Latest / Technology Now / HPE Discover Barcelona: Analyzing Antonio Neri's AI-focused keynote
Transcript
- Aubrey Lovell (00:01): Hello. And welcome back to Technology Now, the weekly show from Hewlett Packard Enterprise, where we take what's happening in the world and explore how it's changing the way organizations are using technology. We're your hosts, Aubrey Lovell. Michael Bird (00:15): And Michael Bird. In this episode, I'm coming to you very nearly live from the floor of Hewlett Packard Enterprises Discover event in Barcelona to bring you the highlights of the keynote speech by HPE CEO, Antonio Neri. We'll be talking about what Antonio's keynote says about the company's direction in the coming months, as well as digging deeper with expert analysis from friend of the podcast, Matt Armstrong-Barnes. It's the second week in a row Matt's joining us. Aubrey, I think we'll probably have to get him a dressing room or at least a director's chair with his name on it. Aubrey Lovell (00:46): Indeed. Michael Bird (00:47): Now in a twist from our usual way of doing things, this episode really is coming to you fresh from the oven. I'm recording this episode on the floor of HP Discover less than 24 hours before you hear it. That is pretty exciting, right? Aubrey Lovell (01:01): Yes, it is. And so glad that you're there, Michael. So if you are the kind of person who needs to know why, what's going on in the world matters to your organization, this podcast is for you. And if you haven't yet, subscribed to your podcast app of choice so you don't miss out. All right Michael, let's get into it. Michael Bird (01:17): Let's do it. Aubrey Lovell (01:20): So HPE Discover Barcelona is the company's major annual tech event for Europe, Middle East and Africa. It welcomes thousands of staff, partners and customer delegates from all around the world to meet, network, learn, and most importantly, talk tech. There's some pretty amazing topics and workshops going on, but the theme of the keynotes is what we're going to focus on. This week, we'll be looking at Hewlett Packard Enterprise's CEO, Antonio Neri's keynote, which focused heavily on artificial intelligence and hybrid cloud, with generative AI tools being the dominant thread. Michael Bird (01:54): And that's where today's guest comes in. Matt Armstrong-Barnes is HPE Chief Technologist for AI. So he's the perfect analyst to break down the keynote and tell us why it matters. I caught up with him just moments ago as an Antonio's keynote wrapped up. Hi, Matt. This was quite a busy speech for AI fans. So as we can now hear, there was absolutely loads to talk about. Antonio Neri (02:17): AI will be the most disruptive technology of our lifetime, even bigger than mobile, Web 1.0 and the cloudware. Believe it or not, since the beginning of 2023, the AI market has exploded. HPE has experienced 250% growth in AI orders in that just nine months timeframe. It's just unbelievable. According to a May 2023 study by Accenture, 73% of enterprises are prioritizing AI above all other digital investments. But 89% need help scaling models into production. Simply put, what worked yesterday won't work tomorrow. So the question is, how are we going to meet the demand? To support our customers on their AI journey, HP is developing an AI native architecture strategy that's simple to deploy and consume. Our goal is to enable your organization to leverage the right mix of data, foundation models, tools, and AI native infrastructure to rapidly train, tune and deploy your models to transform your business. And to help you to be successful, HP GreenLake cloud platform unifies the speed and agility of a cloud native world approach with the performance, flexibility and scale you need of an AI native architecture. Michael Bird (03:48): Hi, Matt. What did you make of Antonio's keynote? Matt Armstrong-Barnes (03:51): I think it's great. Antonio has been talking about artificial intelligence as being a critical component of the HPE technology landscape for a number of years. And what he really talked about was how AI can help customers go through the journey of gathering their data, refining their data, building their models, tuning their models and deploying their models, be that at the AI producer scale, which is using massive supercomputing style technology or the consumers of AI who are using it probably at a much more early stage, a bit more experimental. So whether or not you are starting on your AI journey or starting to build massive scale foundational models, the HPE technology stack along with some key partners like NVIDIA, really will enable our customers to be able to go on their AR journey, not only in a successful way but in a highly accelerated fashion. Michael Bird (04:45): Yeah. So the NVIDIA piece was fascinating. Here's a clip. Antonio Neri (04:49): Today, we are going to advance our strategical operation by announcing a new generative AI full solution that's again, co engineer, but is very specifically targeting enterprises. Here to talk more about how HP and NVIDIA have come together to help businesses unlock the power of AI is vice president of Enterprise Computing at NVIDIA, Manuvir Das. Manuvir Das (05:14): What's fundamentally changed in generative AI is that you have these foundation models that somebody else has already done the work to train up for you and you can start from that. So somebody else did 99% of the work, but yet somehow I can do the remaining 1% of the work and now it becomes mine. So in any large company, there are hundreds or thousands of in-house applications that people build. So really what generative AI is doing is creating the next wave of such applications where instead of doing a SQL query to your data warehouse, you take any employee in your company, let's say you are in PR and you're writing the next press release, you want to be able to write that press release on the basis of the press releases you wrote for the last year. And so you need a little generative AI assistant that'll do that for you. (06:04): So how does the developer build one of these things? What you do is you have these things called embedding models. The assistant converts your question into an embedding. You do a search, you find the best fit answers, you feed it back to a large language model and you get a beautifully curated answer. That's what one of these modern applications looks like. Now, if you think about a typical company, you're going to have teams all around your company building hundreds of these applications and different teams in the company. So what does this say to you? You need one platform that all the people around the company can do this work on, and that's what's so exciting about the thing that we've just announced together because the thing you all did Antonio, which is so great, is you've not just built awesome servers, but you've built a software stack for AI. Michael Bird (06:51): So HP is a relative newcomer in the gen AI space. We only announced our product in the field back in the spring. Did Antonio bring much more on that to the table today? Matt Armstrong-Barnes (07:01): Yeah, so what we're seeing is generative AI has been an evolving technology for a number of years. So from a HPE perspective, it's not necessarily totally new, although we did only announce some of the key offerings more recently, but we have this concept of generative AI producers, generative AI consumers and a hybrid between the two, which is organizations you want to do fine-tuning. And what we can bring to the generative AI space is that massive scale for the generative AI producers as well as the smaller scale for generative AI consumers, as well as providing some capabilities around this sort of in-between model. (07:37): Not only can we do that at the infrastructure level, we can also do it in the software stack that's laid on top of that. And some of the things Antonio hinted at, we might see a bit more from Fidelma tomorrow, it's how we can bring services to bare to help our customers understand how they can navigate this quite complicated field of bringing generative AI into all aspects of their organization to tackle some actually quite complicated use cases and allow it to drive business benefit. Michael Bird (08:04): Absolutely. And here's a clip from Antonio. Antonio Neri (08:07): Today, we are announcing new HP services for AI, which offer a broad portfolio of consulting services, workforce training and deployment solution to support you on every step of your journey. And I'm proud of that work because we work together with many partners who are here in the room who also invest in services for you. Additionally, HP Financial Services combines technology insights, financial expertise, and deep-rooted focus on sustainability to create smarter IT life cycles for customers and partners of all sizes. As you can see, today's announcements dramatically simplify and accelerate the adoption of AI for enterprises of all sizes. Michael Bird (08:53): So Matt, if we were giving someone an elevator pitch for HP's direction following today's keynote, what would it be? Matt Armstrong-Barnes (08:59): I think the first thing would be, we've been involved in this, helping customers go on the AI journey for decades. This is not a new technology for us. We can absolutely help them scale their AI initiatives going forward, and we can do that by accelerating them, by providing the right tools, technology and capability in a unified AI enabled architecture. But what customers need is the right foundational capabilities because building all the individual components yourself to plug it all together is very hard. It's very time-consuming and detracts from what customers really want to do, which is get benefit from their AI deployments. Aubrey Lovell (09:40): That was absolutely fascinating. Thanks so much. So Michael, I'd love to know from you what the atmosphere is like. How has the event been so far on the ground? Michael Bird (09:49): Yeah, it is busy. That's how I would describe it. Very, very busy, absolutely packed with people, which is lovely to see. As I said earlier, I watched the keynote and the room was absolutely packed, so packed, in fact... I didn't mean that to rhyme, so packed in fact that they had to have an overflow room. I think two overflow rooms- Aubrey Lovell (10:09): Oh, wow. Michael Bird (10:11): ... for extra people, so it's amazing. It's great. And of course, Barcelona is a lovely city, so that's always nice as well, isn't it? Aubrey Lovell (10:20): It really is. One of my favorite cities. All right, thanks so much and we'll be back with more from the keynote in a moment, so don't go anywhere. It's time for, today I learned, the part of the show where we take a look at something happening in the world we think you should know about, and it's one from me today. You may have heard the phrase elephants never forget. Well, the same is true of AI, and that can cause problems from a data privacy perspective as well as from the huge amounts of energy and resources required to maintain AI training databases. We've talked about the environmental cost of AI on this podcast before. (10:56): Now, a team of British scientists led by Professor Peter Peter Triantafillou have come up with a novel machine unlearning algorithm that lets an AI forget data without forgetting what it learned from it, or having to replace that data from the next iteration of the program. The algorithm allows three types of data to be forgotten, those being biases, erroneous annotations and issues of privacy. And although it is only at the research stage at the moment, if and when it does come into play, it'll mean our privacy can be protected in a more sustainable way, which is ultimately good news for us and the planet. Michael Bird (11:32): Thanks for that, Aubrey. That was fascinating. And I'm still here in Barcelona with HP Chief Technologist, Matt Armstrong-Barnes. We've spoken a lot about AI, but let's move on a little because we also heard Antonio mention some of the challenges around AI including sustainability and data governance. So Matt, what was new here for you? Matt Armstrong-Barnes (11:52): About some key things that Antonio announced around our deployment of large language models. We are doing so in very carbon efficient locations, which means the overall footprint of using both large language models and when it comes to building AI solutions, you can have confidence that the HP solutions are going to be the most carbon efficient way you can achieve that. Michael Bird (12:14): Where do you think this positions HP in relation to the rest of the industry? Matt Armstrong-Barnes (12:19): Our credentials in this space are very strong and we continue to be definitely a driving force in the sustainability agenda, and that becomes especially true when it comes to building AI systems because we have world-leading intellectual property around how to build AI systems and how to tune them in the most efficient, and thus producing the least amount of carbon when it comes to producing your AI systems. Michael Bird (12:46): We also had a fair amount about hybrid cloud as a platform for AI. It's a big focus area for HP with GreenLake, and that also wasn't left out of Antonio's speech. Antonio Neri (12:55): Okay. So building on our announcements with NVIDIA today, I'm also excited to announce several new HP GreenLake solutions. One of the biggest bottlenecks to effectively train and tune foundational models is applying large traditional storage solutions to unique data demands of AI workloads. And so today I'm very pleased to announce we have made significant performance enhancement within HP GreenLake for file storage to address the density and throughput demands for AI workloads. And with our common enhancements available early the first half of next year, we will increase the capacity density and throughput by seven times. Many of you tell us you want to probably train and tune AI foundation models with your data. Today, we're also announcing a new HP GreenLake Flex solution that includes GPU based, HP ProLiant compute with HP GreenLake for file storage. Think of this as an AI optimized private cloud. Michael Bird (13:57): So Matt, what's new here? Where are we headed now and why does it matter? Matt Armstrong-Barnes (14:02): When it comes to hybrid cloud, it's all about deploying your AI in the place that is closest to where it's going to be useful, whether or not that is at the edge. And by the way, we HPE, through the Spaceborne program, run AI workloads on the International Space Station. So when it comes to having credentials about running AI workloads in remote locations at the edge, we've absolutely got the credentials going back to 2017, I think it was when the first time we ran AI workloads in space, meaning that we now need to start thinking about it in a much more hybrid fashion. So moving AI systems closer to where the data is, meaning they could be much more effective. Michael Bird (14:45): Antonio also mentioned that their partnership with NVIDIA was expanding into digital twins. Antonio Neri (14:50): So we have another exciting news, which is we are announcing HP GreenLake Flex solution for the digital twin. Obviously creating a digital twin of everything is exploding, about simulation using generative and so forth. So this solution actually brings together the NVIDIA Omniverse, which is a phenomenal piece of software and our leading HP Alletra storage portfolio that gives the ability to enterprises to design, simulate and provide customers with the ability to optimize assets and processes across the enterprise. So can you talk a little bit more about this new digital twin solution? Manuvir Das (15:25): Yeah, this is so exciting for us, Antonio, because we are all very focused on generative AI, of course. But Omniverse is a big, big effort from NVIDIA because as you said, there's a physical world, all these companies and industries that are in the business of producing physical things that we all use every day. But this is a very expensive thing to do. And the promise of digitization and the confluence of AI is that you can do all this work digitally. You can design your products digitally, you can design the factories in which you build your products digitally, and so Omniverse is the platform that enables all of that. And the most beautiful thing, Antonio, is these sound like two different things, generative AI, large language models and an Omniverse, but it all comes together. (16:10): I'll give you a simple example. Imagine just as a human that you're designing your living room, and what would be more awesome than saying textually, "How about a couch that is like this?" And you just generate a 3D object for that couch and then you place it in your virtual living room, you move things around until you feel like you've got a good-looking room and then you hit go. And that's exactly what we can now do because with the generative AI, we can produce these models in the right format, they can go into Omniverse. So with you, we've got the end-to-end solution. Michael Bird (16:43): That's pretty exciting, right? Matt Armstrong-Barnes (16:44): Oh, definitely. So the evolution of digital twins is going to play a critical role. As we start to bring more physical things into the digital world, it allows to play out a whole bunch of scenarios that we wouldn't have historically thought possible. Not only that, we can augment it with synthetically created data, meaning we can create scenarios in the digital world that we might not want to recreate in the physical or the real world. And we can also use that as a way as of building out scenarios and data that we can then use as a more effective way of building out and future-proofing AI systems. Michael Bird (17:18): That's great. Thank you so much, Matt, and it's been great to chat. I'm going to leave you to explore now and check in with Aubrey in the studio. How are you doing, Aubrey? Aubrey Lovell (17:27): I am doing well. Things are going well here. A little jealous. I want to be over there with you right now because Barcelona is amazing. But it's been great to hear you and Matt chat as well. And you can find more on the topics discussed in today's episode in the show notes. Well, we're getting towards the end of the show, which means it's time for, this week in history, a look at monumental events in the world of business and technology, which has changed our lives. The clue last week was, it's 1993 endeavor to hold onto your telescope. Did you get it? Well, of course, you did. It was the launch of the first mission to service the Hubble Space Telescope by the crew of the Shuttle Endeavor, the mission STS-61 or Service Mission 1 was one of the most ambitious man space operations in history. (18:14): It involved five back-to-back space walks, totaling 35 hours and 28 minutes during which the shuttle and telescope traveled 4.4 million miles and 163 orbits of the earth. It's pretty amazing. The mission set out to repair an area of the primary mirror, which had become flattened by the width of a human hair, which was enough to cause fuzziness in the image. They also upgraded and fitted several components including cameras and sensors while they were up there. In total, there were five shuttle missions sent up to service Hubble, which saw 30 years of service in 2020 and is set to last well into the next decade. Pretty cool. Michael Bird (18:55): That is pretty cool. Thank you, Aubrey. Now next week, the clue is in 1975, this place really took a bite out of the computer market. Do you know what it is, Aubrey? Any idea? Aubrey Lovell (19:07): No, but there's a great band by that. Michael Bird (19:09): Yeah, very true. Anyway, tune in next week to see if you are right. And that brings us to the end of Technology Now for this week. Aubrey Lovell (19:19): Thank you so much to our guest, Matt Armstrong-Barnes, and to our listeners. I appreciate you all so much for joining us. Next week we'll be revisiting HPE Discover Barcelona with a breakdown of the keynote by company Chief Technology Officer, Fidelma Russo. Don't miss out. Michael Bird (19:35): Technology Now is hosted by Aubrey Lovell and myself, Michael Bird. And this episode was produced by Sam Datta-Paulin and Al Booth with production support from Harry Morton, Zoe Anderson, Alicia Kempson, Alison Paisley, Alyssa Mitry, Camilla Patel, Alex Podmore, and Chloe Sewell. Aubrey Lovell (19:52): Our social editorial team is Rebecca Wessinger, Judy-Anne Goldman, Katie Guarino, and our social media designers are Alejandra Garcia, Carlos Alberto Suarez and Ambar Maldonado. Michael Bird (20:04): Technology Now is the Lower Street production for Hewlett Packard Enterprise, and we'll see you next week. Aubrey Lovell (20:10): Cheers.