do you respond to that reaction? And were there any bottlenecks in converting that backlog to revs? Or was it sort of normal backlog to conversion timing?

Well, a couple of things. I see buying opportunities when the price goes down. So if no one else wants it we continue to go private 1 share at a time, as I told you, hundreds of times. So there's no surprise. The -- I can never outguess the market. Larry and I have actually like a perfect record of being wrong on how the stock is going to react. Luckily, you're in that business not us. We're just in the business of building systems and making them available to customers. For us, we are overwhelmed by demand. There's no question. We are expanding out. We came in exactly where I said we would and I thought we would. What happens in all the spreadsheets that are done by investors and by many of you I can't really control. But I can tell you that pretty much everywhere I've told you we're going, we either go there or do better. And I have to just make sure we can roll out and meet the massive amount of demand that is falling on us. But things do take time, including rollouts and making computer compute and GPUs available and we're really happy here. I mean I don't know if you got a sense of what's going on here at Cloud world. But I can tell you whether it was ranged meetings in advance or sort of serendipitous meetings in the hallway. People cannot get to us fast enough. That's really what is happening. They are -- I mean I think Clay mentioned that he ran -- I ran into a customer who's actually demanded a contract by October 1. Demanded. I've never had that before. You would say usually, it's us chasing customers around whole new ball game. So we're very, very satisfied about how we're playing out this year, and we expect a continued acceleration for the year and potentially better. But we got to deliver and we just play it out.

It's John DiFucci from Guggenheim. Doug's presentation is really helpful when you're thinking about margin and profit. But when I think about free cash flow and I think about CapEx, and I'm just trying to figure out that longer term, like -- and I'm not talking next year or the year after. I mean like when you're steady state, when you're not spending to sort of catch up the demand, how should we think about that as a percentage of revenue, even sort of a rough range?

Well, as you saw even this year, we're already starting to throw off cash. This business, when we put in the cloud, the economies of scale are so stunning. And you also have to understand our sales expense, which historically is a pretty big number, just drops to the floor when all it is a renewal. So think about for you all for modeling purposes, you know what the support business looks like, right? And you know what the license business looked like. It's somewhere over there. And that because it's no longer -- yes, there will be new contracts and sales expenses and chasing people around and negotiation, all that takes both time and money and personnel. But when you're just renewing and expanding and when so much of it is automated, the money -- I said this before and I ended up seeing it in the lawsuit, so I'm not going to say it again. The profitability translates into cash and simultaneously saves our customers so much money because they benefit not from their own scale, but from the scale of the entire network of customers and the automation that we built in to many of the services and OCI itself has already shown up as our competitors. Imagine our competitors are showing up because their cost of goods sold is higher than our retail price for our offering. Do you understand how that translates. And so free cash flow is inevitable because everyone is benefiting simultaneously.

Safra, Michael Turits from KeyBanc. Perhaps -- hi, how are you -- perhaps in some ways an extension of John's question, but on the gross margin side for OCI. And clearly, you've got a differentiated model. I think dedicated regions, cloud of customer, many things. How do you think about the long-term margin potential for that model, which is in some sense is more specific than the more generalized models of some of the larger cloud players?

Okay. So I have to admit that when we were first coming up with these clouded customers and Alloy and things like that, and I thought, how is this going to be as profitable as a giant cloud at scale, right? Well, there are a few things that I didn't include in my own model. And this applies, by the way, to our cloud customer or clouded partner like at Microsoft at Azure. First of all, we don't pay rent. We don't pay for electricity. Those are all covered by the customer. Now for those of you who know, imagine, these are 2 of the largest costs for us. Computers, luckily for us, continue -- should I be holding you guys back? How about we do this? How about we give -- Larry, how about you just come right up here? Are you mic-ed? Or do you want to hold? No, they made us take them away. How about we give Larry a chance, yes? And afterwards, it will be me if you still want me. Okay? All right. I know I can't get out of here either. All right. Larry is going to come back, and I will answer your questions afterwards. There you go.

All right. Real-time. The key question was, is there stairs? Sure, right in front. Gentlemen with the hand up.

Larry, Brad Zelnick, Deutsche Bank. Great to see you. Larry, if I had a crystal ball 10 years ago, and I told people what I see is Larry Ellison getting on a plane to Redmond to make an announcement with the CEO of Microsoft, I think they might have put me in a straitjacket and taken me to a psychiatric hospital.

Well, it was a long shot. I hadn't been there for 45 years. So my -- I'd never -- this is my first time ever to set foot in Redmond. And it was exciting.

The question really -- I mean, Oracle is not over its history, been known to be a partner-friendly company, dating back decades ago, with respect. I think more recently....

Well, I'll tell you 1 thing...

I guess the simple question...

We always make sure that our database run everywhere. And in the world of cloud, I think it's very important that our technology is available everywhere. So it's very important that it's available at Microsoft on Azure and in other clouds. Our intent -- it is -- 1 of the most interesting thing about the migration to cloud, like any new technology, you kind of forget what happened in the past. You're so excited about, oh my God, it's the cloud. And NetSuite was the first cloud company. And near to cloud, they kind of divided it into 2 halves. There are application companies like Salesforce and Workday and NetSuite before we bought them, and they're -- and us and Oracle and then there are infrastructure companies. And they think of us as 2 different businesses, applications and infrastructure even though all applications run on infrastructure, all applications are claiming to use AI, which is infrastructure. All applications are doing analytics, which are on data warehouses, but somehow, it's divided in half. It's actually 1 -- I think it's 1 business. And all of those clouds really came out is walled gardens. It was fascinating. People say, oh yes, no everyone's going to -- I guess some people say when you go to IBM for everything. Absolutely everything. You go to Amazon for absolutely everything. They're going to do everything, unless you go to Azure for absolutely everything because they do everything also. Of course, Google then had to go into the everything business. No, no, no, no, take everything, to -- it was a really bizarre thought. But these clouds were not interconnected. They were all independent and you picked 1 and you did everything there. But then it depends on, what do you want to do a Salesforce automation application. When you do it at Google or -- well, no, no, you can do that at Salesforce, but that's a different cloud. The fact that we forgot all about open systems, we forgot all about customer choice. We forgot all about interoperability and standards and all of that. And you have these new walled gardens that people got very excited about. In fact, some companies do things better than other companies. Some people do AI better than other people. Some people do data warehousing better than others. And customers really want to connect to the cloud that has the technology that they're interested in, and that would be multiple clouds. So they have to connect to multiple clouds. And then it's -- it was incumbent on Oracle when it ran on HP servers that we worked with HP, and we're in on IBM servers, and we work with IBM it ran on Dell servers and we work with Dell. So I understand what you're saying, we're not a "partner-friendly company," unless it's in our interest and it's in our customer's interest. But it is in our interest, it's in Microsoft's interest. I think all of these clouds, our plan is to interconnect not only to infrastructure clouds like Azure but this may come as a surprise to Marc Benioff, who's a very dear friend and has done -- and I think he's done a great job at Salesforce. But our intention is to connect to the Salesforce cloud. So if you want to use Oracle technology alongside of Salesforce technology, we're going to make that as easy as possible because we think that's what customers want. So the whole idea of this multi-cloud movement is just not -- not just to connect to Azure. That was a -- Satya was -- and I agree that the customers do want these open systems. But our intention is to interconnect all the clouds, and I think the customers are going to demand that. We've had customers say very large customers say, hey, what you did with Azure can you do this with this other infrastructure company as well, please. And so we're in discussions, and we expect, again, to interconnect to other infrastructure clouds, we expect to interconnect to application clouds like Salesforce, but others, all of the majors, and that's what customers want. That's what customers were used to. Customers had to really get away from that. Okay, I can't really use multiple vendors anymore, if I go to this cloud. I mean there's things like data egress phase. So if you put your -- I'm not picking on AWS. I think it's actually a great company. But you put your data in AWS, oh, you want to move your data out of AWS, you got to pay AWS to move -- egress phase. We don't have any egress phase with Azure, Microsoft Oracle. We don't do that. It's your data. You can move it wherever you want to. That mean, Amazon's approach is, if you move data into Amazon, that's free because we're an awesome company. But if you move it out, no chance. You pay us for every byte -- every bit of data you move out, you pay us by the bit. This is a very strange idea. And we think that's going to break down in the face of what customers had come to expect for over a long period of time and what they expect now. Gentleman back there, I apologize. And otherwise, I'll call everyone in the first row at this table. And just keep on making a circle there. So.

Go ahead, Brent.

I will get back to you, sir.

Larry, Brent Thill with Jefferies. Thanks for hosting us. You said an aspirational goal of $65 billion, which assumes high single-digit growth, maybe an acceleration from even Q1 and what's underpinning your confidence to hit that goal?

Well, my goals are much different than that. I mean, much different than that. The Oracle's biggest customer until relatively recently, in the last several years -- last few years was it paid us about $120 million a year. They were a big database customer. They bought applications. They're paying annual support fees and the annual support fees got up to the -- and we have a few of them get up over $100 million a year. And that was -- that's a really big customer. It's a big bank or a big phone company or something like that. Now, I would say it would be an odd quarter when we didn't sign a single deal for over $1 billion, and we have multibillion dollar customers. It's -- and they're in -- they're in multiple areas, by the way. They're in infrastructure. They are in a specialized part of infrastructure, which is AI training. I think I said we've even had 1 of the other big 3 infrastructure, I consider us the fourth -- we're the fourth infrastructure company. We're doing pretty well. But then there's Microsoft and Google and Amazon. We're all bigger than we are in infrastructure. Google, not that much. And especially if you look at our total cloud business, they're really not bigger than we are. I mean, you conclude applications and everything else. But anyway, One of these 3 infrastructure companies wanted to sign a $1.5 billion contract for us to do AI training in our cloud because our prices were lower than their costs. That's a good sign, and I'm not completely delusional about our advantages in AI training. Our performance advantages, which in the cloud when you pay by the hour or pay the minute -- by the minute, translated to huge cost advantages. And we say we train twice as fast at less than half the cost. It's actually much better than that, it's actually much, much better than that. So we've had several companies sign up for $1 billion recently. And it's stunning. It's absolutely stunning. And we -- and there's really no end in sight for all of this. So I'm very optimistic about AI and AI training. I'm very -- again, NVIDIA is doing their training at Oracle. And NVIDIA sells their GPUs to everybody. But when they decided to actually pick a cloud to do their own work in, they picked us. When Cohere sells to everybody when they decided to do their own training, they picked us. xAI, which I think is going to be astonishingly successful in this business. Remember, the guy who started to OpenAI also started xAI. Same guy, he's pretty good at this. Again, they chose Oracle to do their training. And I can go on. One of the -- as I said, 1 of the big cloud companies also chose Oracle. But it's more than that. If you look at what's happening in a totally different side of our business, our applications business. Who -- there are really 2 application cloud companies of any scale, Salesforce and us. And what we do is very different than what Salesforce does. We do ERP manufacturing. We're now doing robotic manufacturing all sorts of things like they don't do that. They do sales automation, they do marketing and service. So they're quite different than us. on the application side. They also do infrastructure, by the way. They do analytics. They bought Tableau. They do a lot of things that are not applications. They also mix infrastructure and applications. But our market share, who -- I'll ask you, who do you guys think compete with us in ERP in accounting system, procurement systems, inventory systems, manufacturing systems, supply chain systems in the cloud. By the way, Salesforce doesn't do it. They're not trying. And SAP never moved to the cloud. I know they say, they have this thing called S/4HANA in the cloud, but it's very easy to check if that's true. Go on the Internet spend 15 minutes and try to find SAP's cloud. Try to find it. You can find Salesforce's cloud. You can find our cloud, you can find Workday's Cloud. You can find all these other cloud, ServiceNow's Cloud. And they tell you when the next release is coming out and at the time, how much downtime you'll have as the next release is delivered in, they'll provide all of that. All the cloud companies do that. They're just not there at all. Doesn't exist. What they do is hosting, they take the old SAP stuff, and they'll put it on a computer some place for you. But it's certainly not -- they never rewrote it. We rewrote everything. They never -- so my point is not to say that SAP didn't go to the cloud. My point is, say, we don't have a competitor in ERP in the cloud. How big is that market? And ERP in the cloud is very different than ERP back in the SAP days when SAP was #1 in on-premise. What is ERP will automate B2B commerce. So we're working with JPMorgan Chase. I think we announced MasterCard, do we announce MasterCard as another partner to automate e-commerce between businesses and what is e-commerce between businesses look like in the area of cloud. It's an Oracle procurement system in 1 company talking to an Oracle order management system in another company and transacting, buying something, Cedars-Sinai hospital buys a scanner -- an x-ray scanner from Siemens and then you have to finance it. So you go to your favorite bank, it could be Deutsche Bank, it could be JPMorgan Chase, who knows, whoever their bank is, you finance it, then you have to ship it, it could be DHL, it could be FedEx, and that could be shipped. But -- and then you want to know when it's going to arrive. You have to place the order -- Siemens has to actually place the order if it's manufactured to order, if it's in inventory, they just call it billable to promise, so how soon can they get it. They have to build it, they have to then schedule in their manufacturing system. We automate that. B2B commerce is not automated. B2C commerce is brilliantly automated and has been for some time. I mean, Amazon and Walmart, but a lot of people have done a great job of this. And B2B commerce, it hasn't been done, it's much more complex. But we can do it. We can automate all of those transactions. Do the loan origination for the financing, do the insurance, track the packages, the delivery let you know the inventory is that you can have cold reverse auctions, if you want you to do all of that with your systems. Modern manufacturing, it used to be not -- well, it used to be -- none of the ERP systems really gone on to the factory floor. They didn't deal with the [indiscernible]. They did what was called material requirements planning, and they made sure that you had all the parts and all the pieces you ordered them and they were in inventory and available and they would keep track of work in progress, but they didn't run the factory floor. They weren't out on the factory floor. And now in modern robotic manufacturing where everything is digitized, including the factory floor, we are doing that. We're starting to -- and -- but we're integrating the robot factory floor with the MRP system that came before it, with the B2B commerce system that handles the distribution and then the service for that product. We do all of that. So this next-generation ERP system or really the next-generation ERP ecosystem. And I'll use that word a lot today because historically, our industry -- the IT industry hasn't automated ecosystems. They've automated this area in that area. They've never really tried to take on entire ecosystem. For example, in health care, we bought Cerner. Cerner automates hospitals, providers, large providers. So does the Epic and say, okay, Cerner competed with Epic, they automate hospitals. Well, yes, we're doing that, but we're also automating pharma companies. How about the payers, the government agencies who regulate health care. If you're going to have a clinical trial system, you've got to make sure that the regulator who approves the trial or doesn't approve the trial can take your digital information and process it. So you don't have to build systems for the regulators. You have to build the systems for the pharmas. The pharmas use hospitals to run the clinical trial. You've got to make sure the hospital systems can handle the clinical trials. And then there's the -- what about in case of a public health crisis that you have to build systems for the government. So they can manage the bulk of their budget. The NHS's spend's about 50% of the U.K. budget. So what we're trying to do is automate the health care ecosystem, not do just what Cerner did historically, which is to automate hospitals and clinics. But automate the NHS, big chunks of the NHS, collect -- have these databases that collect national population data on all -- so all the health records of all the people in the country that are being treated or are somehow engage with the NHS. All those electronic health direct records are in 1 place. So suddenly, we know the outcomes for entire populations we've never known this before. Recall COVID. Remember we sending the ship to New York City because we were running out of beds in New York, that beautiful white hospital ship that cruise into New York Arbor to make sure that as soon as we ran out of beds, we had, we never ran out of beds in New York. They had no idea how many hospital rooms are occupied. They had no any idea how many people contracted COVID in the last week. They had no idea of any -- there was no place to collect. We were building systems for them during COVID to help them collect that. They had none of that information. The information they had, Mike, was how old 90 days old? It was 6 months, did you say? Okay. 90 -- I was too optimistic. Yes. I mean, it was -- I mean, they were completely flying blind. So it's not what we're trying to do with Cerner, so why do I think health care, okay. So I mentioned AI training, billion of our customers and the amount is going to be invested in the next 5 years is just going to be stunning, both at the infrastructure level and at the application level. I mentioned B2B commerce and modern cloud ERP systems, where we're automating the B2B commerce ecosystems, including logistics. We're working with FedEx is another 1 of our partners, Logistics, DHL, logistics, finance, insurance, robotic manufacturing, all that little business. And then there's this health business, which is the largest business on Earth. But it's the business, the entire ecosystem, and we're working with it's really funny because normally, we say we signed this large contract with this big company. Now we're saying, oh, we just signed this large contract with this country. We've gone to sell from dealing with companies to dealing with countries. There are countries that want a gene that their national health service want to gene sequence the entire population. Because if you want to personalized medicine, and you have all of this outcome data knowing -- I use my favorite example because it's probably the most common drug prescribed in our country, which is a statin to lower cholesterol. And there are several different statins, there [indiscernible] cholesterol or they are a lot of them. Well, which 1 -- which is the best 1 for you or which is the best 1 with the person with your genotype? What is the right statin to give you the best possible outcome and keep you out of the hospital for as long as possible. So you have a longer life, a healthier life it costs the payers less cost you less. No one really likes to go to a hospital. What's the answer to that question. We don't know. We don't put that outcome data. It's in no place. We don't have population-based outcome data like this. So there are some countries that are very interested in collecting all of this information. By the way, there'll be almost every country, every national health service I talked to ours. We have the largest collection of health records in the United States, which is the combination of the VA and the DoD. And we will, I think, this quarter, received 2 awards from governments for health systems for about $1 billion. So why am I pretty confident we can meet -- asking me, I think I never experienced anything like this before. Never see an then like this before. I remember I was talking to Safra and we said -- I said I've seen periods of time when Oracle has been tremendously successful. I think when we won the database wars back in previous generation of database wars the last 2 companies standing that were -- we were competing with were IBM and Microsoft. And I was there for that, and we just won everything. We won -- we just absolutely dominated a huge market share. And now I see this reoccurring in other areas like AI training, health care where -- but the scale of the business, the scale of the then on-premise database business versus the B2B e-commerce business, the AI training business and all the result in AI applications health care from dealing with countries to dealing with groups of hospitals, from dealing with the pharma companies, the regulators and dealing with citizens who are very interested in wellness programs. So everyone suddenly everyone is enrolled in these systems, everyone on earth pretty much. certainly, everyone in the West. And so the opportunities are so gigantic. And I'll ask you another question. Who else is working on these problems in that systemic way? Who else? And I think we're uniquely adapted to do this because we are the only cloud company of scale that does infrastructure at scale and applications at scale. And you can't tackle the health care problem, I would argue. One, you can't -- without having -- solving very serious infrastructure problems. You have to be able to manage population scale databases. You have to have population skilled databases that are available 24 hours a day, 7 days a week. You have to store all -- you have to digitize in store and all those medical image data and use it for AI training, AI training, everyone is very excited about what ChatGPT did. Okay. That was certainly a milestone and everyone's amazed. But their job was much easier than because they -- was made easier because they just had to vectorize language. They had if you will, capture the semantics of language which is much easier, say, than Elon Musk's problem, which is capturing all the images you have to look at to make decisions while driving and generate the controls of the car. If you look at just the volume of training data of image data, how much has to go into the computer to train the computer. It's 2, 3 orders of magnitude more than the language. So if you want to get smart about images and there are lots of applications for smart about images, the most famous is self-driving cars. You have to have computers that are really good at moving huge amounts of data. And what we're very good at is moving huge amounts of data. We're better than that anybody because we've been doing it for a long time because we've built these -- we started building these large databases long before anybody else. And when then we build custom networks to handle these very, very large databases. These already [indiscernible] networks. And it turns out there are other problems that have to move large amounts of data. We just do it better than anyone else. And if you think AI training for language is a test of that or demanding on moving large amounts of data, way to look at medical images and driving data and all of those things. So we have a highly differentiated technology. We have a skill set that spans applications and infrastructure. And we're going after markets that are so large, it's certainly beyond anything I've ever experienced before. And the transactions that we're getting recently, I mean I'll call Safra and said, oh my God, it's like I'm like, [indiscernible] oh my God. I mean we don't get $1 billion transaction. Sorry, I mean, I wish we did. But we never got $1 billion transaction. We never got $8 billion, $1 billion transaction. We didn't have $1 billion customer. And now it's every quarter. And we have customers that are on their way to be multibillion-dollar customers and prospects, again, their countries that are standardizing on our health care suite, countries that are looking at our Oracle Cloud. One more thing I'll say and it's the longest answer, I promise I'll never take this long on another question because I can't. It will be nightfall and you'll be hungry, you want to go to dinner. The other thing that's fundamentally different between us and all the other cloud providers, is all the other cloud providers build these huge data centers, and they're all a little bit different. They built one, and they built another 1 with a few different changes, another 1 over here. And they're enormous, and they added on to them and they made changes. And I remember we were building -- we were back in our Gen 1 days and I was involved in the project. I remember when I canceled the project. There are a lot of really unhappy people. This is crazy, you can't do this. And I said, I don't think what we're going to do is going to work. We can't -- it looks too much like a regular data -- what these guys are doing, and I understand why they're doing it. But it looks too much like a data center, a conventional data center that doesn't understand it. I think there are a bunch of things we have to do, they're different. And I have to think it has to be totally automated as we completely lights out. the pieces in it have to be autonomous. That has to be self-driving, okay? It has -- the operating system, we need an operating system -- we have to build an OS that's totally self-driving, Autonomous Linux. We've got to use -- it's got to run on our autonomous database, our cloud, to run our cloud, you need data. And that data has to be available all the 7 days a week, 24 hours a day, can never go down, and you can't have people doing it. And we can talk about security and all these other things, but you just have to take a totally different approach and you have to completely automate the thing. And you can't have like 30 of them. You got to have 1 every major city in the world. They got to be close to population centers. And because no one's figured out how to speed up the light. Maybe there's some people working on it, but it's a hard problem. So we need to have these close to people that are consuming the data because the cost of building the network to a small number of these gigantic -- the cost of building these data centers, they get larger and larger and larger. The cost of building data centers that are different from 1 another, which means you really can't automate them. Now we've got to build them like their airliners. They're still not much cheap, but a few million dollars, but they have to all be the same, except for scale. We have to be able to put them eventually into every city, every country, every big city, every country. They all going to be -- they all going to be meshed together. They've got to connect the satellites, they got to connect to terrestrial networks. Back to automation, automation, automation. Then we got to be a [indiscernible] up. Some customers are going to want them in their own data center. They're not want to be -- they don't want to commingle their data with other people's data, no matter how secure we get, they won't want to do that. Let me mention 1 customer like that, the Central Intelligence Agency. The Massad. I can go on. But there are other big banks, certain big banks, NRI, NRI in Japan, which runs the Tokyo Stock Exchange. They really don't want people in their data center. If they're running the Tokyo Stock Exchange -- just stand up -- you can't get on to our network. So we had to redesign this. So our model was lots of smaller data centers highly automated, that were really the same, highly automated and all linked together versus a smaller number of much larger data centers. So we had a completely different architecture. And we started over, we call it our Gen 2. Every -- We got to use this -- the already main network every place, automate the -- autonomous database is got to run all the systems in that actually run the cloud. And that's all autonomous we can't use the regulated data -- again to do that. Everything has to be automated. That's what we did. We thought it would give us incredible cost advantages, security advantages, performance advantages, that can't be overcome because light only goes so fast. Anyway, it's another big differentiator. No one else is doing this. We paid a heavy price for it. I mean we got -- our infrastructure. We had the first -- I started the first cloud company with Evan Goldberg called NetSuite. It was the first cloud company. It's not like I didn't think that having a utility model for computing was a bad idea. I did hate the word cloud. I thought it was a very strange world. We -- now used to make fun of the word cloud. But we started utility computing with NetSuite. I'm the believer in utility computing, long believer of utility computing. But if you're going to build these utilities, you have to -- they look very different than a conventional data center. I think they look very different than what most of the infrastructure people are doing right now. So we have enough differentiation, technical differentiation. We do infrastructure and applications. We do lots of small -- lots of cloud data centers close to consumers rather than a few client rather than a few hyper, hyper large data centers. Our stuff is all fully automated. We're trying to automate ecosystems, not 1 part of -- 1 stove pipe in an ecosystem. And it turns out there's a lot of evidence we've gotten recently that what we're doing is working because people are buying it, like I've never seen before. Okay. Let's see, a gentleman back here. I really to go to the periphery. How about this gentleman there and then I'll call, and you next, sir.

Raise your hand, please. What was it?

Guy over here. Wait, I think, about 5 rows back, 6 rows back or maybe they don't want to be...

Was somebody asking a question back here?

Was there an arm up or? Sure. Anyone who's willing to ask a question.

Larry, this is Siti Panigrahi from Mizuho. I want to ask you about database question. No doubt, Oracle runs all mission-critical workload, but we keep hearing new workload, new app. That's where you see Oracle lagging, I would say. So '23 see now you're talking about JSON, vector database and then also now Database-as-a-Service, should you expect now Oracle now regain their database, your leadership -- database leadership now with new app and workload?

Yes. Well, i think Oracle wasn't really available in Azure. Oracle wasn't available in Google, Oracle wasn't available on AWS. And I think that hurt our database market share. So to reclaim our franchise, but I think our database is way better than whatever is second. I don't know what second would be. And I think we can simply reclaim our database market share like we got it in the first place, which was having by far the best database in the world and making it available everywhere. And if our competition is Snowflake, by the way, again, I think the guy has done a good job, but I think we're better than they are by a lot. And -- but they made themselves available on Google and Microsoft and Azure. It's a good strategy. And it had been our strategy, we were famous for being portable. We were the first database company that did that, that put itself in all platforms. So I think a combination of interconnecting to all these clouds. So we can interconnect to some clouds, so we create the illusion our database is running in that cloud. Or the database, we can do what we're doing with Microsoft, which is literally building an OCI data center, the right on the floor and next to the Azure data center. We can do both of those. But it's still going to be an OCI data center with the Oracle database available every place. And I think that will allow us to continue to manage most of the world's valuable data. It was a much shorter answer.

[indiscernible] Global Equities Research. Super exciting announcements, a lot of good stuff I was just wondering, we have this vector as a -- I would call, a similarity search which seems to be an add-on or integrated into your database, I was thinking, would it make sense for you to have vector database as a stand-alone SaaS offering or have a single tenant that can be installed on say, on the bare-metal? And the reason I'm saying that is, this is a very good strategy you have because you are converting SQL Server -- SQL programmers into an AI programmer, but we are leaving the whole 4 million, 5 million PyTorch developers out. So if we have a dedicated stand-alone vector database, you expand your market, you're leaving a lot of money on the table or I should say, in the cloud for other people to eat.

Okay. So this is the idea. Should every data type have its own database. So Mongo -- you've got JSON documents, you should have build Mongo -- use Mongo database for JSON documents. You use Pinecone for vector database, and you have -- the problem is what kind of security does Mongo have? What kind of reliability and robustness does Mongo have? And what if you want to combine, what if you have a query that combines documents that are in JSON documents and semantics that are in vectors. When you start segregating data, you put some of your data here and some of your data there and some of your data there and some of your data there, and you want to ask a question that spans all of that data, like, let's say, a healthcare question. You want -- we'd like to be able to store vectorized images in our -- in electronic health records. We'd like to be able to store genomics in electronic health records. There -- we'd like to be able to store claims data in electronic health records, outcome data in electronic health records, treatment history and doctors' notes in electronic health records. Should we really have 25 different databases, all with different security systems, all with different schedules to be updated? We think that's a really terrible idea, forgive me. We think that should be stored, the databases should be unified so you can ask whatever question you want about whatever data you have and get an answer. -- which you can't do. And then think about the duplication of effort. If we build a separate vector database, we have to -- it has to be -- you have to be able to recover it, but it goes down. You have to back it up, you have to update it. You have to keep track of the users who are authorized to see it, see that data. You have to duplicate 90% of the database functionality that you've already built rather than simply adding yet another data type to the Oracle database and another index type for vectors. So an awful lot of these queries when you're searching for movies, by the way, and these recommendation lists. And actually, it's really a combination of non-vectorized data and vectorized data as part of that query. That recommendation engine uses not just vectorized data, but conventional data as well. And you can't ask the question or you have to ask it across multiple databases. It's much more expensive to do with the way you're -- much harder to develop. They are not very good systems. And I'll just say, we have a compatible interface. We have a JSON-compatible interface with Mongo, run any benchmark and see who runs faster, run any security audit and see who has better security. Any reliability test and see who's better -- scalability, who's better at scalability -- we're much better than they are at that. Now what they did, what Mongo did was they made it very simple for programmers to get and put JSON documents or JSON objects, whether you want to call them, they have a nice document UI. So we gave you the document UI. We generated the underlying relational schema. So with the Mongo database, you can't use SQL to ask queries of the data in your Mongo database. You can't -- that's not part of the Mongo database. So it was much better to add JSON documents and simple get/put logic for programmers, to make it easy for programmers to use the Oracle database as a JSON database, if you want to use it as a JSON database. But it didn't make sense to make that a separate database. That was just yet another type of data that we store in the Oracle Database, which has all the advantages of Oracle security, reliability, scalability. And the fact that all your data is in one place and you can find it and ask whatever question you want. Over here.

This is Keith Weiss from Morgan Stanley. Your excitement about generative AI is clearly evident in the conversation. I was hoping to get your opinion on the debate that we're hearing in the industry about the monetization potential of the large language models themselves.

I missed that last couple of words.

So there's been an industry debate about whether the large language models themselves will -- could be monetized effectively or whether they become commodity. What's your view on that? Is it going to be the applications built upon them? Or do the large language models, is that going to be a commodity or not?

No, I think certain large language models are better than others. I think it's a little bit like -- will database be a commodity? Well I think these things will be differentiated. And I think there will be -- I'm very familiar with several of the projects that are ongoing. And -- some people will do better jobs on -- it's interesting large language models are now dealing with images and also with other things. So we still call them large language models, but they deal with a lot more than language. The -- no, I think you're going to see a lot of progress and different development teams will do better than other development teams and that will make a significant difference in how successful will you be monetizing your technology. But I think you'll also get specialization. So if you're talking about a large language model, which is a foundational model, which is trained on the Internet, they have the same training data. Let's assume that you told the training data constant and it's just a neural network architecture that's getting better or some people have better architectures than others. Let's just say that. But it's the same training data. I think that will be one area of differentiation that the neural network will just be more effective. The other big area where you'll see it is if you have the training data, if you have a huge amount of training data for healthcare, whether it's diagnostic, do you have cancer or don't you have cancer based on this biopsy slide? Whether it's a diagnosis or it's a treatment plan after you've been diagnosed? The more data you have, the better data you have, the more current data you have, the more quickly you can process this, will give you significant competitive advantage. And what we're doing at Oracle in healthcare, it's really interesting. The database, the Oracle Database, the EH electronic health records, getting this population scale data, all of it, not just the vectorized data, but all of the data, yes, you want the vectorized image -- biopsy image. But you also want all of the treatment data, the doctor's notes, the claims data, because you want to do personalized care, you want to do value-based care. You want to know what -- give people the best outcome, but also -- and the best outcomes are almost invariably the lowest cost outcome. If I keep you out of the hospital, you're happy, your payer is happy. So the quality and quantity of your training data is very important. And I think here, Oracle is in an excellent position in certain industries, for example, in healthcare. We're in a very good position because of our acquisition with Cerner, but also because of our clinical trial work with pharma companies and our database work, the fact that our autonomous database can handle population scale data. Most databases can't do this. Try to put population scale into Pinecone. Good luck. So the fact that we can handle this vast amount of data and use that data to inform policymakers in national health services but also use it to train -- improve the training and specialize the training of models, is where we're going to be putting a lot of our effort. We'll do some work with people who are working with foundational models, but we'll be taking those foundational models that were trained on the Internet and understand English and other languages and understand images. And we'll be adding a huge -- a lot of the specialization is a very small amount of data. You add your service or a company adds its service records or an investment bank adds its trading data, so I mean like that. We're going to be adding like the second Internet, the healthcare Internet to all this EHR data is doing the training. So I think there -- it's really this combination of a high-quality neural network that's been trained on something more than just -- I'd say just, the multi-trillion bite Internet. But yes, you'll train it on -- Wikipedia and all the rest of the Internet, but you also train it on the EHRs of 50 large countries in the world. And once you've done that, you can do a much better job of helping doctors treat patients, helping drug designers design molecules that bind to viral proteins and keep the virus from replicating, doing those kinds of things. Having access to that data and being able to feed that data effectively into these NVIDIA super-clusters where you can train these models with this specialized data. I think that's just a tremendous asset and advantage we have that came from our database business, our network and cloud business and our healthcare business, all kind of blended as one.

Alex Zukin with Wolfe Research. I wanted to ask a question about chips, specifically the availability of certain GPUs, both from the aspect of how is that impacting your ability to actually deliver to the demand that you're getting doubling the AI backlog every quarter. And also, the second part of the question, it seems like...

If we had all of our H100s delivered, everything that we ordered, that we had delivered, we would have had an NVIDIA like quarter. Of course, we have to build the data center. We have to plug in the GPUs. We have to build -- or building data centers, the NVIDIA GPUs use a lot of power. We're building -- I very proudly said, we're building the world's fastest computer. I left out the part saying it consumes the most energy of any computer ever built and that we have -- we really built us new data centers to actually hold these things with a power envelope like that. And then we charge by the month. We don't get all the money. NVIDIA had a hell of a quarter. The -- our business model is different, right? We have to build the data centers. We have to build the GPU super-clusters. And then we get -- and even if we signed multi-billion -- several billion-dollar contracts, and they're sitting there and people start training, they still pay us what they use that month. And then they pay us for what they use the next month. So yes, we got to build -- there is a lag between the time we get the PO and the time we can start providing these GPUs. I had a very interesting dinner at Nobu in Palo Alto with Jensen Huang and Elon Musk, where we -- Elon and I were begging. I guess is the best way to describe it, an hour of sushi and begging. But I think it worked out well. I think Jensen recognizes that what actually AI is doing is very important. And insofar as it's possible to prioritize delivery. I'm not suggesting that we're taking it from anybody else, but that we're -- I think we're doing pretty well. The fact that NVIDIA is training in our data centers is encouraging. And we think we're going to get H100s as fast as anybody else. We placed a lot of orders. We've got a lot of delivery. We have a lot of GPUs running already, but not nearly honestly just not nearly enough. And there is a lag from the time you realize. I mean, really this demand just kind of showed up relatively recently. It's not like -- and this is the tenth consecutive quarter NVIDIA has had this crazy blowout. So yes, we have to build the data centers. We have to get to H100s. We think we're in good shape in doing that. We think we're ahead of everybody because a lot of our --. We can build a lot of -- super-clusters in our existing data centers because we have RDMA network every place. But if you're talking about a 16,000 GPUs super-cluster, you're talking about that monster that we're building, then we literally needs -- and the new data center is ready, we have so many data centers under construction. We'll soon be moving from 64 to 100. We have -- a lot of them are being built. But it's some of the biggest new ones are months away -- not years away, months away. But still months away, it doesn't mean it's not like it's -- they're all up and running now. So there is a lag for some of the stuff. And there's also a lag when we get a contract from -- in healthcare. That's, if you will, the interesting aspect that the cloud business -- in the old days when we just sold software, and by the way, that's what's partially what's going on at Cerner. Cerner used to be in the software business. Now Cerner is in the cloud business. We're in the process of moving every Cerner customer to the cloud. We'll have the majority of Cerner customers in our cloud by the end of this calendar year. And you'll say, oh my God, how much is that going to cost you? I mean they're paying by the month by -- writing checks. We've been doing -- we've been going through this transition of all of Oracle's businesses as they move to the cloud, which means we get -- we actually, in the end, we make a lot more money. We deliver a much better service and a much better product by moving it to the cloud, but it's a different business model. We don't get the money on day 1. We get the money over time. And -- but in the end, we've signed a lot more contracts and we have much happier customers and all of that. But yes, it's just the cloud business model, you take a company like Cerner from selling software to cloud, there -- it does slightly change that. So in spite of that, we're doing very well. We're getting enough new business to cover, virtually make that transparent. Right in the middle.

[ Marshall Bush]. Since you participate in almost every layer of the tech stack, what layer do you expect to capture the most value over time? And what Generative AI application do you think will be the first killer app?

That is such interesting question. So let me just kind of rephrase it. If you just say cloud, you don't even count applications, right? I mean, as you say cloud, you think AWS and Microsoft, those two giants. And so even though the applications pioneered the cloud. Salesforce was doing this year after NetSuite was doing it. I'm very familiar with Salesforce. Marc Benioff copied what Evan Goldberg was doing, and I was an investor in Salesforce as they move to the cloud. They were very interested in what Evan and I had done with NetSuite. And they said, okay, we're going to do that. We're going to create a Siebel in the cloud. So the cloud all started with applications. But when people just talk about the cloud without any further precision, they think infrastructure. So I'll answer -- which part do I think is ultimately more valuable, the infrastructure or the applications sitting on top of the infrastructure? And my opinion in terms of monetizing this, ultimately customers want specific applications for their industry, industry-specific applications. And I look at all of our technology is enabling technology for us to solve problem, application problems. And Generative AI is much more interesting when I say I'm using it to design small molecules as antiviral drugs versus, I got ChatGPT and it can talk. And it can answer questions about Wikipedia. Well, that's really cool. And it can write poetry. It can write poetry in the style of Chaucer or Shakespeare, it's really cool. It can write music. Hollywood's worried that it can write scripts. Okay, scripts, they're cool. I like movies. I got a couple of kids in that business. Hollywood writers are little panic now. They're on strike because they're afraid for their jobs with this. But -- let's see what business do I want to be in? Do I want to be in kind of chatting with a computer and asking them, hey, do you like my shirt? Do you think I should wear this, which I'm wearing jeans, let me share the whole thing. Or do I want to design a molecule that stops COVID-19 for replicating and also knocks off SARS and MERS in the same time. That's pretty good. And then what about curing cancer versus writing a really good piece of poetry in the style of Chaucer? I didn't like Chaucer. I like Shakespeare. I don't like Chaucer. I think I'd rather have the pharma application and the drug design application. I think it's more valuable. But the other technology, you can't do it without the other technology. So in a way, it's just like you're saying, which is more important, where they're both essential. You have to have both, but which business would I rather be in? Where do I think I'm going to get the most reward, both emotional -- okay, I'm a scientist? Do I want to work on the basic neural network? It's pretty cool. Go to work in DeepMind in London and work on that? Or do I want to work on the pharma company designing molecules and specializing the neural network -- with that information and trying to build those drugs. I want to build those drugs. So, I want to build cars that don't require fossil fuels, don't require service, that are inexpensive that everyone can afford, don't require government subsidies. People want to buy, they are fun to drive, all of those things. I'd rather solve the Tesla problem. The Tesla problem. Now he had to solve a lot of fundamental technical problems. And by the way, to pull this off. I mean he had to build a robotic factory. He had to build -- I don't know if you know this, the new Tesla factory in Texas is the biggest building ever built in the history of human kind, built in 18-months. And he had to be his own contractor because no contractor could build a building that big in 18-months. He had to build -- do you wanted to build these cars, you got to build buildings, you've got to build robots to build the buildings. You have to redo battery chemistry. You have to solve an incredibly difficult AI problem for self-driving. How many lives you're going to save with self-driving worldwide? But you want a successful car company. If you have cars that don't self-drive in a few years, I mean, you have no chance of competing. What do you mean you got a car -- now, why do I do I have to drive this car myself? But being in the car business, by making 20 million cars -- 20 million electric cars a year, you are solving a good chunk of the climate change problem. And I think he'll be reasonably prosperous as an individual for having done that. I think he's just going to be fine. So Yes, the underlying technology is -- are the key enablers, whether it's battery technology or lightweight steel alloys and all the stuff that Tesla -- lots of these new technologies are key enablers. But it's the application, building the cars, building the pharmaceuticals, delivering the future to humanity, building these robot greenhouses, which we are working on which I think. Providing food supply, a reliable food supply in drought-stricken areas like Ethiopia and Somalia. I mean that's a lot of robotics and sensors and AI and plant genetics and all of this -- they're a lot -- and then material science, I mean these are not greenhouses made out of glass. They're made out of incredible shrink wrap plastic, I won't go into the detail. It's a robot factory that builds them, all of that. But I want to solve the food supply problem. I want to solve the -- another viral drug problem. I want to solve the energy problem using these 21st century tools. And I think the reward, but whether it's emotional reward or financial reward, it's going to be in the actual solution, the application of the technology as opposed the development of the technology. Over here.

Okay. Maybe last question, unless Larry, you want to continue, it's completely up to you.

I'll go on for a while. -- until my answers get really incoherent.

Brent Bracelin, Piper Sandler. Larry, I think it's been 5 years now. You've been talking about the merits of OCI. I was a skeptic, I'm a believer now.

Thank you, seriously. I don't blame people being skeptical because we were late. There was a reason why we were late, and I told you -- but I understand. I understand. But by the way, I too require evidence that we're doing well. I say it, and I say, do I really believe that. What evidence is going to -- I use to convince myself. And I'll -- and now we have -- I actually talk about this with Safra. When another hyperscaler comes to us and wants to do AI training in our data centers and the -- for $1.5 billion, and they're a significant competitor and the contract is ought to be a secret, of course. And that's a sign. We're doing okay, that our stuff is probably pretty good.

Based on the people here, it looks like there's more than me that's just a believer. But I wanted to bring up the database part of the business. Full disclosure, I'm a skeptic there. As we think about the database market, what's changed over the last 10 years. My question for you is, is 23c and portability to these multi-clouds, your OCI Gen2 moment to reclaim and If so, is it take a couple of years? Or do you think the adoption cycle to reclaim your place, is that #1 database player? Could it happen faster?

Okay. So it's -- 23c is very important. That's the merger of object and relational, that's really important. Multi-cloud is really is -- I think, really important. But the big deal is the Autonomous Database. The Autonomous Database is a tiny fraction of our installed base. It's very interesting. It only runs in the cloud. It does not run on-premise. You can't upgrade. You can't -- our huge installed base of Oracle cannot move to autonomous without also moving to a cloud, and it turns out only OCI. And we -- Fusion doesn't run on autonomous database. NetSuite doesn't run on a Autonomous Database. A lot of our own stuff we didn't move to Autonomous Database for a bunch of reasons. Now Autonomous Database is so good, and it's going to be -- we're moving everything. Fusion is going to autonomous database. NetSuite's going -- everything. All of our cloud technology is moving to autonomous database. Our cloud itself actually runs on the autonomous database. To feed these huge AI models, to train these huge AI models, you need a database that can transfer enormous amounts of data, vector -- you can vectorize and transmit enormous amounts of training data in parallel. And the autonomous database is a really good at that, it's really good at that. And we automated all of our data centers with the autonomous database because there are no DBAs. No one has anything remotely like this. Almost all security problems and the worst security problem is losing your data. Almost all security problems are caused by human error, driver crashes, plane crashes, data center crashes, loss of data. It's always -- there was this famous one where this bank running at AWS, lost all their credit card data, that's in your wallet. So they lost everything. And I called the CEO, he didn't call me back. The -- but Amazon published this really interesting thing. They said it wasn't our fault. They made a -- it was the guys at the bank that left that basically made these mistakes when they configured the system. And they lost their data, but it wasn't our fault. They misconfigured the system. We don't run everything in Amazon. Our customers run their stuff. They rent from us and they run their stuff and they configured it wrong, and they lost all their data, wasn't us, not on us. It's on them. Well, guess what? It can't happen with the Autonomous Database, customers can't -- don't run it. We don't run it. It runs itself. It patches itself. It tunes itself. It secures itself. You don't do anything. You can't come near it. There's no steering wheel, there's no break. There's no clutch. There's no controls. It runs itself. No human labor, no human error. No human labor saves a lot of money. If you don't want to lose your data, you got to be willing to spend a lot less. There's nothing close to this out there. So I think if it's available at replacement, it's going to be available every place. And it's that much better than everything. We really haven't lost our franchise. We've lost a lot of new apps like Netflix doesn't run on Oracle and that. And there are other companies like Netflix. I get that. Some people like Snowflake, it's really interesting. I mean, again, I respect what they've done. But I know the guys who did it, they was a team out of Oracle. They can't do any of the stuff that we can do. They didn't even do transactions. Look, they did some interesting things. Do I think we can be as dominant as we were not that long ago in database, we still store almost all the world's valuable -- most valuable data. But I think we can be more dominant because of autonomous database. It's just that it has to be available -- it has to be easy for everybody to get. But once you use it, ask Clay Magouyrk, who runs our cloud. I didn't order him to put autonomous database into the cloud, he tried it and said anything else is insane. This is -- I don't have DBAs. It never makes a mistake. It runs all the time. It never goes down. And it's much cheaper to run because it's truly elastic. Who says the databases aren't elastic. I know Amazon calls itself elastic cloud. And some of the Amazon's elastic. DynamoDB is elastic. But DynomDB is not Oracle. And the Oracle Autonomous Database, when it's not running -- if it stops running, it occupies -- it uses exactly 0 servers, it's serverless. It gets as many servers as it needs, while it's using them and it returns them to the pool when it's not using them, nothing does, nothing does. And it does it more or less instantaneously, in milliseconds. So there's nothing like that, that doesn't require DBAs. You don't have to reserve computer shapes to run it. You don't pay for what you don't use. Again, dramatically more secure because there's no human labor, no potential for human mischief. So I think we get that franchise bigger than -- and as much -- in the franchise, these days is much bigger than -- I mean our database business is growing. You can say we're losing share, okay, because the overall market is growing faster than we're growing. But our database business is still growing with Oracle available everywhere and autonomous database being incomparable, and us solving the vector problem and us being able to move vast amounts of data, solve the training data, the data problem not only for language but for images. Do I think that we're going to be successful or as successful as we were in the database business? I think we're going to be more successful, because the technical demands are greater. And the more demanding the application is the better we do and the better we distinguish ourselves. Right here. I will not be offended if a few of you leave. If all of you leave, I'm going to leave too.

Stefan Slowinski from BNP Paribas. Just want to ask about your use of Gen AI in-house. I believe in your keynote, you mentioned up to 30% productivity benefits in software development. So just wondering how we're using...

I apologize. Let me just say, I think NRI said they lowered their development costs by 65%. Once again, I'm -- we think we can get a 10x improvement in productivity. I know that sounds bizarre. What it will mean is not that we will lay off 90% of our programmers, we're not going to lay off any of our programmers. We're going to do more. We're going to do a lot more. And our applications will be much broader in scope and we can take on -- like we're taking on the healthcare ecosystem, we'll look at doing similar things in industries that aren't quite as complex as healthcare. We picked [indiscernible] and now we took -- pick them on purpose because they are so complicated, but utilities, banking, we're going to go after other industrial ecosystems. Now that we have a tool set that can actually automate not just individual companies but ecosystems, networks of companies and government agencies. I'm sorry, I didn't let you finish your question.

No, just wondering if that's what you're going to do, you're going to accelerate development or to generate more cost savings, and anywhere else are you using Gen-AI internally to drive efficiency and productivity benefits?

I think for -- I think the most conspicuous for us is all the new markets, that let's just go into. We never would have gone into built a first responder system. This is just a net new idea. I was personally involved in building the first responder system, and it was inspired -- Mike and I worked to discuss this. We looked at some incidents that occurred in policing in our country where there was this -- I don't want to go into detail, this incident in Memphis when someone -- the police -- by the way, I don't think it happens all that often. But when it happens, it's horrific, right? It's terrible, because we trust the police, we give them a tremendous amount of power and occasionally, they're human beings and they make mistakes, right? So the fact that young police officers are out in the field without supervision is troubling, things can go wrong. So our approach to solving that problem, and then they have these body cameras that record, you can turn them on and off and all -- so we just -- we came up with this idea, what if we had body-worn computers, 5G computers, and we had just cameras and microphones and if an officer is in a difficult situation, and it's always on. But it's not recording on the police officers body, it's transmitting 5G to a command center, and it's put up on a screen. It's also being recorded. But -- so the police officer can get assistance in difficult situations, get advice in difficult situations, gets supervision in difficult situations. The police officer is never alone. Same thing with a fire department, they can get situational awareness. They can see if there's the window shift, there is something like that. But we now have these technologies, we can build these real-time audio/video networks where we can take the images, what's going on and what the police officer is seeing, what other police officers are seeing, put it up in the command center and a senior person, more experienced person can weigh in on what either stand down and wait for help or whatever the advice is there's -- you're never alone. You're always part of a team, and there's always supervisors unseen. And that by way is also true for medical technicians and ambulances. You are moving a patient to the hospital, patient is bleeding and what should I do? I mean, it would be you have to clamp off an artery that's been severed. It could be -- you need to use tranexamic acid, which is the best chemical we have right now to get blood declot, to stop bleeding. There are a variety of things you can do. But the EMT in the back of the ambulance is in constant communication with the control center at the ER room in the hospital and there are medical professionals helping the EMT care for the patient while the patient is in transit. So we now have 5G networks, satellite networks. We can guarantee that you never lose communications, you're in constant -- that assistance is always available. And the situation doesn't get out of control. The EMT doesn't make a mistake and lose a patient. Police officer doesn't make a mistake and create a terrible problem for the community, that a person in the fire department doesn't make a mistake and get trapped because the window has shifted, and now the fire is coming in their direction when they thought they were in a safe space. So we can -- so we built that system and we built that system very, very quickly using code generation, generating the application. And this is -- we never could have taken on that problem. Same thing the greenhouse, all of the applications we're building in the greenhouse, I shouldn't say all of them, but all -- because we're using NetSuite in the greenhouse and some -- but all of the new stuff, all of the new stuff, all that we we're using, we're using application generation. So we wouldn't have taken on those kinds of problems. It would have been foolish for us to try to take on those kind of problems. If we didn't have a much better tool set than we had even 5 years ago. So it just opened up all of these opportunities for us to tackle problems we just never had the bandwidth to tackle. Now our new toolset allows us to do that. Right there.

Kash Rangan with Goldman Sachs. Good to see you, and it's fantastic to see you, attend every single Analyst Day for the past several years, and it's so much fun. I hope it's fun for you as well.

It is. I am enjoying this. People ask me why I do it? There's a lot of answers, but one of them is I really enjoy. I like -- I like the people I work with. The work is challenging and rewarding, and I'm not going any place.

We see Larry Ellison university every year. It's so much fun. My question for you is, there are some people that believe that Generative AI is going to lead to so much efficiencies that the employment picture, whether it's sales, marketing people, customer support developer, that population is going to shrink. I'm just curious to get your thoughts whether you believe in that doomsday scenario? And the other question is, besides Oracle, who wins in Generative AI and who loses among the tech companies that we've all grown up with?

Okay. I think by and large, Generative AI helps people do their jobs. People with really hard jobs. Now let's say, certain kinds of customer service. We've pretty much crowd-sourced some of our customer service already, right, to other customers. Customers will talk to other customers or write questions down and it gets crowd-sourced. Frequently asked questions, Generative AI is great at doing some of those things. But, I don't think we're going to run out of questions. I don't think that's going to happen. And some -- and eventually, human beings are going to be in the loop. For the docs, where we now are using generative AI to write doctors' orders, but we're not writing doctor orders. We're writing drafts of doctors' orders, right? We're not doctors. We are not licensed -- the computer program is not a licensed physician, cannot discharge someone from the hospital. They're not authorized to do it. What we can do is write a coherent order draft of the coherent order, so the doctor can come in, look at that, make a couple of simple edits to the order, approve the order, sign it and get that done very quickly, which frees up more time. So what is the doctor going to do with her extra time? Well, she's going to spend it with patients. Sure she is not in love with writing discharge orders or a bunch of other things. So we're going to take people who are very busy and allow them to concentrate on the parts of the job -- on parts of their job that are most important and help them handle some of the other tasks, which are complex, but can be automated away. Therefore, we're going to recapture their time, and they're going to be more effective in what they do and they're going to enjoy their job more as well. So I think robots make your job easier. I pointed out, people worry about robots in agriculture, but what we do in these automated greenhouses, we lift rows of plants up out of the growing area, move it into the harvesting area. There are no people allowed in the growing area because people can contaminate the food. So it's a -- growing area is a clean room. We lift up this big heavy -- this robot lifts up, rows of plants, moves into the harvesting area or transplant area. And then if you're involved in harvesting, you're sitting down and you're in an air-conditioned room. This is not something that Cesar Chavez would have recognized. This is -- these are not people leaning over in 115-degree temperature, bright sunlight all day in a backbreaking difficult, difficult job. Now the agriculture jobs are much more pleasant, but they haven't gone away. They've changed, but they haven't gone away. So I think there are robots, whether they're large language models or there are robots who are lifting plants and helping transplant the plants. We're not going to take the human beings out of the loop. The human beings are going to be doing different things. They used to turn over the soil with a shovel and then now they drive fancy tractors. They don't even drive the tractors anymore. The tractors drive themselves. But the farmers are more productive than ever. And that's fine. I think it's going to happen across the economy. And we will be as human beings will live more prosperous and happier lives as a result. Oh, the winners and losers in AI? Well, okay, winners, Tesla. Big winner NAI. Very, it's going to be very difficult for the other car companies to compete with the self-driving part portion of it. And there -- I don't know if you have noticed, they're finding -- even the electric part is not so easy. Building a few electric car is pretty easy. I'm not going to pick on Rivian, but building thousands of electric cars, not so hard, building millions of electric cars. Let me show you the robotic factory. So, I think Tesla is a definitely going to be a winner. There are a lot of good -- no, actually, they're not. There are a few good groups. I think Google is pretty good at this. I think Facebook is pretty good at this. I think Open AI is pretty good at this. I mean Google and I say Google, I really mean specifically DeepMind in London. I think, obviously, NVIDIA. I think we're going to be very successful with this. Cohere. Then there's this whole application space, I think you're going to see new generation pharma companies that are based entirely on computational molecular dynamics. In other words, they're not going to be designing molecules in traditional chem labs -- they're going to be designing molecules on computers and then making those molecules and testing them in toxicology and then go straight into Phase I clinical trials. I think that's going to be made. But you'll see new generations of pharma companies. And some of the traditional pharma companies partnering with these new AI pharma companies. Oh my God, it's going to be -- in terms of applied AI, it's going to be all over the place, right? So -- I mean it's such a broad-based fundamental technology, it's going to impact -- again, are the writers in Hollywood wrong for being worried? I think some are probably right in being worried and others have nothing to worry about. So there are going to be some jobs that are lost. But overall, in terms of productivity in the economy, the productivity in the economy will just skyrocket, I think. Over here. You two guys are very close to each other. We'll do both.

Derrick Wood at TD Cowen. I heard back on the app side, I heard Steve mention at a keynote a couple of days ago that 5,000 customers have migrated to Fusion. Fusion has been out for a little over a decade. I think I went back to my notes and I have it in 2014 that you guys had 27,000 total application customers. So that's less than 20% of customers that have migrated over the last decade. I'm just curious, as you look at the next 5 to 10 years, do you think, we see a linear kind of chipping away at migration over time? Or do you think that we see more exponential kind of adoption curve?

Yes. I think there's now more -- every year, there's more and more differentiation that what we have in the cloud becomes that much better than what we had on-premise before with the e-business suite. But you're right, people are very happy with the e-business suite and people and some of them have yet to migrate. But Safra and I were just talking about some of the big companies, some of the very large companies who love the e-business suite, I mean really and are experts in the e-business suite know more about it than we do. So are now migrating to the cloud because the new version of Fusion, some of the things -- some of the B2B commerce stuff that we're automating just never existed. It would have been impossible to do with the old on-premise technology. So Fusion continues to get better every year, and the e-business suite doesn't. It's a little better, but it isn't making these huge strides. So I think we're going to be -- we are -- in fact, we're seeing acceleration there, larger companies. And we're definitely seeing -- winning share from other companies. And I say Berkshire Hathaway, all the Berkshire Hathaway utilities picked us, and they really were a combination of us and some -- I can't really mention the name of the company, but they're a German ERP company, but they were half us and this German ERP company and now they're standardizing on us. I think you're going to see more and more of this. Fusion is very successful, but nothing like it's going to be. I mean it's just still early days, still early days. The gentleman right next to you?

Larry, Sebastien Naji from William Blair. I wanted to ask about the Gen AI opportunity in OCI. But specifically, I think it's very clear that on the training side, Oracle has pretty substantial advantages versus some of the CSPs. As this market sorts to shift more to Inference. How does Oracle carry some of those advantages over and remain differentiated versus the CSPs in Inference? And maybe as a corollary to that, just generally, how do you think about how that market ends up in steady state? Is it a 50-50 split? Is it 90-10 to the other way? What are your high-level thoughts there?

Well, thank God for the RDMA network. The RDMA network helps enormously for inference. You don't necessarily need GPUs, but you're still moving a lot of data, for inference. And so the Oracle -- OCI became obviously good when faced with this AI training problem. It exposed -- well, OCI really is different. And because we are faster and the faster will help us in inferencing as well. So we think we can hold on to a significant share. Furthermore, these training never stops. We think, okay, well, I've trained my foundational model, I specialize it with a little bit of data. I'm done. I'm just doing inferencing, inferencing, inferencing. No, you improve -- when you improve your neural network, you come out with the next version of the neural network, which comes out in 6 months. You have to. Oh my God head, redo the training. So as you constantly upgrade this technology, you have to retrain the next generation, the next version of the neural network. You have to specialize the next version of the neural network. We're always collecting more EHR data, more medical research and scientific research is being published and you have to put that into the system. The training never ends. But we do quite well on the inferencing. We're already -- we've built some of the inferencing systems, and we're very competitive and cheaper on inferencing. Not as much as we're cheaper on training, but we're significantly cheaper and faster on inferencing as well. So I'm pretty confident -- and then the market is going to grow at this crazy rate. So I think -- I think we're going to do fine in the AI space. But there are a bunch of spaces also adjacent to AI where this is helping, this AI is helping the Oracle brand, for example. Just the general notion that, okay, we had one person say, I wasn't really a believer in OCI and now wait a second, maybe you guys do have a few things where you're better. So our brand is greatly enhanced as a result of this, which helps all of the other businesses. So no, we think this is sustainable and it will compound and that our future is bright and Safra is telling me if I don't get out of here, she's going to be the next one to leave. So thank you all very much.

Okay. We have covered a lot of ground today. Those who stuck around, I just say thank you. That obviously was a very extended session with Larry. We don't get those often. And so hopefully, it was a value to you all. Everybody, safe travels going home. If you have any follow-up questions. Just give us a shout tomorrow or day after. Thank you again. Take care, Bye-bye.