Oracle Corporation (ORCL) Earnings Call Transcript & Summary

Good morning, everyone. We are about to get started, just letting everyone else -- all the other participants get on, and then we'll get underway. Thanks. All right, everyone. We still have a few more people trickling on, so we'll give it just a couple of minutes. Hello, everyone. It's time to get started. My name is Mark Andres. I'm a Global Vice President for Defense and Intelligence at Oracle. I'm co-hosting this event with my good friend, Steve Durr, Vice President of Cloud Operations and Engineering at Oracle. We're excited to speak with you today about the recent news of Oracle Cloud being approved for use on top secret missions. Just look at the world around us with the horrible conflict in Ukraine, with what erupted over the weekend in Israel and now warships heading toward the conflict as they always do. We're reminded of the criticality of being able to provide the best capabilities we can provide at the highest classifications. Just 6 months ago, I was CIO at MGA. During my tenure, we really only had 1 cloud service provider on contract at the top secret level. And now we have -- we are 1 of 3 CSPs capable of providing top secret cloud services. When I think about national security, I think about data volumes, the use of new technologies such as AI, the increasing risk of cyber security. All of these things led to this promoting of hybrid and multi-cloud environments to address these demands. I've only been here a few months but I can assure you that from the very founder, Larry Ellison, to the CEO, Safra Catz, there's an unwavering commitment from Oracle to support these critical missions. All right. What have we got for you today? We planned an introductory brief by Steve Durr, kind of an Oracle 101, what is Oracle Cloud for top secret, where are we available, what are our day 1 services. And then we're going to have a discussion on what you should be thinking about that differentiates Oracle Cloud. Every cloud is different, each brings special attributes, and we're going to dig into Oracle special sauce. Then we're going to open it up for questions from all of you. You can post your questions at any time in the Q&A block. Some of these will answer in the queue back to you in the Q&A block and others will take verbally. On an administrative note, the session is being recorded so that we can make it available for those that weren't able to attend. However, once Steve and I finish the opening discussion, we'll stop the recording as we move into Q&A out of respect for the price. So with that, let's get started. Steve, over to you.

All right. Thank you, Mark, and thank you, everybody, for attending this afternoon. Let me go into presentation mode here. So as Mark said, I'm Steve Durr, I'm VP of Cloud Operations and Engineering for Government & Defense and Intelligence here at Oracle. I've been working on the deployment of our classified clouds for a number of years now. And finally, that came to fruition at the end of July and August with accreditation of both our top secret and secret environment. So what we're going to do today is talk a little bit about Oracle Cloud in general and then the top secret environment itself and what services we're bringing to market. So for Oracle's global cloud deployment, we're at 45 regions total across the globe in our commercial environment. And this commercial environment is very duplicative across all of these with equal capabilities in all of the regions and a consistent price point across all those environments as well. In addition, we have our partnership with Microsoft that we like to always highlight with the 12 Azure interconnect regions on the commercial side. Those are regions that are jointly located with Oracle, and we're able to provide high-speed, low-latency interconnects between Microsoft's regions and ours to offer joint services. We also go beyond just the network connectivity, and we talk when we deliver identity as well as the ability to jointly provision services across that foundry. We'd love to bring this capability to the Department of Defense and intelligence community as the need arises. From a U.S.-based dedicated cloud region's perspective, we do have multiple regions across all classification levels. So we have 2 regions that we dedicate the FedRAMP High. Those 2 regions are open to state and local customers as well. When you get into the Department of Defense regions, those are physically isolated from each other. They are located in 3 geographies across the U.S. for redundancy and they are fully impact Level 5 approved and authorized for use for national security systems and control and cost side information. When we get into the secret environment, there's 3 of those as well, geographically dispersed, 3 different power grids. And then for the top secret environment, that is also spread across 2 different power grids for survivability. From an operations perspective, you'll see the cloud network operation centers on the screen as well. Those are ICD 705 facilities. They're isolated from each other between secret and top secret. We did build our secret and top secret data centers to an ICD 705 standard. So at both secret and top secret ICD 705 we do have separate CNOX at each of these locations, one for secret and one for top secret at each one of the locations on the screen. From an Oracle Cloud basic perspective and the foundation, so what we've built and what we brought to market for all of those commercial regions that you saw, the 45 cost and then the national security regions as well as our unclassified government regions. We brought the market a bare metal cloud. So the entire environment is bare metal by default. We don't have a bare metal service that we offer in specific regions or localities. The entire cloud is based off of bare metal. And we did that in order to provide superior isolation for your workloads as well as superior performance. So we, as a cloud service provider, are not co-located on your host, we don't run our cloud control plane software, taking up virtual machines on your host, you have the entire host. And we enable that to the off-box virtualization on the left-hand side, so that's our ability to run a separate cloud control computer as well as a network virtualization host that encapsulates the traffic and keeps it secure between clusters of your host as well as to shared services like Walker Object Storage. When you're done using that bare metal host and you turn it back into the environment for reuse, we have a set of custom security chips. Since we build our own host, we're able to heavily secure those hosts and make sure that they're white before they get returned back into the pool at that same classification level. The other thing that we've done to enable the high performance of our cloud is non-blocking networks. So you'll find that we have a non-blocking network architecture that interconnects our standard host, I'll call it, from a layer 2 virtualization perspective. And we had to enable that capability for things like Oracle Database and real application clusters to ensure we could meet the maximum availability architecture requirements of our customers for mission-critical workloads. And then on the RDMA cluster networking side. So we're using RDMA over cluster Ethernet to interconnect our high-performance compute, our GPUs and our Exadata infrastructure to ensure you have the highest performance 100-gig network with low latency to support those mission-critical applications. On the flexible infrastructure side, that's fairly new. Back in the early days of cloud, you used to get a fixed shape that would come with a certain amount of compute and a certain amount of memory, and that's all you would get. In the case now, we're able to flex that infrastructure where you as a consumer can independently tune your compute and your memory to meet your application and cost optimize what you're trying to do. So if you have a compute-intensive workload that doesn't require a lot of memory, you can tune your environment to that to enable the best performance and price for your workload. From a cloud service perspective, so we are a hyperscale cloud service provider. As part of our accreditation, there are over 50 services. They were part of our initial accreditation. So we're not starting out with 1 or 2. We started out with 50-plus-out of the gate, 57 to be exact. And those services range from your basics of compute and storage and networking. So your block, your object, your file storage systems, the networking I mentioned already, a wide variety of compute processors, including ARM processors, Intel, AMD, and in addition to that, obviously, we're a world leader in data management and we have a large-scale data platform that runs all manner of databases from SQL type databases, JSON and XML and really pulls that in. I'll talk later on about the modern data platform. From an analytics and data science perspective, we have a suite of capabilities with the cloud analytics that we're bringing to market. And our acquisition of data science a number of years ago has turned into one of our most popular cloud services. On the DevSecOps environment on the top left, we have a full suite of the DevSecOps capability that we bring to market. And we do run container native services. We have the capability to support that to your normal open source tools. And then on the machine learning and AI front, Oracle has been doing machine learning for a number of years inside the database itself as well as to a certain extent within our applications. With the artificial intelligence aspects, we're able to enhance even more of the database capabilities as well as bring in an artificial intelligence component with our partnership with Cohere. And on the native VMware front, that Layer 2 networking that I mentioned earlier, that's very good for real application clusters and maximum availability architectures of the database. That also works very well for native VMware. So you can use the Oracle Cloud to extend VMware from your on-premise environment to the classified cloud and manage all of that with your V center capability. So Mark, over to you.

Great. Thanks, Steve. I thought it was really important to do that upfront. A lot of what we're getting when we're out there talking to our customers is you have a cloud, and then we get, well, is it a database cloud. And so hopefully, you realize that this is a full service, full scale globally distributed set of capabilities being offered on all security fabrics. So thanks for going through that. So this next part I was going to try to talk a little bit more about what are the differentiators. As I said, each cloud differentiates itself a little bit. And these are some of the things that we really thought were really distinguishing for Oracle Cloud that we want you to think about. But before I even get into these, let's start with cost. There's not a CIO on the planet that doesn't manage cost every day as a part of every decision, every program build, every quarterly review. And so while cost isn't specifically listed here, I would encourage everyone to revisit what they're paying today for cloud in a world where multiple cloud providers with different cost models are truly driving competition and value differentiation to the government. We'll hit on it briefly at the end, but I'll give you 1 example that when we talk to people, they question whether or not it's true, we say, for example, we have no egress fees for classified missions. And they go, what you really mean by no egress fees, and we mean 0, we mean none. And we felt that was important because that might affect how you're -- it could suboptimize your architecture. It could make you think about cost in a different manner, both near term and long term. So we'll go into those cost drivers later. I just wanted to highlight that before we got into these. When I was at MGA, my most critical missions ran on Oracle, when you look at these kind of high-performance database systems that require 0 tolerance for failure, Oracle, for decades, has been just the leading provider of the most advanced data platforms in the world. So like Steve was saying, we want you to think about Oracle Cloud as a data platform that has a cloud-based compute and storage built around it. Everything in Oracle for multiple decades is rolled into how this team design and build the cloud. The next thing is this concept of a Generation 2 Cloud. Because it came slightly later to the market, it was able to do what no other CSP can do, which is learn from everything that went into first-gen cloud providers. One of those that Steve just highlighted was disability to easily lift and shift to bare metal compute without reconfiguration. That was always tough. And he also highlighted that it made it easier through a close partnership with VMware. So this is one that I really think would have been a great benefit to me. I think the other thing I like is that Oracle Cloud is uniquely able to extend its hyperscale environment into your data center for hybrids, or all the way out to the tactical edge. And we'll go into this a little bit, but you're going to see it's -- when we're talking about extending from hyperscale, let's say, to an on-premise OCONUS, we mean truly moving our Oracle Cloud infrastructure out there. So pay close attention to that. The other one for me, I had no cost model for HPC workloads other than on-prem. There just wasn't a real price to performance solution out there. So most of the stuff we did was on-prem. And I think Oracle Cloud comes in with both a design, a technical design and some cost drivers that really make it something that I would have looked at heavily had it been available to me. The rest of these, we're going to go into Oracle Cloud's approach for Zero Trust, which is being reviewed by DoD CIO for its value as well as many of these enterprise-level applications such as human capital financials that are available. So Steve, if you can go to the next slide, I want to start with a deeper discussion on AI. Okay, this slide captures our strategy and approach to data and AI. And it's broken out into these 3 areas. And I'll begin with the infrastructure that Oracle Cloud as a second-generation cloud brings. As I stated for me, HPC on-premise because of the pricing performance ratio couldn't be met in the cloud. That paradigm has been disrupted. And Steve is going to literally walk through what that design means and what it means to you. But again, with cloud, and in particular, HPC and in particular, with big data, AI and ML, both model training cost and inferencing, we're hoping to provide lower cost while giving you the resiliency, scalability and the ease of use that you'd expect from cloud. The second area builds upon the simple fact that Oracle is a data platform company recognized as a world leader in data management, data security, data replication, open interfaces, et cetera. And data management is the critical first step in AI implementation. The better you control and understand all the data you have, the more rapidly you'll be able to advance in AI/ML outcomes with greater confidence in what those outcomes are generating. Data management is just so important as you begin your AI/ML journey. And I also mentioned how open Oracle's architecture is. Oracle allows for rapid integration of prebuilt AI technologies, the ones that you like, you may want to bring in as well as close partnerships with leading AI companies such as NVIDIA, Cohere and Mosaic. So think of Oracle's approach to AI as rapidly adaptable to how you want to approach AI. And then finally, Oracle strategy focuses on all of these enterprise applications that many of you have running today. PeopleSoft moving to HCM, Oracle Financials, supply chain. All of these are SaaS offerings that when we bring those to the government, we bring the learning and data and advances that come from a global -- literally hundreds of thousands of customers globally to embed AI capabilities and make them available to the federal government as well as the ability to customize them for your specific needs. So with that, Steve, I think I'll turn it back over to you to give them a little more detail on why this works so well inside Oracle.

Very good. Thank you, Mark. So on one of Mark's earlier points, right, the -- in order for AI to be successful and be able to train models that don't give you hallucinatory results or another bad outcome, the curation and handling of the data sets that will be used to train the AI models is greatly important. So what Oracle brings to the table in that environment is on the left-hand side, you'll see there's a set of data producers. We can integrate them with any database, doesn't have to be just an Oracle database, any database, we can pull data from, any set of applications that you want to provide that has a data set, we can also pull that in and also across clouds. So there's multiple blogs and press releases out there about how we integrate in to pull data from other cloud providers as well as any event sensors or real-time streaming data that you want to pull in, we can do that with open source Apache Kafka type technologies. When we absorb that data, GoldenGate is a very familiar technology. It's been around for a long time. It's our primary adapter to be able to pull in multiple data sets into the Oracle database. And then on top of that, there's the Oracle Data Integration Cloud service that allows us to capture the rest through a common set of APIs. Within that lake house construct in the center of the screen, this is where we'll pull the data in. We curate it, we tag it, we secure it, and we present in a catalog or an open search type format where on the analytics side, it can be consumed by using Oracle Analytics Cloud or can be consumed through our data science service or through other AI services as we bring those to market. In addition, you can tie into this lake house and with your own custom applications using open AI, open APIs to be able to pull data from that into your own application. So we're really trying to provide the ability to pull in multiple data sets, have that data set be trusted in order to train your AI algorithms on top of that. And all of that takes a significant amount of horsepower. So to Mark's point, cloud-based HPC and GPUs have been a bit of a challenge from a price perspective and performance. So Oracle has brought a cloud of the market. We have a very specific RDMA cluster and networking capability that we have deployed across all of our classification levels. That was one of the table stakes requirements as we build out clouds for government specifically across the unclassified environment and secret and top secret, was to ensure this functionality was embedded into the environment. And with that, we can run up to 20,000 HPC CPU cluster within each of those regions, you saw on that map. And we can expand that from there, but the initial capability consists of that 20,000. We can also handle GPU clusters as well as being able to interconnect into the Exadata platform over that same RDMA cluster networking. So that gives us a very high speed and very performant environment to work on. In addition, because of the cost and performance ratios of this, this is where companies like NVIDIA run their DGX cloud service on top of Oracle Cloud commercially. We're working to bring that DGX capability to the classified side and many other AI companies are using Oracle's Cloud because of the performance and cost benefits they get when they're leveraging high numbers of GPUs. In one of the benchmarks that we recently completed for the intelligence community, our cloud was 4x faster at 120th of the cost. So we were using less CPUs to get what we needed. And with that, it ran a lot faster and also drove down the cost. So it's a significant cost savings. It's not a 10% or 15% cost savings. It was 120 for that particular benchmark. We talked about VMware before. So the Oracle Cloud VMware service or solution, is the ability for Oracle Cloud and the way that we've established our joint service with VMware is to be able to run VMware as natively as possible on top of our cloud using common tools that the VMware operators are used to using every single day like vCenter, utilizing the vSAN capabilities and the VMware vSphere along with vMotion to be able to move virtual machines between on-premise and the cloud seamlessly, either for geographic survivability or for expansion and scale or as part of your migration strategy to 100% in the cloud. So what we've been able to deliver across the classified environments as well as unclassified is the migration without modification. So you're just using VMware and vCenter to operate the environment. There's no retooling or retraining. Your VMware operators operate the same way they do today. The costs are extremely predictable on our cloud service. So there's not a lot of variability to that. So from a budgeting perspective, it's easy to budget for. And then it does operate in the top secret environment. So some clouds have challenges when they cross that security boundary. Ours do not, because we build our cloud organically between unclassified, secret, top secret from a capability perspective. So if the cloud service operates on the low side, it will operate exactly the same on the high side, which provides you the ability to also do development and testing on the low side with a high level of confidence that will work on the high side. And the next area is Oracle Cloud SaaS. So as Mark was mentioning earlier, Oracle has a long history of business applications from PeopleSoft to Siebel to other capabilities running across government enterprises. And the first one that we're bringing to market from our SaaS portfolio is HCM. So we're bringing human capital management and the capabilities you see here on the screen to market at the top secret level. We expect to begin the joint test team in environment testing that will be in the first quarter of 2024. So we hope to have this available in the 2024 calendar year for everybody to consume. And then we'll move on to other capabilities that you see. The advantage of this is that you migrate off of your older systems into this new modern environment, take advantage of AI and ML and what we're building into the applications to help you create a business advantage with your business application data, something that's been challenging to date. So we're removing that challenge for our customers. From a security perspective, we talked a little bit about this, the second-generation cloud and the bare metal host and this just kind of visualizes what we've done. So on the left-hand side, you see the standard virtualized environment that the standard clouds were built on. And there, you have the cloud service provider sharing the host with you with their cloud control software and a specific VM, doesn't give you full control of the host and it also takes away some performance given you're sharing the CPU and the memory with that cloud service provider or with other tenants. What we've been able to do when we re-architected the cloud, is we've created a bare metal foundation, so you get the entire host that you see on the screen that's dedicated to you. You can virtualize it. You can change the software and the operating system, the software. You can bring your own hypervisor, you can containerize it if you want. So it gives you the same look, feel and capability as if you had an on-premise server. If you would like to, you can virtualize that environment using our virtualization capability that we have in the cloud. So you still get the benefits in isolation and security of a dedicated host, but you're also able to virtualize it for your use using automated tools in the cloud. And then from a security perspective, you see the isolated network virtualization on the bottom. That's where we encapsulate the traffic leaving that host ensure that the policy enforcement exists to route that traffic only to other hosts in the cluster or to share resources like block or object storage in a fully encrypted manner, both in transit and at rest. So in order to support staff workloads, the extra isolation, the control of that bare metal host becomes critically important to ensure that soft data doesn't leak on to any standard classified network. We're very confident in that capability and willing to test that. Then across Zero Trust, it is a critical component of protecting the DoD, intelligence and U.S. government on an agencies data. We have to assume that compromises will happen. And from there, we want to limit the exposure to the rest of the environment based on that compromise. So we have a host of capabilities ranging from the user through device and applications down to the data, all the way down to visibility and analytics across our cloud to support the pillars of Zero Trust. We've been working with the Department of Defense, Zero Trust Performance Management Office for over a year. We started in July of 2022. We just completed our Zero Trust Red Team testing in the month of September. We did very well in that environment with the Red Team test that consisted of people from DoD [indiscernible], NSA, et cetera. So it was a very robust test that was done and these cloud services that you see here, many of them that are security focused are free of charge. So to Mark's point earlier around cost savings, we don't want you to have to make a trade-off between security and cost when it comes to protecting mission-critical environments across the Department of Defense and intelligence community. And moving into the extensibility of our cloud from the core to the edge, so what we've done is you have your global cloud regions, you can equate that to both the larger math that you saw with the 45 regions on it or you can equate that to the CONUS environment for the government dedicated clouds that we have and no specific cloud rounds with multiple regions in them. The other offering that we have as we move our cloud capability to smaller and smaller form factors is a dedicated region. We can do that at the unclassified level or the classified level where we can take a full service suite of our cloud capabilities, deploy that in your data center, CONUS or OCONUS and give you a full suite of our cloud capabilities. So this isn't a limited set. This is the full set of cloud services, the same as you would get in the CONUS. We can do that on a 13 rack minimum footprint and then we customize the environment based on the services you want to scale. Do you want more compute? Do you want more GPUs? Do you want additional storage, Exadata service we'll customize the hardware to meet the specific mission application that you want to accomplish at that location. And it's only remedied by your power space and cooling within your facility. The scalability of the dedicated region is the same as any of our CONUS regions and can scale well beyond any government facility that I've seen anyway. From our cloud and customer perspective, so this is an extension of the cloud where we bring specific componentry into your data center, much smaller scale down to a single rack. We can extend from the cloud to get the same service capability, but you get a specific function such as Exadata or a private cloud appliance in that environment. As you move closer to the edge, you start to get into smaller footprints. So you see the Conix container at the center -- bottom of your screen, that's where we deploy the -- our edge cloud services into that container, fully transportable, both air and ship, comes with built-in HVAC, security, shielding as well as biometrics and video for monitoring the environment and then we can put either roving edge devices into that environment or the roving edge device can be clustered on its own, as you see down in the lower picture where you can stack them in a forward operating base environment or in a tent. We've deployed Roving Edge devices on aircraft as well as ships for testing during specific exercises or you can run it as an individual Roving Edge device, as you see there. In addition, we offer a very small ultra device, that's a backpack portable battery-powered solution, so you can put it in a backpack and take off with your data and analytics embedded in that capability. So we -- our mission with the cloud is extensibility and extension from the core to Edge with the same capabilities and putting the power at the tactical edge where the fight is going to happen as well as making sure that we can synchronize data from the cloud to those devices, using technologies like GoldenGate or we can operate in a DoD environment where there are disconnected comms or intermittent comms where we're able to continue to operate also with our Roving Edge device in Ultra, you can leave them disconnected from the cloud permanently. So for certain missions, you may want to preconfigure the Roving Edge device and then deploy it and never connect it back to the cloud. So we also offer the ability to do out-of-band updates to the device where you never have to connect it back to the cloud. And with that, I will turn it over to Mark for the final turn.

Thanks, Steve. I want to full stop a couple of those. I was very impressed when I got here to see essentially an out-of-the-box Zero Trust approach and then to have that tested not as red teaming your infrastructure, but red teaming and end-to-end Zero Trust capability and saying, this is good. So I know all of the federal CIOs have Zero Trust mandates huge with DoD. This is a huge accelerant for that. And then the last one, I thought Steve did a great job talking about it. But when I was talking about having compute from hyperscale to the Edge, a lot of people have that, but then you take all of Oracle's data technology and replication across that as core functionality of Oracle, it really is a big differentiator. Okay. Last slide is price. So I'll hit on these, but after I do -- as I do, what I'd like you to think about is don't just believe us, follow up and run your own price models, work with us on what the workload is you are exploring, either if you're an SI supporting the government or you're the government, come back to us and say this is where we are. Here's what we're doing and we'll work with you on what does that look like in Oracle. The first one is commercial price parity. This gets to the fact that Oracle as a company, made a commitment that when they go to develop a new cloud capability, that capability is required to be able to run on all from commercial to top secret, not they're going to push it out to commercial and then some day to figure out how to run it on top secret. So by doing that and by having a common compute environment for every secured domain, it means that we can deliver capabilities quickly to the government. The government will still have to certify those and that can take time. But from Oracle's perspective, we come in with a capability, it can be applied in multiple security fabrics. With the contracts that have come out, they have significant discounts applied to those. Those are discoverable or if you can't find them, let us know. No change in price when you're operating on secret or top secret for the same service. So no uplift. I already hit on egress fees. To me, that is so important, not just [indiscernible]. Why would the government -- have you pay to host your data and then charge you to move it. It seems a little counterintuitive to me, but there is, not at Oracle. And I would ask you to think about what does that mean from a mission perspective. if you've got a high throughput system, pushing data at different places, there's the early egress fees. If you are really wanting to explore, as I started this off with, hybrid or multi-cloud environments, database on one end, the app on the other, storage somewhere else. Egress fees won't complicate that. Steve already hit on the security services. The majority of those they're there. it's good for you, it's good for us. We want your data to be as secure as possible. So we've made those -- the majority of those free to you, you don't pay additional licensing for those. And then this last one applies to -- it depends on your ULA contract, but there's an incentive that if you have a contract for Oracle software, there are ways in which your spend in the cloud further offsets that price. So discounts the price you paid for your ULA. And I think that is it.

So next slide, we're going to go into Q&A. I see the team has been already answering several in the box. Yes, and we just stopped the recording. So standing by.

I think to full stop on Mark's comments on the pricing because we get this question quite a bit. Is that our commercial pricing starts with list price and then we discount that pretty heavily for the major contracts that are well known out there. And then in addition to that, we do not charge uplift. So if it costs you $100 to run it, get at the unclassified level and you do development there, you can expect to, say, pay that same price at secret and top secret. So there really are no uplift and are no egress fees. It makes a significant difference as a government mission owner and a program manager to be able to budget as you move your workloads from the low side to the high side. Well, I think, let's give it a minute. I am not seeing any new questions in the Q&A. So if you have questions, please feel free to ask.

All right. Well, and you see on the screen where if you get a question after this, that you didn't want to post in the chat or -- we got another one there going on, but be sure to grab that gdi-events_us@oracle.com, and we can follow-up quickly afterwards.

Mark, we had 1 question coming in.