Hewlett Packard Enterprise Company (HPE) Earnings Call Transcript & Summary

Hello, and welcome to Hewlett Packard Enterprise Discover 2024 Las Vegas. I'm Daisy McAndrew coming to you from Sphere. We are only moments away from the first-ever keynote address at this iconic location. HPE President and CEO, Antonio Neri, will be taking to that stage in just a moment. But I've got time to get us a little bit of insider intel. Delighted to be joined, as you can see, by HPE Chief Marketing Officer, Jim Jackson. Jim, before we talk about content, let's talk about this place. Why here?

Absolutely. First of all, hi, Daisy. Good to see you. We're setting a new standard for brand storytelling this week. And we've got one heck of a story for our customers, and we needed a venue to match that moment. This place is incredible. It's filling up. We're going to have a sold-out house.

It really is buzzing. All right. I'm hoping you can spill the beans a little bit for me, Jim. What can we expect to hear from Antonio?

Yes. I can't give away all the details, but everyone is going to hear about bold new AI solutions that reduce complexity and help our enterprise customers accelerate the adoption of generative AI. So we're excited about that. Antonio is backstage. He's excited. He's ready to go. He's going to take the stage shortly. He's going to be joined by Jensen, who is the CEO of NVIDIA. They're going to talk about our partnership, which is over a decade long, and how we're co-developing new solutions for our customers.

Okay. We've got a lot to look forward to, not just here at Sphere, but the rest of Discover. What else is going on?

Yes. I mean, it continues. We've got 2 more days. We've got the biggest technology showcase that we've ever done. Seven acres, 285,000 square feet of technology that people can touch and feel. We've got over 400 demonstrations, hundreds of sessions, where attendees can go deeper on our technology and trends. And then we're going to have some fun, Daisy. We've been working on this for months. We are beyond excited to do the first keynote in Sphere. I hope everybody enjoys the show.

I'm sure they will. But on that note, Jim, we haven't got a lot of time so I think it's time that you and I went and found our seats before Antonio takes to that stage.

Absolutely. Let's do it. Let's do it.

Ladies and gentlemen, up next, Antonio Neri.

Thank you. [Presentation]

Please welcome, HPE President and Chief Executive Officer, Antonio Neri.

Wow. Wasn't that amazing? Welcome. What an amazing venue. Big moments require big venues. Welcome to my living room. I hope you're going to have an amazing experience here today. It is an honor to be with you at the Sphere and give the first ever, ever keynote in this space. Thank you for joining us today and for being part of the Hewlett Packard Enterprise community. At HPE, we believe intelligence has no limits, and what a great representation in that video. When the human mind is in concert with AI, there is nothing we cannot achieve together. Since ancient times, human beings have had a thirst for knowledge, which has driven us to explore the world and push the boundaries of what is possible. Starting with Hewlett-Packard Company's founding over 80 years ago, HPE has been at the leading edge of innovation, empowering enterprises with inspiring tools to spark imagination and turn ideas into reality. Bill Hewlett and Dave Packard transformed the movie industry with the audio oscillator and accelerated finance and engineering with the groundbreaking HP calculator. And today, we are at the forefront of reshaping life sciences with the advent of Frontier, the world's fastest supercomputer. HPE has consistently provided the essential technologies that transform enterprises to an entire industry. Everything we have accomplished has led us to this moment today. We are, again, leading a revolution on what we anticipated and adapted for the rapidly transforming HPE into an entirely new company. I stand before you, full of inspiration, as we arrive at the AI solutions that will accelerate the generative AI industrial revolution. Through breakthrough organic innovation, targeted acquisitions and key visionary partners, we have propelled our Edge-to-Cloud innovation agenda forward and strengthen our Edge-to-Cloud journey now leading to AI. This morning, I will make bold announcements that bring us closer to the potential and the promise of AI and catapult the enterprise of today and tomorrow to new heights. Let us begin with how HPE works with organizations of all sizes across industries to harness the AI as a force for good in order to change lives for the better. From renewable energy to helping solve food scarcity to drug discovery, HPE enables our customers to harness the power of AI as a driving force for good. For instance, HPE proudly offered its patented innovations and AI systems to accelerate the world's response to the COVID pandemic. We provide a supercomputing software and application expertise as well as our systems to help research port, run, and optimize essential applications for their research. HPE exists to advance the way people work, and AI accelerates our purpose in profound new ways. To be successful, we all must be purpose-driven and we need AI that we can trust. That's why in 2019, Hewlett Packard Labs established 5 key AI ethics principles: the first is privacy; second, AI must be human-focused; third, AI must also be inclusive, ensuring access for everyone; fourth, AI must be responsible; and last, AI must be robust and continuously quality tested. At HPE, we are stewards of AI, upholding our principles with unwavering integrity. We are all deeply aware of the transformative power of AI, but AI is hard. It is complicated, and it comes even with full of risks. It is tempting to rush in with AI. No one wants to get left behind. Yet innovation at any cost is very dangerous. Example of what can go wrong happen every day. We now question everything we see in here, and cyber threats are everywhere. In fact, a recent HPE survey of 2,500 IT leaders across industry found that 94% say that AI has increased their enterprise security risks. Following our AI ethics principles, Hewlett Packard Labs is researching, developing and deploying technology to help protect privacy and stop attacks. With enhanced security, our enterprise on-ramps to AI provide the guardrails to help ensure data sovereignty and brand protection. This empowers you to sense, learn and react smartly so that you can turn insight interaction in an instant. HPE is a complete AI solution partner for your enterprise, regardless of the size. When we connect everything inside your enterprise with the right expertise from Edge-to-Cloud, business transformation happens, and intelligence has no limits. It all starts with data. It is your most valuable asset. I have said for years that one day, data will be recorded in your balance sheet, and that day is coming. It's coming really fast. When you can extract all the knowledge that lives in your data, we believe your enterprise can truly have no limits. HPE is a trailblazer in the world of technology, and we are there for you to serve as your trusted guide. For most of you, utilizing generative AI at the edge is a top priority, and that means training existing models and running inferencing at the edge to solve business problems like content generation to accelerate product design and optimization, enhancing customer support experiences while reducing cost through conversational AI; using computer vision for quality assurance, security and streamlining operations; and even to building virtual assistants to increase productivity. In fact, if you have a question for me, just ask my virtual avatar later today on the show floor. Live demonstrations of these examples are waiting for you in the HPE Discover showcase, alongside smart solutions for integrating inference and fine-tuning with your own data across your enterprise and so much more. Whether you're building a small or large language model, HPE has the immense AI compute power and speed to handle the massive amount of data and software needed to train and tune AI models. The Frontier supercomputer's process speed is so powerful. It will take every person on Earth combined, all of us, more than 4 years to do what it can do in just 1 second. That is exascale, a computing achievement equivalent to landing on the moon. Our supercomputers are marvels of engineering, serving as global leaders in performance and leading in sustainability. We have democratized these advancements, making them accessible to enterprises of all sizes. We are proud of our supercomputing leadership. Today, we have 4 of the top 10 fastest supercomputers in the world per the TOP500 List. It is what positions us to lead in the generative AI into the future. We have solved challenges previously thought impossible. And what we innovate for the world's largest systems today with the power of AI system for your enterprise tomorrow, that's the advantage that we have. As we look forward, the next generation of accelerated compute silicon will require more power density. AI systems generate a lot of heat and waste. HPE is at the forefront of applying the latest liquid technology in terms of cooling and has more than 2 decades of experience. Today, HPE has 7 of the top 10 supercomputers on the Green500 List in part due to our liquid cooling leadership. HPE has decades of experience in the design, manufacturing and management of liquid cooled systems, including the data center infrastructure, to reliably deliver the highest level of compute performance. We also have hundreds of patents in the realm of liquid cooling because we are not just concerned with having the fastest or the most popular AI system on Earth. We are driving innovation that consider the impact on the planet for future generations. And when you start listening to our planet, it talks back. It talks back to us: oceans, glaciers, desert sands and all the living things have stories to tell. These immersive sounds are what inspired the researchers at the Purdue University Center for Global Soundscapes, and for a good reason. By listening to our planet, we can discover things we cannot see. Our work at the Purdue University is an example of an AI inferencing. We are helping the university to preserve these stories by collecting, tagging and applying AI to massive audio data sets so they can understand, we can all understand, the patterns and help preserve a beautiful planet for generations to come. To date, the Center for Global Soundscapes has gathered 6 million pieces of quality data from 600 different places in the world, over 1.2 petabytes of data using HPE infrastructure, software and services. Carnegie Clean Energy is a great example of a training use case. It relies on HPE's AI expertise and AI systems at the Pawsey Supercomputing Research Centre in Australia to design AI-powered buoys that generate electricity using the ocean's waves, by using reinforcement AI learning to adjust the rhythm and the size of oncoming waves, continuously optimizing the energy production, which has proven to be a very exciting sustainable solution. These examples are a small sample of sets that we have in the vast potential for AI inference, training and tuning, which demonstrates the power of our unique strategy, which is only possible when the connectivity and the data aggregation work as one with accelerated computing. HPE is using AI to change lives for better. The next story hits close to home for many of us. In 2020, over 55 million people were affected by dementia around the world. Sadly, this number doubles every 20 years, and it is predicted to reach 78 million in 2030. But there is hope. A groundbreaking study by DZNE, a leading research institution initially focusing on German patients with dementia, has now expanded globally. Swarm learning is an AI technique developed by HPE and is designed to keep AI training results private, but enables hospital research institutions to keep their patient data secure while privately sharing the learnings, the insights with other teams around the world. Swarm learning is a revolutionary shift from sharing AI training, learnings and insights without sharing an individual private data. Whether it's from Osaka to Auckland and Helsinki to Cambridge, MIT and Stanford, swarm learning is leveraging the memory-centric computing of HPE's AI systems to accelerate early detection, prevention and, ultimately, the cure of dementia and other devastating diseases. That is why DZNE established a swarm learning center to bring together the biggest minds around the world to combat dementia. This is AI at scale, working across borders and institutions to build a better future for all. But no one can do it alone. Collaboration is a must. And at HPE, we have the means to connect the world. Today, the aviation industry is accelerating its focus on reducing its overall carbon footprint. GE Aerospace is at the forefront of this effort. It is leveraging our supercomputing technology to create an AI digital twin to design a revolutionary new efficient engines. Leveraging Frontier's ability to process billions upon billions of operations per second, GE was able to simulate engine air movement with an incredible detail. Let's take a look. [Presentation]

Albert Einstein once said, "Creativity is intelligence having fun." It is obvious that GE Aerospace is having a lot of fun exploring the future of flight with our innovations. As I have shown you where insight lives, HPE is there with the right AI strategy. Bringing excellence in engineering, delivery, support as well as creating smarter IT life cycles, we are your trusted partner in AI. Quality is woven in our culture. Innovation is baked in our DNA. Our track record proves it. There are millions of illustrations of this, from enabling interconnected retail at massive scale at The Home Depot here in the United States, decarbonizing the world through more sustainable IT at Danfoss with our HPE GreenLake Cloud platform, or enabling Carestream Health to use machine learning to reduce training times by 73% and detect tumors earlier to help save lives. HPE helps ensure enterprises of all sizes drive innovation with the power of AI. I spoke earlier about the innovation that HPE has brought to life throughout our storied history. As we accelerate into the generative AI era, even greater innovation will be required to empower your success, and it is HPE and our community of partners who will bring that to you. AI will require hybrid cloud. But AI is not one single thing or a monolithic workload. To deploy AI, you must orchestrate hundreds of micro services, multiple AI models, specific accelerators and connect many different data sources, all of which are highly distributed across your hybrid IT estate. That is why you need a hybrid strategy. At the same time, you must maintain data governance, regulatory compliance and security end-to-end, making on-premise private clouds essential to your hybrid mix. As I mentioned earlier, AI also require a data-first approach. And when it comes to your data, speed matters. According to IDC, 50% of executives says that the data loses value within hours, and only 7% of IT leaders says they can access real-time data. If you don't have your data at the source, then you cannot get it fast enough to use it. Generative AI is a distributed workload. Connecting all your data is essential, which means you require a strong networking foundation from edge to cloud. Unlike other providers with siloed solutions, HPE provides a leading modern and also AI-driven networking fabric. Today, our HPE Aruba Networking portfolio enables customers to deploy modern AI-driven networking. From Walt Disney World Resort here in North America to Engel & Völkers, Tencent and Bancolombia, leading organizations across the world rely on us to connect their businesses with security-first, AI-powered networking. And we cannot forget our HPE Aruba Networking customer community, who we call affectionately Airheads. It is 210,000 members around the world that share information and inspire each other with their ideas. Let's give them a warm welcome here to HPE Discover, which is their first time. Thank you for being here today. Many of you are familiar with Land O'Lakes, one of America's leading agriculture and food cooperatives. Land O'Lakes has a big dream, to connect a dairy and co-op locations to reliable broadband. They turned to HPE Aruba Networking for a solution that will provide more secure indoor and, actually, outdoor wireless connectivity and support mobility for people, devices and machines. By using our intelligent AI solution in its dairy manufacturing and processing plant, Land O'Lakes has greatly improved what we call pick, pack and ship operations efficiency. But it didn't stop there. Land O'Lakes wanted to bridge the gap between technology and its farm operations, bringing Wi-Fi to rural communities that lacked access. HPE Aruba Networking went above and beyond to address the digital divide, empowering farmers with applications that could help them in the daily operations, but also in their personal lives. Now by gathering more secure data from silos to soil, Land O'Lakes can utilize AI to anticipate crop yields. The data is protected and belongs solely to the farmers in the field. Together, we are laying the foundation for AI, revolutionizing the agricultural industry and empowering farmers with the data and innovation they need to thrive. With manufacturing, quality assurance and latency matter. HPE placed cameras at 5 different angles on Relimetrics' assembly lines. The camera feeds images into a series of neural networks, where patterns are recognized, and the computer actually decides to pass or fail in the moment. They went from 21 seconds down to 1 second, which is a 2,000% improvement. I think that's the power of processing data in real time at the source. That is about as real-time as it gets. Earlier this year, we announced an intent to acquire Juniper Networks. Deal is expected to close at the end of this calendar year or early next year. We expect to harness the combined capabilities of both companies to create an industry leader with a modern, more secure AI-driven networking portfolio. This will complement our overall business strategy to accelerate the deployment and adoption for both hybrid cloud and AI. And what that means? It means for enterprises, managing, simplifying complex network architectures and extend it into new areas such as private 5G for environments where you need persistent dense connectivity. And for service providers, we expect they will see improvements in automation and orchestration of virtualized network functions to scale their infrastructure, reduce costs and accelerate the speed of delivery for new services to customers. To train your AI model, you will require the right accelerated computing and storage infrastructure. HPE is the leader in accelerated computing from edge to exascale. And together with our partners, we will continue to engineer high-performing and sustainable AI systems. We continue to expand our HPE ProLiant and HPE Cray AI servers portfolio to support the latest silicon innovations that will demand direct liquid cooling. We will be time-to-market with our key partners. AI also require a modern approach to store and protect your data. Our new HPE Alletra Storage MP platform provides high performance for data-intensive workloads with highly efficient data reduction. HPE Alletra Storage allows you to scale up and down independently to optimize your individual workload requirements and now extends data replication to the public cloud with its software-defined storage capabilities. Building an intelligent data foundation will be the key to your data-driven business transformation. The latest release of our HPE GreenLake file storage is specifically designed to meet the demands of our structured data for AI as well as our recently announced HPE GreenLake block storage for AWS, a solution that helps you seamlessly manage block storage across your hybrid cloud environment. That is what makes HPE a leader in hybrid cloud storage and your trusted partner to advance AI. BMW is a great example of our customer that is taking full advantage of our unique innovation. The premium car company brings together the future of mobility, electrification and sustainability. Together, we embarked on a history-making collaboration that unified vehicle trace from the global BMW test fleet. This enables a series of initiatives, including automated mass data analysis, an essential part of the vehicle development process. Having a solid data foundation is a critical step to accelerate the deployment of AI. Let us take a ride into the future of mobility. [Presentation]

Isn't that amazing? What an incredible look. The driving experience that await all of us. But listen, we have not stopped driving innovation just on Earth. We are out in space, too. As an example of the most extreme of edge inferencing, a new chapter has unfolded with our HPE Spaceborne Program. As we sit here today, HPE Spaceborne is orbiting more than 250 miles above us, enabling astronauts to process data in real time as they collect it in the ISS, International Space Station. Why? Because they are conducting research in space. In 1969, HPE's technology helped support the Apollo mission land the first U.S. astronauts on the moon. I'm personally very proud that HPE will partner with Venturi Astrolab and its FLEX rover program on the upcoming Artemis mission to bring edge computing and AI capabilities to the moon. Working together, we are building a logistics network that will help move supplies and collect data so that what we can learn up in space will benefit us all down on Earth. A very special thank you goes to our partner, Kioxia, for joining us on the journey to the moon, so thank you very much. But this all leads us to the most important component, which, I believe, is our people. HPE has a dedicated team of experts, including our partner ecosystem, who are committed to supporting you throughout your AI and hybrid cloud journey. Expertise has always mattered. You need a partner who understands your business, who has the right technological solutions and who can help you create investment capacity so you can pursue your business objectives. Together, we are One HPE, building a bridge to the future with you. With our unique innovation and expertise, we make the impossible possible. HPE is your trusted partner in AI. We stand ready to help you make history, too. We are not just witnessing the future, we're shaping it with AI. For more than a decade, NVIDIA has been our visionary partner, who share our purpose and commitment to innovation. Over the years, HPE has partnered with NVIDIA to achieve amazing success with our customers and delivering best-in-class AI and supercomputing solutions. One great example of that is the newest Venado Supercomputer at the Los Alamos National Laboratory, which is the first U.S. supercomputer to feature the NVIDIA Grace Hopper GPUs. Another example is our recently signed $200 million deal to build a new supercomputer for one of Japan's largest research institutions, AIST, which we will build in collaboration with NVIDIA, using our HP Cray XD systems, featuring NVIDIA H200 Tensor Core GPUs. At the end of last year, we jointly announced 2 new innovations. First, we introduced our turnkey generative AI training solution built on NVIDIA Edge GX reference architecture designed to train and tune generative AI for large enterprises. And second, we announced a co-engineered preconfigured generative AI solutions specifically for enterprises to quickly fine-tune foundation models using private data that can be deployed anywhere. HPE and NVIDIA have a proven track record for delivering innovation. Today, we are taking our partnership further. It is with great excitement that we announce NVIDIA AI Computing by HPE to accelerate the generative AI industrial revolution. And here to tell us more, I am very pleased to welcome out a man and a friend who needs no introduction, the founder and CEO of NVIDIA, a driving force of innovation that's changing the world, Jensen Huang.

Did somebody say partner?

Yes. Look at this. Isn't it nice?

Hello, everybody.

Isn't it amazing?

Go, HPE.

Go, NVIDIA. Isn't this amazing, Jensen?

Yes, it's great. Nice living room. Did you say living room?

Nice living room, my personal living room, yes?

Okay. I like it.

Yes. Maybe we'll play something later.

Okay.

Something more interesting. But let's adjust to that.

So all day long, you've got all these people staring at you. That's a good way to live.

Yes. I couldn't believe. I mean, listen, we have people kind of stuck in the bridge -- at the bridge, right? And they were telling me, we need to wait another 5 minutes or so. I said, great, let's wait. We want the house full. Look, it's full. In fact, there were supposed to be not all people on the site because that kind of is not the best view. But boy, we have every seat pretty much taken. It's amazing.

Well, they're here to see you.

Yes.

Yes, you guys are doing a great job.

Yes, thank you. Thank you for being here. It means a lot to us. But honestly, it means a lot to them, our customers and partners. Listen, we just announced the NVIDIA Computing by HPE. And obviously, our focus is to capitalize this amazing opportunity, which is AI. But let's talk a little bit about what it means for enterprises. What is for them at the end of the day?

Well, when you take a step back, AI, generative AI, what's going on in the industry today hasn't happened in 60 years. This is the greatest fundamental computing platform transformation in 60 years, from general purpose computing to accelerated computing, from processing on CPUs to processing on CPUs plus GPUs, from instructions, engineered instructions to now large language models that are trained on data, from instruction-driven computing to now intention-driven computing. Every single layer of the computing stack has been transformed, as we exactly know very deeply well. And the type of applications that are now possible to write and develop are completely new, and the way you develop the application is completely new. And so every single layer of the computing stack is going through a transition. And that's why, from an industrial perspective, it's such a big deal. On the other hand, what's really amazing is a whole new industry has emerged. What used to be the IT industry, and we've seen the creation of a $3 trillion industry, which is really driven by the last Industrial Revolution, where we invented the idea as an industry of mass producing software. Software is invisible. You can copy as many versions as you like. And yet, we've created a $3 trillion industry on that idea. Now we have a new idea. And this new idea is quite extraordinary in the sense that what is being manufactured is still invisible, but it's embedded with intelligence. We call it tokens. And so what is really extraordinary is the IT industry and the work that we're doing together, on the one hand, can address enterprise computing, GPU clouds, sovereign AI, but now we also have this new industry that's emerged called AI factories, where we're producing intelligence in high volume. And so our industry has been transformed really, really profoundly, and this, we're at the beginning of that.

But Jensen, I mean, 2 years ago, we were not talking about this. But you and I had many conversations over the years, and you have this amazing conviction, there is a better way. And it's not just about technology, it's about transforming the business. And I think the power of this AI revolution, that's why we call it the generative AI industrial revolution, as you just talked about it, is transform the business, is transform society. So when you put yourself in their shoes as a customer, what do they need to think through to be successful as an enterprise? What are the key ingredients, if you need to think about it?

Well, theoretically, what AI needs are 3 components. Of course, the model, the revolutionary large language model and generative AI models. Second, computing to go process that, but very importantly, something that all of the enterprises in the world possess. It's -- and I heard you say it earlier, and it's completely true, their most important asset, their -- for companies, it's your important asset. For countries and different regions, it's your natural resource. You need these 3 things. You need model technology, you need data and you need compute. However, that's on a theoretical basis. What you need on a practical basis is you need a model stack, you need a data stack, and you need a computing stack. Each one of those stacks are incredibly complicated. And so the work that we've done together is really about productizing, and on a really massive and comprehensive way that our partnership is working together so that we could productize the model, we call them NVIDIA NIMs, and the model stack, the data stack, which comes from NVIDIA AI Enterprise. And the data processing, the vectorizing of your data, semantic embeddings, the retrieval and the deployment, we call that NVIDIA AI Enterprise, and the entire computing stack, which has all of the software that goes along with supercomputing technology. And one of the reasons why this is built in the clouds today because the stack is so complicated. Between our 2 companies, we've really turned these 3 stacks into deployable solutions that you can engage right away.

So the NVIDIA AI Computing by HPE offerings and services, one of the key elements, it will be available through our joint go-to-market strategy that really spans the sales, the channel. So here, you have 1,500 of the most important channel and partner ecosystems as well. And we will provide training, obviously. But also today, so you know, we're making an announcement that Deloitte, HCL, Infosys, TCS and Wipro are coming along our journey. And that's really significant because at the core of this, they also need to transform the business. And obviously, there is a variety of workloads. And so when I think about that, first is training and certification. So together with Jensen and his team, we are going to train the entire HPE sales force. We're going to show up in front of you as 1 company. And when I think about that, there's thousands and thousands of people, thousands and thousands of people just on the specialization side as well as on the presell side, which is a huge important role. And listen, when you engage HPE, you're going to be working with AI experts that have the skill to help you design, together with our partner ecosystem, deploy and operate these amazing generative AI solutions. Together, we are making significant investments to train you as well and the services that you need across the life cycle to turn this NVIDIA computing initiative a reality. So where you have skill gaps, we have the experts to help you ready to support. Second, we're also working with Jensen and his team to develop multilevels of comprehensive and optimized solutions to deliver that amazing innovation. And obviously, Jensen, you are on the forefront to keep announcing this amazing architecture. So obviously, we are now into the edge family, but the Grace Hopper, and now also Blackwell, and the other ones, Rubin and the like. So that's super important. But also, at the same time, we also need to think about managing the life cycle of this solution. And that's why HPE has unique differentiation with HPE Financial Services because, ultimately, as these systems come into life, we can find the right smarter IT life cycle solutions so that we can help you create the capacity to invest in these amazing innovation, but also make sure we find the right use when the asset comes out of your environment. So Jensen, this is amazing because I'm super excited about this joint go-to-market. But can you please elaborate these 3 elements we talk about, right? Because ultimately, this ends up on an integrated co-engineering solution we're going to announce today.

Yes, these 3 stacks are incredibly complicated. First of all, you're sitting on a mountain of really valuable data. You want to fine-tune your proprietary model. We have a large collection of public, proprietary and partnership models that we've created, and we take this entire complicated stack and we turn them into NVIDIA inference micro services, basically a model stack. This model stack is essentially an API. It's kind of like a chatbot, except we've integrated everything and made everything work, and it's now essentially a giant container of hundreds of dependencies, incredible amounts of technology, but easy for you to use. You download it, you could fine-tune it, guardrail it, you can deploy it. And that microservice is now one of the stacks. The second stack that you have is your data stack. This data stack, you want to vectorize, bring semantic meaning to it. And so we have technologies for helping you do that we call NeMo. You can then take this data stack and connect it to your model stack. This way, your model stack becomes an intelligent AI that can interact with, chat with, retrieve information from your company's private data. And then the third thing is, of course, the entire computing stack. Between us and HPE, we will turn you into a world-class AI cloud, a world-class AI company. And then, of course, those 3 ingredients are fundamental to AI, but ultimately, you also need the expertise, the domain expertise. And this is where this network of partners connecting with all of this technology with a giant go-to-market, we can now bring AI to the world's enterprises for the very first time. This is a massive partnership and a really comprehensive solution that we can take to the market. So it's a very big deal today.

And we appreciate the vote of trust and confidence in us because, for us, obviously, partnering with the best has always been one of our key objectives. But you talk about these 3 major elements, I call it. And while others have talked about integrating these AI solutions, at HPE, we wanted to create something truly unique with you. And so we did not just bring together the pieces, I always like to talk about widgets or doorknobs in the back, that's very complex for you. That's not how we work. If we want to accelerate this generative AI solution, we have to deliver you an experience. And so that's why, together with NVIDIA and their amazing portfolio of software and, obviously, silicon, the networking components, we are bringing all this together with the HPE systems, HPE services, the ability to manufacture and service this. I think people forget that. Because once we go to direct liquid cooling, and Jensen and I were talking early on, you need to have a water-cooled manufacturing capability. And HPE has one of the largest water-cooled manufacturing capabilities on the planet. On the planet. Why? Because we have to do it for supercomputers, right? And so now we bring all together, we democratize with all the things we talk about, and that's why today, it is a pleasure to introduce what we call the first-of-a-kind turnkey private cloud AI solution co-developed with NVIDIA. We call it the HP Private Cloud AI. And this solution provides the deepest integration. And this is very important. It is the deepest integration to date of NVIDIA AI Computing networking and software with HPE storage and, obviously, our servers and the HPE GreenLake cloud. Because, ultimately, you have to deliver that experience as a cloud experience, right? Unlike any other AI solution out there who are services-led, [ mid-tender ] in front of the customer, bring it together, we deliver this as 1 integrated product and is ready to run out of the box because that's the key. When I say turnkey, I mean, the simplest experience today for deploying and operating the AI NVIDIA software stack in the industry. So how this work, Jensen, because you and I went through it, right? Is it plug it in, a few clicks?

Three clicks, I understand.

Three clicks?

Three clicks. All it takes is 3 clicks.

Stop there, 3 clicks?

Well, the thing that's really amazing is you say it works right out of the box. Inside that box are CPUs, NVIDIA Grace CPUs, NVIDIA GPUs, networking, NVLink Switches, either InfiniBand or Ethernet Spectrum AI switches. And all of that technology with all of that software all integrated, number one. Number two, we've made it so that you could have small, medium, large...

Extra large.

Yes, extra large. And as you know, the more you buy, the more you save. And so everybody...

Buy more. In fact, you need a few of them.

Yes. The more you buy, the more you save. And so you want the extra large version, but you can start anywhere. And this is really the big deal that we've integrated everything into a really easy-to-use private cloud AI solution. And all of the complexity that goes into all the technology has been provided to you, hidden from you, and it's helping you operate it over time. Remember, you're not buying a computer that you're using. You're buying a service and you can own a service that you can operate. And so that's a very, very big deal.

Yes. I'm excited because, obviously, he is million tons better engineer as I am, but when I think about technology, it's always about the experience, right? So that's why the HP Private Cloud includes comprehensive, multilayer from both HPE and NVIDIA, right, because we have to deliver this through a cloud experience. And also, the most secure sensitive data models, which includes, by the way, built-in AI, data compliance and explainability, that's very important. You want to know for where these models are coming from. Automated AI pipelines with complete lineage, traceability and auditability for the entire AI life cycle. This is one of the things that enterprises, they're worried how this thing works, where the models are coming from, how to track the data. A single dashboard, think about that, a single dashboard to monitor all your AI applications and services. By the way, it supports multi-tenant collaboration with unified role-based access control to isolate data and purchase across different teams. It is, as Jensen say, all built in. And at the launch, as we just kind of hinted to you, HPE Private Cloud AI will be available in 4 configurations. We are in the United States so we need to have 4, not small, medium. We need to go a little bit large, but 4: small, medium, large and extra large. And you can see here, each one of them is flexible, is modular, has the ability to expand our capacity over time, but maintain that consistent cloud-managed experience with HPE GreenLake Cloud. Many of you are already HPE GreenLake customers here, whether they're using it for deploying an access point or a storage array or a server or a switch. Now you can do the same with this private cloud solution. Listen, some customers start small, maybe to try, but also do some inferencing. But also, as the use cases multiply, you need other things like RAG and to fine-tune into 1 solution. I think we are going to offer the most comprehensive solution on the planet because, also, we can offer as a self-managed or as managed services as a part of HPE GreenLake. And at the same time, you maintain the consistent operating model, which, by the way, is ready to adopt whatever they come up with, which, I know, you have a lot coming up with, which means he's ready to adopt the latest technologies that Jensen and the team are coming through. So this is, in my mind, is a big deal. So when you think about that, Jensen, and think about the next generation, which, obviously, will support Grace Hopper 200 as well as the network, what do you think the enterprises will be looking next once they have this simplicity? It's all about understanding the use cases and getting the factory mold to deploy. I think that's what we're trying to do here.

Yes. It's a really big deal for enterprise today. Look, the fact of the matter is we're sitting on a mountain of data. All of us are sitting on a mountain of data. We've been collecting it in our businesses for a long time. But until now, we really haven't had the ability to refine that data, discover insight and codify it automatically into our company's digital experience, our digital intelligence. Every company is going to be an intelligence manufacturer. Every company is built fundamentally on domain-specific intelligence. For the very first time, we can now digitize that intelligence and turn it into our AI, the corporate AI. The thing that's really profound here is that you could start anywhere, small, medium, large, extra large. They're architecturally completely compatible. And so if you change your mind and you want to do more, everything that you've built yesterday is going to run seamlessly across the larger systems. And when we go to the next-generation system, everything will be CUDA-compatible. This is one of the great disciplines of our company by putting everything on top of CUDA this revolutionary invention that we made 20 years ago, and having the discipline to maintain CUDA compatibility over all these years, some 15 generations now, we can make it possible for all of our customers to invest at any point, at any scale and carry that investment forward. Now once you do so, remember what AI is. AI is -- and you mentioned life cycle. It's a life cycle that lives forever. The thing that we are looking to do in all of our companies is to turn our corporate intelligence into digital intelligence. And once we do that, we connect our data and our AI flywheel so that we collect more data, harvest more insight, create better intelligence, which allows us to provide better services or create -- be more productive inside the company, run faster, do things more efficiently, do it at a larger scale and, very importantly, create new products. And so all of that is possible because of the things that we're doing today.

The net of all of this to me is that you've got to embrace this technology. We understand people have been a little bit wait-and-see. It wasn't easy for them to embrace it. But with today's announcement, we're making it really easy for you. It's just a rack or maybe 2 racks that you can deploy with everything you need fully integrated. And to me, that's super exciting because we give you the tools to accelerate your journey to transform your business. But today, it's not just that. We talked about NVIDIA Computing by HPE. We talked about the announcement of HPE Private Cloud AI. But also, today, HPE has committed that we will beat time-to-market with NVIDIA Blackwell and Rubin GPU architectures, in addition to all the software that Jensen and team will continue to bring to market. So as you adopt this 1 operating model inside GreenLake, it comes with it, right? Very simple. And we'll be the leading integrator of the most advanced AI systems at any scale in a sustainable way, which is something we both care about. Because I argue, the sooner you adopt this accelerated computing, the more efficient and more sustainable you will because you eliminate all the waste.

Liquid cooling results in higher performance, but also lower infrastructure cost. The overhead of all of the liquid to air, air to liquid, all of that is now gone, liquid to liquid. And so the future of liquid cooling is going to result in everything from better performance, lower infrastructure costs and lower operating costs.

And HPE is uniquely positioned to deliver direct liquid cooling.

Nobody has plumbed more liquid than Antonio.

There you go. And by the way, you need to come see our factory so you see how good we are on that one. But listen, HPE delivers 100% liquid cooling, including the CPU and the GPUs and the fabric. I don't think people understand that the fabric today is also liquid cooled. They think about a switch there with a bunch of Ethernet ports. Actually, it's not. We also run direct liquid cooling to the fabric. And listen, we hold numerous, numerous, hundreds of patents in this particular part of the portfolio. And listen, we have the expertise to build these large systems at scale, as I show early on. But also, we are really proud to bring this innovation for you, together with Jensen. And I think our combined portfolio can achieve amazing results. And by the way, we do it in a circular economy because we can actually capture 100% of the heat that gets wasted, and we have some amazing use cases. We take actually the heat generated in some areas, and we either power greenhouses in the winter, also power heating systems in buildings so they don't waste other things. So listen, Jensen, know we can spend here the whole day. But to wrap us up, what is the key takeaway you want the audience here with an amazing living room of this partnership?

Well, one, the era of generative AI is here. You must engage the single most consequential technology in history. You must translate and codify your company's intelligence into digital intelligence. And you must find a way, as quickly as possible, to turn that flywheel of your company's experience into your company's AI as quickly as possible. That's number one. Number two, our partnership with HPE is about simplifying all of this incredibly complicated technology in 3 stacks, the data stack, the model stack, the computing stack, turn it into this private cloud so that you could operate it or you could get the benefit from it with all of that complexity being extracted from you so that we could help you apply all those great technology for you -- in service of you. And then lastly, this giant go-to-market.

Exactly. People. People matter.

This giant go-to-market, yes, this network of ecosystem, of partners and service providers, we're going to, because of this, make it possible for the first time to bring generative AI to every single company in the world. Thank you.

Now, Jensen and I are going to be on the show floor later, not demonstrating because we don't have time, but going through the demos, which are amazing. We have an amazing show floor. Please go there. But listen, what Jensen and I do is in service to you, our customers. When we connect the power of our joint innovation, the expertise and the greatness of our partnership with NVIDIA to your enterprise, intelligence has no limits, right? Intelligence. So thank you, Jensen, for inspiring us. It's great to have him here, an amazing innovator, a force for change in the world. And for me, it's a pleasure and honor to always to learn from you. Thank you very much.

Thank you, Antonio. Go, HPE.

Go, NVIDIA.

All right, guys. We're on your team.

So again, at the center of what we have developed with NVIDIA is the most simple experience to advance AI. That's what it's all about for you. We deliver the experience through our HPE GreenLake Cloud, and it sets HPE apart from all others. HPE GreenLake is the leader in hybrid cloud. One of the crown jewels of HPE GreenLake is our acquisition of OpsRamp, which we have now fully integrated into our cloud platform. It provides full observability for multi-vendor environments, including some of the other vendors like Cisco, Dell, NetApp, Pure and many, many others, as well as your public cloud instances. And today, we announce 3 major updates from OpsRamp. The first one, OpsRamp now supports full-stack AI infrastructure to workload observability. So Jensen and I just talked about. That means you can observe now this private cloud AI solution, which includes NVIDIA GPUs, AI clusters, NVIDIA DGX systems, NVIDIA Mellanox, InfiniBand and Spectrum-X Internet switches. Second, we introduced an AIOps copilot, an AI copilot feature. It is actually a natural language interface that enables enterprises to identify, predict, and solve IT problems more quickly by converting machine learning data into actionable insights. The operations AI assistant combines the observability signal from specific AI models developed by OpsRamp, with a generative AI conversational assistant to digest large data sets and provide insights in real time. And finally, to provide more security, observability across your workloads, we announced a new integration with Crowdstrike APIs so customers can see a unified service map view of endpoint securities across the entire hybrid IT infrastructure and applications. As you can see, we are not just keeping pace with the future of AI operations, we are leading the industry. And then there is the data we just talked with Jensen about that, right? So you need to protect the data. And so our acquisition of Zerto provides edge-to-cloud disaster recovery and cyber resilience. Another aspect of our HPE GreenLake differentiation is also our cloud-native approach, providing you with flexibility and choice of run-time environments, including bare-metal containers and virtualization. And today, we announced a new HPE-developed virtualization capability for our private cloud portfolio. What is it? It combines an industry standard KVM hypervisor that has been hardened for enterprise with HPE's innovative cluster orchestration software to support the high performance and availability requirements of enterprise workloads. And it complements the HPE Ezmeral software container servers and extends to cloud-native and AI workloads. The adoption of HPE GreenLake is accelerating. Today, HPE GreenLake connects more than 4 million network devices and a lot of storage and compute systems, supporting more than 34,000 unique customer organizations to transform their business. Just look at what HPE GreenLake is doing for the Harry Reed International Airport right here in Las Vegas as you came through. It is powering a future-ready, more secure infrastructure to meet the evolving Wi-Fi needs of the millions and millions of travelers who must stay connected. And when it comes to world-class sports teams and sports venues, HPE GreenLake is in a league of its own. We are fanatical about optimizing sports experiences and deepening connections between the fans and the teams they love. In Rome, the 2023 Ryder Cup leveraged the power of HPE GreenLake to deliver an exception fan experience to more than 250,000 spectators over that weekend. The world-class goal, of course, was transform into a smart city, a super smart city, bringing unrivaled secure connectivity to fans so they can follow the dramatic moments at every hall, for sure, for the United States players. It didn't work well. But for Tottenham Hotspur, HPE GreenLake and our networking infrastructure enables the Premier League Cloud to deliver an amazing differentiated fan experience and generate, at the same time, greater revenue by using data. Technology is integral to growth, innovation and global expansion of sports organization as well. And our demonstrated leadership in this space caught the attention of a very iconic football team. Just last week, I announced that HPE signed a partnership agreement to become the new official Edge-to-Cloud partner of Espai Barça. HPE will provide state-of-the-art technology for Espai Barça. The teams completed renovated stadium, which is in progress, which is called the Spotify Camp Nou, and the new Espai Barça mega complex, which is the largest and most innovative sports and entertainment space in any European city, a truly spectacular venue. Powering the reimagined fan experience is HPE GreenLake Cloud, HPE Aruba Networking and our hybrid cloud solutions. And in the future, they will be also deploying AI. These are cutting edge innovation that will take the match operations during the day to a whole new level and create an immersive experience that will leave visitors in awe. Let us take a look at how we supercharge the fan experience of many with HPE GreenLake. Let's play the video. [Presentation]

So today, you heard many amazing stories for customer success like this, but HPE doesn't do it alone. Our industry-leading partner ecosystem stand with us, and they built their business strategies around our innovation. Many of those partners have generously sponsored this week, including Emerald-level sponsors: NVIDIA, Intel, Kioxia and Microsoft. I would also like to thank our Platinum sponsors: AMD, AWS, Red Hat, Samsung Semiconductor and SK Hynix. We appreciate each of you for your partnership in making HPE Discover a reality. I also want to thank my team. Listen, being here at this amazing venue takes an enormous amount of planning. And obviously, I'm incredibly privileged to be here in front of you. And special thanks to Jim Jackson, our Chief Marketing Officer, who architected this. It is incredible. And all the teams, from the communications team to the business units, to put me in this position to speak to you. But listen, partnerships and connections are the lifeblood of all endeavors. HPE ignites and propels your interconnected journey. In this week ahead, you will find everything you're looking for to make that journey. Leverage your time here and talk with our experts. Bring us your biggest, boldest ambition, and we will bring it to life. As we stand on the cusp of an AI revolution, our ambition knows no bounds. It drives us to reach for the stars, to solve the unsolvable and to transform ideas into reality. But ambition alone is not enough. It requires a catalyst, a force to amplify and accelerate it forward. Technology is that catalyst, serving as the bridge between aspiration and achievement. For more than 80 years, HPE has been at the forefront of innovation. And today, our more than 60,000 team members around the world are laser focused on helping you realize your biggest dreams. Let us embrace these amazing technologies, not just as a tool, but as a force that unlocks our human ambition. Thank you very much. Enjoy the rest of the week. It has been an amazing privilege to be with you today.

Okay. Thank you, everyone. So thank you for joining us for the Investor Relations Summit at HPE Discover 2024. I know most of you, but I'm Shannon Cross, HPE's Chief Strategy Officer. And I'm really pleased that you've joined us in person or tuned in via the webcast for HPE Discover. It's a unique opportunity to learn more about our strategy, our innovation and how we're capitalizing on market opportunities. Before we jump in, I'd like to thank Jeff Kvaal for his time as Head of IR and wish him good luck as he leaves the company to pursue another opportunity. As always, inquiries can be directed to any member of our team and our contact details can be found on our IR website. Now we hope you enjoyed the keynote. It was amazing and the Executive Summit with Antonio and Jensen, which was also amazing. And we look forward to diving deeper into our announcements in this session. In addition to Antonio, we have the pleasure of welcoming 3 additional executives to our management -- well, 2 of them -- he's coming. We'll see. Things are running a little over, but Phil should be here. On the stage with me, we have Fidelma Russo, who is our CTO and leader of our Hybrid Cloud business; Neil MacDonald, who runs our server business and, of course, Antonio and Phil Mottram, who runs the Intelligent Edge business may or may not make it here as he's...

He's doing the keynote right now.

So I got to do this quickly. Please do let me remind you that HPE Discover is about our strategy and our innovation. We will not be updating financials, including guidance today. I will start with the disclosures. This event may include forward-looking statements involving risks, uncertainties, estimates and assumptions and if the risk or uncertainties ever materialize or the estimates or assumptions prove incorrect, our results may differ. Perhaps materially from those expressed or implied by such forward-looking statements. HPE assumes no obligation to update such statements. Please find more information regarding forward-looking statements on our website at investors.hpe.com.

With that, I'll kick it off with a question for each of our executives and we'll open it up to audience Q&A. So I'm going to start with Antonio. So I mean, we just saw you and Jensen. We obviously saw the keynote, I guess, maybe if you could briefly just point to what are the key announcements and key attributes that we have that will allow us to win with model builders, sovereigns and enterprise in AI?

Well, first of all, good afternoon, and thank you for joining us in person. I know many of you are on the East Coast and you have to travel quite a bit, so much appreciated. I feel being together here, actually, you can touch, visualize everything because sometimes we talk about after earnings, we do follow-up calls on a regular basis, but you can't really visualize everything we do in a vast portfolio. To give a sense, the show floor there is 7 acres, and it's really covering every aspect of what we do. But today, it was all about AI, obviously, because that's the topic of the moment. But without forgetting that we still bring innovation across the rest of the portfolio. And so when we think about the announcement today, obviously, with Jensen, we announced the NVIDIA Computing by HPE. Underneath that is the work that Fidelma and Neil have been doing around this fully integrated solution specifically targeting the enterprise. Because when you think about the customer segmentation, I think about 3 different unique customers, you talk about hyperscalers, model builders, those are very few. You can count it sometimes with a finger of 2 hands, and they demand enormous amount to accelerate the compute. You're talking about maybe hundreds of thousands of GPUs. On the other spectrum, we have the hundreds of thousands, if not millions of enterprises, they may need only hundreds of GPUs. And in fact, if you go look at the configs that Fidelma has worked as a part of the private cloud, and it's a small, medium, large and extra large. And they are very much tuned to specific type of use cases, whether it's inferencing or fine-tuning or a small training because a lot of customers are going to do the proprietary language model, but they're small, not these large language models in the trillions of parameters. So that was the core, but the core of that was obviously the experience, easy to buy, easy to deploy, easy to manage. And ultimately, as we were doing the tool with Fidelma, she reminded me one thing, it's going to be 3 clicks, 24 seconds to deploy an AI application. And for enterprises that were in the room with you early on, that's the name of the game. They don't have the time, they don't have the expertise. They are a little bit kind of wait and see, but they all know they cannot left behind. So that was the core. In addition to the fact that together with NVIDIA, we already sell, as we said in the earnings call to large hyperscalers, including Microsoft, obviously, that extend their Azure cloud to our HPE infrastructure, the cloud colo. We actually deploy and run all of that for them. And then obviously, service providers we like. We talked about Scaleway as an example of it, but there are many, many others that we talked about over the last few quarters. And then sovereign AI. And sovereign AI is going to show up more like a supercomputer, honestly. Because sovereign AI to me is not different than a [ Fortis ] system or the U.K. Bristol National Laboratory, where they use it for public sector use cases, perhaps weather forecasting is a great example, but they will open it up for private sector as Frontier has done with GE and many other companies. And HPE has all the technologies from training to tuning to influencing from hyperscaler to enterprise and sovereign clouds to do it. And one of the things that Jensen was super excited is, obviously, for him, the transition to Blackwell, Rubin and Vera requires 100% directly liquid cooling and Neil, I'm sure, will talk about that what our differentiation is. But we didn't stop there. We talked about enhancement to our GreenLake platform with a Copilot for OpsRamp. We now integrated Zerto, we integrated OpsRamp. We integrated the CrowdStrike, APIs through the API, so we can manage endpoint security. And so we keep advancing. In addition to that, if you really pay attention in the last 6 weeks, and we did this purposely, we've made a series of amazing announcement in our networking business, including WiFi 7, private 5G, security, Copilots for Aruba Central, which is fully integrated inside HPE GreenLake. So a lot to go see. I know there is a private tour resorts eventually. You will be blown away by the variety and the broad -- how broader portfolio is, but we always integrated to the same experience, which is HPE GreenLake, how you pay for it as a customer is your choice. You can pay as a CapEx plus a subscription or you can pay as a full consumption in an IS type of approach.

Great. Thank you. So Neil, maybe you could dig a little bit more into the key capabilities and technologies that we have within our AI systems business and why that should allow us to win in the future?

So as we shared in our earnings announcement, we're participating at scale in the model builder segment of the AI market already. And there, we leveraged our history of running some of the largest, most complex infrastructures in the planet from our leadership scale high-performance computing business. We bring to that the ability to host these systems on AI for per customers and to collaborate on other business model be on straight CapEx by leveraging OpEx and as-a-service models. When you think about the sovereign builds that are becoming increasingly prevalent around the world, such as the U.K. I deal with, that Antonio just mentioned, those are very much an adjacency for us, right beside our traditional leadership scale HPC business in supporting national research institutions around the world, where we build on the momentum that we've had with the top 2 exascale systems in the world and 7 of the top 10 of the Green 500 supercomputing list. So for us, that's a very natural adjacency because what's happening is that these national research organizations are pivoting their focus to embrace AI as an integral part of the scientific research agenda, and those are the customers that we have been serving for decades with our Cray business. When you think about the enterprise, it's all about being time to market with the right technologies on the one hand, about democratizing that liquid cooling expertise because very, very quickly here, we're going to be dealing with accelerators, which are so energy dense that the only viable approach to deploying them is to go to 100% liquid cooling, no fans. And you'll see on the show floor, the Venado system from Los Alamos National Labs as one example, very similar architecture to what we've deployed in sovereign AI deployments around the world, which has about 100% liquid cooling and enables us to deliver not just the technologies of today, but the accelerators that are coming in the coming quarters and years with incredibly high energy densities. But really, the enterprise also needs systems that can fit into an enterprise environment. So we've announced today, for example, our HPE ProLiant Gen12, our very first Gen12 system, the DL384 that is based on NVIDIA's Grace Hopper, GH 200 platform. That enables enterprises to adopt that technology with the management tools and the format that they are expecting from a traditional X86 server, but no longer to be shackled to that architecture, being able to embrace all of the advantages of accelerated computing that you just heard about from Antonio and Jensen. But really what the enterprise needs is simplicity. It's not about just an accelerator. It's not about just a server. It's about the fabric, it's about the data. It's about the data management tools, the models, the integration into the model workflows, the guard railing, the governance and so on. And so with all of that, really the most important thing that we're showing this week about the enterprise opportunity is what Fidelma is building on top of all of those elements that I've described in the core server business.

And that was a perfect segue to my question for Fidelma. So can you talk a bit about the private cloud AI and how it's differentiated. We did a demo the other day. We have an AI council that we're helping within -- to run across the company. And during the demo, it was really interesting to me to see as sort of a non-techie person, how -- don't tell Antonio that, how simple it was. And then I've heard from others that really, there's nobody out there doing this from a competitive perspective. So maybe you can provide a little bit more color?

Yes. So thanks, Shannon, and I think everybody has heard a lot this morning, and I would encourage people to go to the show floor, you'll see it. There's a demo on the show floor, there'll be a demo tomorrow. But really, what we've done is really simplify for people. You don't have to be an IT expert and you don't have to be an AI expert, okay? So that was the premise. And so what it does is take away if you're embarking on your GenAI journey, you want to deploy a chat bot, you want to deploy within your organization, for instance, a Copilot, for instance, for cogeneration, you don't have to think about what compute, what storage, what networking, kind of, what operating system, what could -- I mean, if you think about the list of questions you have to ask yourself before you even think about how you get it all up and running, before you think about is it actually working, you will be there for months. And then you have a whole bunch of consultants that come in and a whole bunch of service providers. And then you have to think about is it going to be successful? By the end of this thing. That's why back in 2017, 2018, there were a lot of failed machine learning AI projects. And so what we did was we took a step back and along with NVIDIA, we said, how do we make sure that we take that burden. We size everything, we look at it and it's ready to run, and so that's what we've done with private cloud for AI. And so you come in, you size -- we have sized it. We look at the use case and you put it into our configurator and you come out with either a small, medium, large or extra large, and it's up and running with 3 clicks and 24 seconds. And so -- and of course, we have services that go around that, but they are really on top of the basic system. And so that's what the value prop is versus what you hear in other instances where you have the different pieces, but now you get to put them together. And so they're very, very different and you get a different outcome depending on how you put them together. So this is a guaranteed outcome. And so that's really what we've done with private cloud for AI.

Great. Maybe I know you want to open up for questions, but I know Fidelma and Neil eventually need to go up to do their keynotes. But one thing, Fidelma, you want to touch maybe is also the other part of the hybrid cloud announcement, which is the virtually...

Yes, virtualization piece. And one of the pieces I did want to talk about in private cloud AI, before I get to that in just in seconds. One of the really critical pieces in AI, and you heard Jensen talk about it as, in enterprises, there's a lot of data in people's enterprises. And with AI, it's data hungry. The benefit of AI is how do I get insights out of my data. And you have to be able to access your data, that data comes in lots of formats, structured, unstructured. Within private cloud for AI, based on our Data Fabric technology, we have integrated a data lake house. So within private cloud for AI, you are able to access easily, you can connect to all of your data across your enterprise. And so when you buy a private cloud for AI, you have a very simple way of accessing any of your data within your enterprise. And so that is critical to having a very, very effective way, and it is unique in the marketplace. The second announcement that we made is that we offer choice, and we always have within our private cloud portfolio of different virtualization options in the marketplace. But we have talked to a lot of customers over the last number of months, and they have asked us HPE to provide a virtualization capability, which we announced today within our private cloud portfolio. And so in the second half of the year, we are going to provide based on an open source, a hardened open source hypervisor, KVM, that we are going to provide this as a capability within our private cloud portfolio. And so we have added some cluster technology that we have within the company, and that will be available in the second half of the year. So we're very excited about this.

Great. And we have Phil Mottram here. And I have a question for him and then we'll open it up to the audience. So Phil, since the keynote was all AI. Can you take us to networking and tell us a bit about what the highlights were from the atmosphere?

Yes, sure. Yes. So yes, sorry, I was late. I was just finishing my keynote to the Atmosphere conference where we got about 4,000 or 5,000 attendees every year. And actually, the feedback, by the way, for them attended because this is a first, obviously, having Atmosphere as part of Discover, but the feedback has been super positive. We've been talking about a few things, Shannon, on keynote. One was security, and this is where we acquired a company called Axis Security about a year ago. And what we're betting on here is the convergence of networking and security going forward. So Gartner estimates that 3 years from now, 60% of enterprises will want to buy network and security from a single vendor using a model called SASE. And so we acquired Axis Security about a year ago. And actually, in the last 12 months, we've added 300 new companies onto that platform. So we talked about that. We then talked about private 5G. Again, that was a result of an acquisition that we made about a year ago of a company called Athonet. We don't see private 5G replacing WiFi, but we do see it living in parallel to WiFi for some use cases. So think about use cases with very large outdoor spaces, and also for very high-speed moving environment. So think about warehouses, ports, distribution facilities, defense organizations. There's a lot of interest in private 5G. And what we talked about on the stage today was the fact that we're making it as easy for customers to supply as WiFi. So it's part of Aruba Central on the same contract, same support model, et cetera, et cetera. So the takeaway is trying to make private 5G easy for enterprises to consume. And then last but not least, we didn't spend too much time talking about it today, but we've obviously had a big push to expand what we do in the data center space and 650 Group when they produce their market share data in Q4 of last calendar year highlighted us as the fastest-growing data center networking provider in the market, albeit off a relatively small base versus some of our competitors, but still making inroads into the data center market. So yes, that's what we've been talking about.

Okay. Great. Well, we're going to open it up now to questions in the room. When we come to you with a microphone, please introduce yourselves so everybody on the webcast can hear your name and your firm. And I also can't really see very well. So -- yes, let's go and start with Michael over here and then. The lights are very bright.

Mike Ng from Goldman Sachs. I was just wondering if you could talk a little bit about the type of enterprise that might want to access AI services with HPE, private cloud for AI versus buying and owning physical AI infrastructure for on-prem. And then with HPE private cloud for AI, delivering services and more of an IaaS approach, does that impact your ability to sell into AI CSPs that might view you as a competitor?

Fidelma, you want to take that?

Yes. I think -- Let's -- so when we sell private cloud for AI, what you purchase and you can purchase it multiple ways. You will end up with infrastructure on-prem, okay? So you can end up with infrastructure on-prem or you can choose to have it as a colo like an Equinix or a Digital Realty or your colo of your choice. You can buy it on CapEx, okay? Or you can buy it, purchase it through -- on a consumption contract. So it isn't -- we're not going to be in competition with anybody, with the kind of -- with the MSPs or anything like that. So I think that's really the answer to your question.

I think the way to think about it when the customer has decided that they need this infrastructure on-prem, whether it's their own premises or a colo. Ultimately, we get them this integrated solution, and they can pay the way they want it. On a GPU per hour basis, by the way, it's not different we do today as a part of GreenLake when we sell a server or a storage on a per gigabyte or a number of cores in VM or PCB today. In fact, our private cloud stack is built on the private cloud business addition that Fidelma has been building for a number of quarters now, which is one of the fastest growing products. And now we are just AI piece of it. But the AI is inclusive for all the tools and the software we talked about it. But then you have also a private cloud for virtualization or containerization of some cases, that will optimized. So it's an extension of the portfolio, target specifically for AI.

Okay. Great. We move over to Tim.

Yes. The way to that -- I think Shell, we need to sit there because there is all the questions here, on this side. I think.

Tim Long, Barclays here. I wanted to ask a 2-parter on kind of the AI server business. First of all, if you could talk a little bit about differentiation. A lot of the activity for companies like HPE, the enterprise stuff seems like it's more on the come, a lot on the sovereign AI, some really large enterprises. These cloud AI companies. So in a world where a lot of the componentry and technology is coming from NVIDIA. Can you talk a little bit about differentiation within that -- those kind of customer segments. And then related to that, you also have a lot of ODMs and EMS companies as well competing in that area. Can you talk a little bit about kind of value chain and profitability and where -- how HPE sees that developing is probably much different than the Cray's supercomputer HPC of the past. So if you can maybe touch on both of those related?

You want to start, Neil?

Yes, I'll start. So if you think about it, there's multiple layers that you can participate in, in serving that community, Tim. The first, obviously, is in piece part infrastructure. But beyond that, these organizations need more. They need the integration and delivery of essentially warehouse scale computing facilities, which also brings in fabric, it brings in the cooling, it brings in the delivery, the deployment, management services around that, potentially hosting that on their behalf on an as-a-service basis in a facility that we run. And we participate all the way along that spectrum at various scales with various customers in that space. So we have a very diverse set of models and participation that give us opportunities to add value beyond core componentry. And as you think about the sovereign space or you think about the enterprise space, there's a need for much more than that in all the areas that Fidelma talked about in the context of private cloud AI with all of the other value that needs to be added around it.

I think, Tim, I look at this way. First of all, we have a great portfolio of patents that obviously we're using today. Our leadership class system give us the right to play and win in this generative AI transition. Second is that you need to build the systems at scale, and it's very hard to do. Jensen was blown away when he walk briefly to the floor because he said, boy, look at this DC power, all these hoses, jokesly he says, the future could be the [ plumbing ] but it's true. And when you think about building the systems, you need very large water cool factories. Hewlett Packard Enterprise has one of the largest water cool factories, actually 2 now, 1 here in the United States, which obviously serves the sovereign requirements of the United States, as you can imagine. But also there, we are building these generative AI systems that we are deploying in the data centers that Neil talked about it. And in fact, quite a significant percentage of those systems get sold on a GPU per hour basis, which includes the services component. And that's why we started disclosing the last quarter announcements with Marie, how much of the services pull-through start coming through because that's a unique expertise, which is the last piece of this, right? You need to have the running expertise. I remember deploying with Neil and Dam Trish, one of the largest exascale, boy, you have to learn chemistry. It's like, I love aquariums, right? I always make that analogies. When you started the new aquarium, it's very complex. You have to build the bacteria. Well, this is not different. You have to build the bacteria on thousands and thousands and thousands of nodes, so the system can find the right balance. And Jensen walking by and said, " Antonio, what is the right temperature you guys optimize the FLOPS " and I say, well, depending on what you're doing, maybe it's 89-degree Fahrenheit or what I mean is Celsius, so I translated for him. I think he's still on Celsius, I know why. But I -- actually where I come from. But it was -- that level of expertise is unique. And let's remind ourselves, you're talking about potential in a very large load on thousands of GPUs and they all have to work consistently. Our software also plays a big role because how we do checkpoint also really matters because you either want to start the model by new and it stops and you start the game and it stops. That's a waste of time and money and energy. We now have to run the model from beginning to the end. In fact, it's one of the critical metrics, right? So we have all that expertise. Now we are packaging all other expertise for enterprise in the thing that Fidelma just talked about. It's one thing. But for these sovereign clouds and service providers, you need a level of capability. And I think it's very special. None of the other vendors. I mean, if you think about some of the vendors, you sell servers. We sell a solution that includes servers and includes also our storage capabilities.

Great. Ashley, do you want to talk -- how about Wamsi real quick and then we'll go.

We have time. We're going to come back all the way here.

Wamsi from Bank of America. Thanks for the presentation. Really nice keynote and really nice fireside informative for all of us. I guess, when we step back and think about AI, there's a lot of concern that in the industry broadly about profitability. And if there is this race to the bottom, clearly, you've shown a lot of IP, your liquid cooling expertise. So how do you think about profitability of the industry? And if I could just ask about and for HPE, and just on the sovereign point, I heard you say about this being more supercomputer like, why not sort of more broad than that in the context of sovereign? And what do you think the TAM is for sovereign?

When I think about sovereign, right, I think about the type of use cases that run there and the access to these systems require a little bit more work. They are becoming service providers themselves. The software layer around it, it's a little bit different than just open end GPUs to people. And so that's why it's a little bit from a system architecture, it's obviously more looking like a generative AI type of thing going forward. But from a -- but sometimes it's also simulation. So you have to find the right hybrid approach. And a lot of what happens in sovereign AI was there will be people doing large language model. For example, there is one large supercomputer we deployed for the European Union in Finland and they already developed a large language model for Finland, the language, right? But also, they are using also for simulation and like weather forecasting, but they also injecting generative AI to speed up the model. So you have to find architecturally is not just a bunch of GPUs in a system. And if you look at Venado is a great example, right? Yes, it's the Grace Hopper, the latest. But the way the team architected is also to address all the things because it goes beyond generative AI to biology and all the things you need to do. From a profitability perspective, I think we demonstrated a significant discipline. I mean you know that we have booked so far cumulative $4.6 billion. We have a pipeline that's multiples and multiples of the current backlog. And we stay in that 11%, 13% for the server business and partly is because obviously, we are also including these additional elements, which is not just the server, right? And that's the key here. And then as more of these as-a-service become reality, actually, that's sticker because ultimately, we sell a mix that software and services becomes a bigger portion than just infrastructure, right? And that's why we are -- with Marie and the team, we have incredible discipline. If we have a path for profitability over the life cycle deal, we take it. But so far, it has been very good, I will say. And so we don't chase just volume for the sake of volume.

Aaron Rakers at Wells Fargo. I wanted the one point of clarification there. So when you sell the AI platform, the solution suite, are you going to piece out other buckets within the other categories or segmentation? Or is it all in that server category when we're thinking about this margin profile that you're talking about? And then my question actually is on the fabric side. I'm curious, you talk about Quantum InfiniBand with NVIDIA. You've talked about SpectrumX on the Ethernet side. HPE's had a slingshot capability internally. You also are acquiring Juniper. So I'm curious of how the lines and I think you've even talked about data center networking more and more over the past few years. So how do you see those lines of delineation playing out between doing what NVIDIA is doing full stack with their networking versus what you guys do proprietarily?

So there's multiple questions in that question, and we'll ask eventually Neil to answer the NVIDIA within the stack itself with Fidelma and then I will take more of the data center view of this. But fundamentally, maybe start with that part. Over time, these architectures are going to converge, right? They have to. Because today, if you are within the rack itself, as we said, we want the simplicity of the experience. That's why you have the SpectrumX with MB links, the GPU and the CPUs that now have with NVIDIA and then obviously has the storage piece of this. Let's not forget about the storage component that Fidelma has and then all the software. All of that goes to the hybrid cloud segment that we report, okay? That's the enterprise. Everything that Neil sells today for the hyperscalers, it goes all to his segment reporting because it's a server with additional services. But when it comes a private cloud stack, it's part of the hybrid cloud integration, right? Now it has a server in it, it has storage, it has all these things and has a lot of software as well to it. Now the Juniper acquisition, as I said in my keynote is on track to close by the end of this calendar year, beginning of calendar '25. We don't see so far -- we are very -- any issues. We are very well underway with our regulators. Some countries already have provided some approvals already. But obviously, we need to wait for the United States, European Union, and that's going fairly well, I will say. But nothing gives us the pause to think something is going to go wrong, not at all. So that's on that one. But -- and I think the convergence, right? Now obviously, we are arm's length, we cannot engage Juniper. That's the process. But I envision a convergence of the data center networking and the stack over time and integrate it with NVIDIA aspect because today, they have almost like 2 different control planes. And that's why this experiment of, call it, although it's a true product now, right, is important because as we integrate with GreenLake, we can integrate the rest of the architecture over time. And then Neil can talk about this Slingshot versus everything else because I think it's a little bit different positioning, I will say.

If you think about networking in the context to be on, the first thing to recognize is that there are 4 different networks in AI systems. You get fabrics that connect accelerators together. You've then got fabrics that connect sets of accelerators together. Then you have data center networking that interconnects this and then you have whatever you're using to get to your data and your storage. So it's a very heterogeneous environment. I think sometimes, when this threat of conversation comes up, it comes up through the lens of, well, what's the one binary answer, there isn't going to be one binary answer. The biggest shift that's going to emerge is that InfiniBand, which has been used in some of the highest performance systems in the world, including some of the highest performance AI systems in the world, is something of an alien technology for many enterprises. And as a result, Ethernet is much more absorbable, but Classic Ethernet is not sufficiently performing for these workloads, which is why you see the innovations and technologies like SpectrumX or technologies like Slingshot or efforts like the Ultra Ethernet Consortium. So over time, you'll see richer, more capable Ethernets among some of those fabrics, but it's still going to remain very heterogeneous of the different layers because they're designed for different things.

And today, Slingshot has been very much focused on massive scale, massive scale, meaning these supercomputers like Frontier, the Capitan later on, Aurora, they are all enabled by the HPE Slingshot fabric. When you go to the footprint that Fidelma is just introducing here is actually that connecting these silicon together, right? And that's why you buy an 8-way server. And that little chip that sits in the middle is the one, the fabric that connects the systems. But then you have the storage, you have to connect. So our opportunity with Juniper when becomes part of us, is streamlined that data centers is working and integrate fully down with NVIDIA as we go forward and make it easy to manage because today, you have 2 different consoles. Tomorrow as a part of GreenLake and the work we'll do it even with Phil today with Aruba Central, one, but it has to be an Ethernet base. I think we have a question.

We have a question, although I want to be mindful because Neil has a keynote.

What time you need to go?

At 12:30.

I need to go now.

All right. He has to go to keynote. Still get closer because you are too far.

We will keep going. We just need to make -- good luck. Break a leg, not literally.

Okay, I don't think this is too hard a question.

Two hard question?

No, it's not too hard, since everybody's left. So Lou Miscioscia, Daiwa Capital Markets. So 2 things. You talked about educating the sales force. Just wondering how far along you are with that journey and the situation of helping you increase your sales in AI and solutions. And then the second half is just like, obviously, the partnership with NVIDIA can't do everything yourself. Wondering where you are partnering with the other service providers out there, whether it's the IND ones, like Infosys or the U.S. ones like Accenture, just wondering if you've actually brought that to bear to these solutions, too?

Yes. So today, we have obviously a crown jewel in the companies of sales force, which is both our direct coverage and our indirect coverage to our channel ecosystem, distributor value-add sellers and also our solution integrators. Some of them is what you named. So from a direct perspective, we had already in place for years, a dedicated sales force and solution engineering for HPC supercomputing and AI and obviously, has been focused on specific segments of the market. For enterprise now, we already started 60 days ago, start enabling the sales force with this solution. So we are ready for today and the availability of the product has happened in the next 60 days or so is basically train everybody, because whether you're an account manager or whether you're a server, seller or whether you're a storage seller or you're a GreenLake private cloud seller, all of them will be selling this, meaning we are leading with this solution in enterprise and only when the customer said, well, I got it, but I just want a server, okay, I will just sell a server, or I want a [indiscernible] unstructured data. Here you have it. And so to give a sense, that's thousands of people. Our entire sales force is very large, in fact, represent a significant percentage of our population. So we're going to train everybody. But it's not just the seller. I think one of the most critical roles here is the presales or solution architect. And that's why we ask our partners to bring their role here. And if you go one of the sections in the 7 acres, there is a section there where we're hosting training and certification sessions as we speak. And then obviously, we have deployed our specialization also inside this channel partners' offices. And I'll take, for example, CDW a great example here in the U.S. right? They have offices all around the United States. We deploy people in their offices, not just in Chicago in the headquarters. So that's what we are doing from a training certification perspective. I forgot the other part of the -- oh, the solution integrator, the same thing. I mean, I have been very clear with my partner ecosystem since I became CEO. We have been -- we are and continue to be a partner-led company all the time. More than 70% of our business goes through our partners in some geos, it's 90-plus percent of the business, and Phil's business is 90-plus percent of the business. Everything he talked about, he has to sells enable everybody. So I always say the same thing, whether you're an HPE batch in person or you are an Infosys or you are a CDW or TD SYNNEX or RO doesn't matter. We treat all the same. And we have events throughout the year to train them on our technology, which is obviously very, very broad.

Great. We'll go to Simon and then Meta.

Simon Leopold with Raymond James. One of the things you haven't talked about in this context is the application of HPE Financial Services. And I think what I imagine is with your enterprise reach and the scale of AI projects financial services could be more significant as a differentiator, could become bigger. So maybe if you could elaborate on how that fits into this overall strategy and the growth angle?

Yes, absolutely. Thank you for asking the question. We don't think of HPE Financial Services as a financing entity. We think of them as a smarter IT life cycle entity. The way we manage that business is on assets fundamentally, yes, we finance upfront, and we make good margins for our shareholders. But at the same time, we actually make more money on the back of those assets when they come out of the customer data center or whatever it is. And what we have seen so far is that again, go back to the customer segmentation. A lot of enterprises will buy on CapEx. That means they may be using leasing, CapEx leasing. Fine, we do that every day of the week. But a lot of them are going to go through the GreenLake model, which, again, last quarter, we already exceeded $15 billion in total contract value. And all of those that went through HPE Financial Services, all of them, from a as-a-service model. And those assets are going to come out at some point in the life cycle, which means we're going to make money on those assets later on even, right, not just on the current contract, new contracts we're going to establish for the secondary market. And I can tell you right now, there is a huge demand for HPE Financial Services. However, enterprise is all good. We understand them. They have -- they're worth everything, the paper and the like. And on some of these other companies that are just emerging, you have to really apply a lot of discipline because sometimes you have these companies want, okay, I want 20,000 GPUs, great. I want to finance it. Okay, great. Show me your credit rating in many ways. And so you have to be really disciplined about that. So clearly, it is a differentiator, not just from financing. I argue from an asset life cycle management from a circular economy and from a sustainability perspective.

Maybe just add a little bit on the networking side as well. So we use HPE Financial Services for our network as-a-service offer, which appeals to customers who find that either difficult to get hold of CapEx. All customers are very interested in sustainability because as Antonio touched upon what HPE Financial Services do is in a typical year, they'll take back about 4 million devices and they're not just HPE or Aruba devices, may be Cisco, Dell and others. And I think in 84% of cases, they'll recycle and refurbish the device and sell it on to another market. So we find customers are interested in us where they're short on CapEx or they're interested in sustainability.

We manage more than 4 million systems every year through that entity. So we have Meta.

Yes, we'll go to Meta for our last question.

Great. Antonio, you've talked to a lot of customers here and just in terms of what are the inhibitors to them investing in AI. You made major strides in terms of kind of easing their transition and maybe leaving some fears about future-proofing. But is it -- use case is it not wanting to make big investments and not know if something else is better is coming around the corner? Just what are you detecting in the conversations with customers about what they're going through and thinking through how they're investing in AI?

Thank you. I think it's fairly simple to me. Number one is defining the use cases for the return on investment, right? I mean, sometimes if you go back in history and say, well, I'm going to invest this much money in IT, I'm going to save this much labor and the labor funds, another way to get fungible somewhere in the organization. I think the CFOs are being very stringent with them about it, but it's really the use case definition for that investment and how they measure the return is number one. Number two is cyber, I think. Understanding the prominence of the model, how they protect the data and how they make it compliant is the second piece of it. Third piece is, obviously, depending on the geo you're in, obviously, you have all these data regulation that comes with it, data privacy and the like. That's why I spent a lot of time this morning talk about the 5 key principles of AI. And if you go to Europe, they went way ahead in terms of AI policies and the like. And so the prominence of these models is one area that they really want to understand. The second is the guardrails around those models, to protect the privacy and so forth. So -- but once they understand that, and that's why this expertise upfront in the consulting is so important. That's why we went ahead with Fidelma and our services team to establish these key unique partnerships with HCL, TCS and Infosys and the like, then it's about the speed of deployment because ultimately, we can do workshops with them, to finding use cases. And after that, they say, do a proof of concept, let's deploy. And once they start showing the results, the light bulbs comes up. I mean Shannon talked earlier on inside HPE, we have multiple AI use cases today, multiple. They're all on-prem. Obviously, we brought them our models on-prem. We trained with our data and we maintained those guardrails to the principle I talked about it. But now use cases are coming out of the w****, all of it. And the question is how you prioritize them for the best return. And so maybe a month or so ago I was in London, I hosted 25 customers that were here in the audience, some of them early on. And what they told us, we trust you, we trust HPE. We trust you because you have secure infrastructure. We trust you because of your values, and we trust you because of your services organization can help guide us through it. We understand the technology, everybody understand technology. But what we did today is accelerate that adoption by making it so simple. It's not going to be a technology problem. It's going to be more a business problem. Because to me, AI at the core is a business productivity tool and second, a decision-making tool. Once we cross the barrier is, we are fine. And that's why customer segmentation really matters. You have 3 curves, Meta, I think. You have these service providers building the models. It's going like this, and then you have what I call the enterprise system beginning an inflection point, which is very exciting to us. And then you have the sovereign clouds by definition they get on it. Otherwise, they will be left behind in this new environment we live it. So it's very exciting for all of us to be here today.

No, that's great. I'm very happy to have joined .

No. And so I know you did that right, but I will say I hope you get the sense that HPE has been transforming itself over the last 6 years. We have 2 of our Board members here with us, the Chair of the company, Pat Russo and the Chair of our HR Committee, Pam Carter, they have been incredible supporters. And listen, for me has been an incredible privilege over the last 6 years to be consistent on the strategy and building -- every year, building a block against that strategy. But I hope when you go through the tour, you're going to see everything coming together. Obviously, there's a lot of interest and about AI, and you will be blown away about the AI technologies. But don't forget how we bring it together with the rest of the parts that enterprises will need and I think HPE is uniquely positioned to capture the massive opportunity and do it profitably to the question that once they stated because it's not just about selling infrastructure, its selling the experience with the software and services you need.

Great. Well, thank you so much, Phil and Antonio. We're going to head to lunch now, but thank you.

Yes. Thank you.

Okay.

And enjoy the concert tonight. It's going to be fabulous. Let's see how good they are compared to my keynote.