0:002020 we started working on autonomous vehicles with Mercedes. As you could see Nvidia's life journey the entire 33 years I've spent a lot of time in Germany. So I'm delighted to be back. Thank you. [applause] I'm delighted to do GTC here this year.
0:23This last year, we started the new journey to building a new industry here in Europe. The beginnings of the AI economy. In the last year alone, we've invested nearly 10 billion dollars in Europe across startups, big partnerships, small partnerships all over all over Europe. This last week alone as I journeyed from France and UK
0:51to Sweden and now here in Berlin, we announced the investments of 14 startup companies. Overall, since the beginning, we now have 12,000 plus partnerships across Europe, and we're all working towards one goal. Now after decades after decades of advancement in the computer industry, we are now at the beginnings of a new industrial revolution. This image was generated by a computer.
1:23It was prompted by me. Nice simple prompt about AI powering a new industrial revolution generated this. I gave it a few partners that I wanted to highlight, companies that we're working with. And this is an extraordinary thing in the mind of the AI. This is what it means to build a new industry, generation of tokens, the produce production of intelligence which
1:51integrates into every single industry powering a new industrial revolution. A new industrial revolution happens when the new technology affects how we build something and what we can build.
2:11How we build something and what we can build. I think in every single way artificial intelligence will affect everything we build and in fact enable us to build new things in this way because everyone will use it. Every company will be powered by it because it's digital intelligence. The foundation of everything. Every single person will use it. Every company will be powered by it. Every industry will be
2:38transformed and new ones will be created in this way. Like the electricity, it is a new industrial revolution and we're at the very beginnings of that.
2:51The question is what is AI?
2:55And this is a big deal if you want to be part of creating this future. Understanding what is AI is vitally important. And this is really a simple framework. We were early to observe that in fact AI is a manufacturing industry.
3:13AI is not just software technology. It is software technology but not just AI of course runs on a computer but the way that that computer operates is fundamentally different. Rather than a tool that's used by us it is now part of a machinery and infrastructure factories that are producing intelligence that are integrated into every single industry
3:40and application. A simple model is to think about as think of AI as a five layer cake. Energy powers advanced computers that are part of an AI factory that generates tokens, digital intelligence from the models that run on it that are integrated into every application. A five layer cake to be part of this AI economy. You don't want
4:07to just be consuming the AI. You want to be producing the AI. We could do all of that here in Europe. And in fact, we've already started. Europe is building its AI economy and it's five layer cake represented here. We've got great partnerships. And when you look across this five layer cake, you can recognize that in fact it is a new industry because there's growth at every single layer.
4:36We're creating jobs, creating growth at an extraordinary rate. Companies here in Europe, Schneider, Seaman's electric, incred Seammen's energy incredibly well around the world. We're the computing layer. We integrate into AI factories that are now flourishing all over Europe. Startup companies like NBS doing incredibly well. a new company called Firebird which is setting up a new cloud
5:05in the Central Asia region and even Cassava in Africa. every single telco in the future and I predict this every single telco in the future with a 100,000 sites representing some 100 gawatts of distributed telco base stations and data centers will in the future evolve into a distributed AI network.
5:31So the AI infrastructure layer is being built out which is then above that the AI models. It's really important to recognize that long before the AI models emerge, the infrastructure has to exist.
5:45Just as before you can have an internet, you have to have the networking. Just as before you could really enjoy the capabilities of electricity, you have to have electrified houses, electrified power plants. And so the models are the next layer. It's also interesting to observe that the model layer is not just language models. It represents language, biology, chemicals, physics, robotic
6:13actions, animation, and it represents models of all sizes. These different industries are going to be powered by different modalities of AI models which then integrate into the AI natives, the startups, and there are so many here in Europe. So the be the Europe's AI economy is starting and it's really accelerating. It's great to see at the foundation of it where we contribute most people think is GPUs. It is true.
6:45We started in 1993 building a graphics chip which emerged into the GPU the programmable shading GPU. Over the time over time we added CUDA acceleration to that reinvented computing as we know it which then brought forth artificial intelligence but Nvidia today is really a full stack endtoend complete AI factory platform
7:14the breadth of technology of our company over the years has really expanded everything from GPU computing Vera Rubin is our latest generation to the Envy Link switches that connect them into one giant GPU, the Vera Bluefield 4 that we recently introduced and I'll say a few words about that later to Spectrum X which is our Ethernet solution making Nvidia the largest networking company in the world
7:42today to Grog 3 a decode accelerator for generation text for token generation to Vera four Vera Bluefield 4 storage acceleration with I'll say a few words in that in just a second and then Vera RTX compute and I'll mention that in just a second but this now represents the breadth of technology necessary to
8:08create the AI factory but nothing works unless you have software that activates it all the first part is the system software that runs the AI factory these systems are gigant gigantic. A 1 gawatt data center, 1 gawatt AI factory costs about $50 billion.
8:30Land, power, shell, all of the compute, networking, storage, power, cooling, all connected together, multiple acres of AI factory, just turning it on and operating it is a marvel in itself. The mag the magnificent amount of software necessary to energize this factory to put it to use and produce tokens is an extraordinary feat. Nvidia offers the
8:59entire full stack software. On top of that are what we call the CUDA X libraries, the algorithms of our company. What NVIDIA truly is is an algorithm company that designed silicon to accelerate them. Of course, one of them is KDNN, what many researchers recognize as the library that brought forth that started the big bang of modern AI. KudnN, it's QDNN is to AI
9:29what SQL is to the world's database industry. It's a domainspecific language, a domain specific library. Just as OpenGL and DirectX enabled real-time computer graphics, KUDNN enabled the AI computing industry, 350 other libraries.
9:51Those libraries are in fact the crown jewels of our company. Everything from computational lithography to signal processing, image processing, data processing, particles, fluids, sparse solvers, incredible breadth of libraries that many of you in the audience have connected into your company. This is how we open new markets. This is how we connect NVIDIA into algorithms that
10:19enable you to do computation at incredible speeds. And then finally on the right hand side on that side the models this is the new software the new engine of new applications and it sits on top of and it collaborates with the CUDA X libraries. These models Nvidia have been working on across multiple domains. With these models it opens up new opportunities and I have lots to say
10:48about that later on. Well, every single generation Nvidia started with Pascal. The AI era started with the Pascal new what is called DGX1 system. DGX1 brought to the world the world's first AI supercomputer. I introduced it 10 years ago. Delivered the first one to OpenAI.
11:17started their computation journey from Pascal to Volulta to Aier then to many of you who heard about us for the very first time Hopper the current generation Blackwell and then now Vera Rubin in full production notice six generations of architectures with each generation we advanced the technology in the course of that last 10 years
11:46we've improved improved AI performance by 1 million times. If you would take a step back and think about what would Moore's law have done in 10 years time, it would have the first five years would have improved by a factor of 10. the second fact second five years by another factor of 10 a 100x improvement in 10 years time in the same period of time because of the way Nvidia extreme code designs were innovating across every
12:14single computer across every single stack of the software optimizing new algorithms creating new algorithms while creating architectures to accelerate it working across that entire domain what we call extreme code design we were able to accelerate AI AI by a million times and we're going to continue to do so.
12:35This is what Nvidia builds today and this is the foundation by which we can enable Europe to be part of the AI ecosystem, the AI economy. Now each generation in the beginning Pascal was GPUs, Volulta was GPUs, Amper was GPUs and Hopper was GPUs. We invented the new DGX form factor. Then it was Blackwell that
13:02introduced to the world the first rack scale computing. And I'll show you that in just a moment. Rack scale computing turned one entire system into one giant computer, one GPU. And now with Vera Rubin, we're extending it to processing beyond models, beyond large language models to now processing the entire agent. Now, we'll refer back to this slide later,
13:32but remember this, the agent has to think. The agent has to think, and that's where the model comes in.
13:43perception, reasoning, planning and using tools. All of that runs on the large language model and the models are designed to do that. But the agent also has what is called a harness. Basically this exoskeleton of the digital agent. Just as we have the brain, the rest of our body represents our exoskeleton. The agent is the exoskeleton that orchestrates the
14:12work that allows the agent to actually engage memory, active memory, working memory as well as long-term memory and activating tools. Then the third part is a computer that processes knowledge designed for data processing. And then lastly, a computer dedicated to CUDA X libraries. And its job is to run tools. And these computers now represent the entire agentic system.
14:42So Ver Rubin is in full production. It's nice to kind of see it. This is the first computer in the world that we've created. No cables, no wires, completely 100% liquid cooled. And it's just really delightful. Whoops.
15:01Really delightful to be able to pull it out. And this is one Vera Rubin computer. Okay. So, this has two GPUs. Yep.
15:13[applause] What used to take in the case of Grace Blackwell two hours to assemble one node. And when you see it, you'll you'll know what I mean. Cables and wires everywhere. 80% liquid cooled. Now, Vera Rubin is 100% liquid cooled. And it takes less than five minutes to assemble it. And so incredible, incredible achievement. Of course, this translates
15:40to higher cycle times in manufacturing, more better, much much better reliability, and of course, a lot better performance. This is the MVLink switch tray. And what this does is it connects 72 of these GPUs. There are four of them in here right now. It connects 72 GPUs into one gigantic GPU. Okay, so each GPU is 20 pedlops connecting 72 of them into one gigantic system.
16:23And this is our new Vera computer.
16:40Be go easy. Let her let her give her a chance. Everybody has a chance to say something. But today, let me [applause] No, no. Take just I'll talk to you later. I'll talk to you guys later.
17:09All All important issues. All important issues. But let me get through my keynote. We have we have so this this auditorium is packed. We have a hundred other watch parties around the world.
17:27a 100 watch parties, 15,000 people are watching us right now. And so I'll get through the keynote. There's some really, really fun things I want to show you in the end. Okay? So, you just have to you have to I have to torture you through some education first and then the entertainment is coming. This this our brand new This is our brand new Vera computer. Vera CPU is an extraordinary CPU beast. Eight instructions per clock.
17:56It has the world's first LPDDR memory sub memory system so that we can have high performance very high bandwidth as well as extremely excellent energy efficient efficiency.
18:09The the onchip fabric is the fastest of any chip in the world. As a result, data processing is off the charts. Agentic sessions starting and processing is off the charts. And this is designed for doing exactly that. This is our Vera computer. And this computer here, this computer here is the Spectrum X. This is the world's first co-packaged optics. And you could
18:38see it here. I'll show you. Some of this some of this area falls into the ground. And some of this area you can walk on. You I'm just not sure which. All right. So, so if I if I accidentally disappear, you guys know, you guys know us.
19:03It's not a stunt, it's an accident. And so uh you could see here uh these are the lasers and these lasers come up through optics and it directly connects directly connects to our chip and it connects using a process technology we invented with TSMC called coupe and so silicon photonics are directly attached right onto our chip. Otherwise, what you would see is
19:31all of this would be PC board and PC board's transmission capability is much worse than optical cables. And so we now have extremely high data rate, extremely high quality data that comes directly onto the chip. And this goes directly into otherwise transceivers without uh going through without going through a PCB and a transceiver goes right into the optics. Okay, so this is the world's
20:00first CPO. We're in complete volume production and uh 512 port at 200 Gbits per second. The world's most advanced Ethernet switch and we're really proud of it. Thank you. [applause]
20:20And so so what that what that gives us what that gives us is the ability to uh process the thoughts the large language model uh process the agent harness run data processing for its knowledge and then run accelerated computing for all of the tools that represents essentially the agent workload and I'll show you more in a second while Vera Rubin is in full production production and the performance is off the charts.
20:50Every single generations, as I mentioned, through extreme code design, we're able to achieve performances that are well beyond reason, well beyond what Moore's law would have would have dictated. And as you know, Moors law has really largely ended. And so we need these extreme code-design ideas working from chips to systems systems to system software to algorithm to the entire AI factory. Well, Vera Rubin is another giant step
21:19on top of Blackwell. Blackwell was about 30 times faster than Hopper in doing a gentic in doing large language model processing. And now Vera Rubin is another 30 times greater than Blackwell in just two years time. Let's take a look at how Ver Rubin turns into what is a DSX full stack AI factory.
21:49AI is creating the largest infrastructure buildout in history. The factories are massive. The investments are significant. We have to build it right. But great chips alone don't make a great AI [music] factory. NVIDIA DSX is a full stack AI factory reference architecture co-inovated with the world's leading infrastructure [music] companies. Compute, networking, storage, software, power, [music] and cooling
22:17co-designed as one system, optimized from design to operation. DSX starts in a digital twin. Performance, power, water, and scale [music] validated virtually before construction begins. PTC turns equipment designs into simulation [music] ready assets. DASO systems connects rack and facility design through modelbased systems engineering. Schneider Electric [music] simulates power distribution and liquid cooling. Jacobs brings it all together in the Omniverse [music] digital twin. Seammens evaluates thermal, acoustic, and grid impacts. And Procore [music] supports construction phase planning.
23:01Power limits how much an AI factory can produce. DSX Max LPS is an end-to-end [music] control loop that dynamically steers power to match workload demand. It uses 30% less power for the same output.
23:17DSX rethinks [music] cooling. NVIDIA Vera Rubin systems are designed for closed loop liquid cooling with coolant temperatures up to 45° in suitable climates. Outdoor dry coolers release heat into the air without evaporating water. At 2 gawatt, DSX [music] can save 54 million lers of water a day. That's the daily [music] water use of 180,000 households. But an AI factory can take several months to build. Until it is operational, it produces no tokens.
23:52[music] Today we are announcing Nvidia DSX Pod bringing this full stack platform into a factory-built integrated system. It starts with Vera [music] Rubin NVL72. Seven liquid cooled compute racks connected with 17 km of Spectrum X cables form a Vera Rubin DSX pod. Vera Rubin DSX [music] pods are integrated, fully cabled and factory tested, then delivered in transportable [music] containers.
24:24Dry coolers replace traditional evaporative [music] cooling. 800 volts DC connects to the racks for efficient power delivery. Once site [music] utilities are ready, estimated on-site deployment goes from several months to days. NVIDIA DSX AI factory is responsible by design. And now with DSX Pod, AI factories can be built at the speed of light. More intelligence per watt. Less [music] time to first token.
25:04[applause] So in 2016 we introduced the world's first AI supercomputer. It was in one node, one GPU node. Then three years ago we introduced the Grace Blackwell MVLink 72 rack scale computer. Both times we had to reinvent not just the computer, the software and how it works but also the supply chain.
25:41In the case of MVLink72, we took the entire rack which would have otherwise been assembled on site and we assembled it inside the supply chain. We essentially manufactured supercomputers at scale. It is the reason why we're able to move so fast and be able to provide for literally several hundred data center buildouts around the world at the same time. Every single computer
26:10manufacturer in the world now manufactures MVLink 72. The scale at which we are now producing computing is unheard of ever, unimaginable. Well, even then, when you bring these systems onto your sites, there are still months and months of assembly. And you just heard 17,000 kilometers of wiring to connect just Spectrum X for the scale out computing.
26:4517 kilometers, 17,000 kilometers. It's crazy number. And so 17 kilometers, incredible numbers. Now suppose we were to do that all inside the supply chain. Run everything up, build everything up, make sure that it runs completely, get it qualified, burnt in, and by the time we ship it, it is fully functional. The amount of cabling and onsite assembly is going to be greatly reduced. But there's something much, much more.
27:17In fact, what's going to happen here is that by moving this into the supply chain, the amount of idle computing, the amount of idle computing that's going to be sitting inside at your data centers will reduce dramatically. Remember this, each gigawatt, each gigawatt data center is $50 billion.
27:40The revenue generation you expect is tens of billion dollars per year. And so if we just said it's $40 billion per year of revenue generation that comes out of that infrastructure. 40 billion if delayed by just three months. That's $5 billion. Excuse me. What is that?10 billion worth of revenues. $10 billion worth of revenues will make anybody anxious.
28:13And so by moving this into the supply chain, the speed by which we can turn on go from installation arrival of computing to production of tokens will go from many months we think something along the lines of four to five months down to about a month. Okay. So this is a complete transformation of the world's supply chain and this will change the way that people install and build data centers
28:41and hopefully as a result velocity goes up the wasted idle compute the wasted idle capex will reduce and everybody becomes more profitable and so this is MVLink 72 now into a large cabinet system we call it DSX pod Okay, so this is the next click of Nvidia computing. [applause] Now what you make and how you make it makes you special. Just as here in Europe, you the work you know makes you special.
29:28You design machines.
29:30You operate factories.
29:33You discover medicine.
29:35You have extraordinary craft. That craft, that specialty makes you special. Well, what makes Nvidia special are really two things. The first thing is that our architecture is the only accelerated computing architecture that is general purpose.
29:55We don't specialize on just one ar one algorithm. For example, transformers. We specialize across anything that could be paralyzed. Could be fluids and particles. It could be quantum. It could be quantum chemistry. Could be image processing. Could be data processing, structured vector data processing, processing of all kinds. And because we've been working at it for so long and because Nvidia's architecture has remained consistent for literally two
30:24and a half decades, the number of algorithms, the number of domains, the number of industries we serve, we accelerate has grown tremendously. our install base, our general purpose architecture, our dedication to this architecture and our collaboration with all of the world's industries has made Nvidia's architecture literally everywhere.
30:48The second thing is how we do it. Of course, we have the architecture that we developed, but we also build it now in a full stack complete AI factory. There are many companies in the world who want to design their own chips but most of the companies in the world just wants to deploy the AI infrastructure so that they could run their business on top of it. The ability to design a chip and build a supercomputer write all the
31:18software is really completely out of mind for most of the companies around the world. And so we now have the combination of these two ideas, CUDA, full stack AI factories, and it exposes us to markets that otherwise is impossible. We're able to partner with so many companies here for this very reason. If I were to go around this, starting with CSPs, this is where most people think the vast majority of AI is
31:47today. But as you can see in the beginning of our keynote, I mentioned a whole bunch of companies who are not those. So CSPs remain very large businesses for the world's computing.
31:58But there are AI natives, companies that are starting all over the world. There are so many that are here. Neoclouds because every region wants to participate in the AI economy. Every region wants to ensure that the valuable resources that they have, the energy that they have is dedicated to companies in their region. For example, Mistrol here in in Europe, Mistrol in France, Nibbas here in Europe, so many companies
32:27around Europe that are building up neoclouds, OEMs and ODMs who are building computers for the world's enterprise, AI labs. Of course we know about last last year all we spoke about was open AI to today we spoke about we speak about open AI anthropic but we also talk about the incredible meta muse gro bot and of course Gemini so many different AI labs
32:56are now participating at the frontier and then very importantly and this is ultimately the most important for AI to succeed for Europe to benefit from AI and to have really created an AI ecosystem and AI economy, you need all of the end markets, the end industries to adopt AI and to succeed with AI.
33:18Whether it's in the automotive industry because we're building self-driving cars or the way you manufacture with robotics, the way you discover new drugs in healthcare, your financial services in inst uh industry, all of these different industries have to become AI companies. every single one of them.
33:37Well, we have an unique ability through what I just mentioned about CUDA and the way that we build computers. We have a unique ability to serve literally every single segment of AI.
33:53And so one architecture, many markets, and because these infrastructure are so significant and so expensive, we need to make sure that they're as productive as possible as soon as possible, that they're fungeible that you can use it for every workload from the beginning of your AI journey, data processing to training to post-raining, inference and agentic processing, every single step of the AI journey to every
34:22type of AI, chemicals and physics to languages to every single industry and the algorithms that they use. And so Nvidia's architecture is funible and as a result of that continuous innovation, continuous updating of software because the installed base is so great, we have such an incredibly rich ecosystem of software. Then this architecture is also durable, fungeible, durable, productive.
34:50Well, that lays the foundation. That is the infrastructure. Now, what do we do with it? This last year, we saw a transition in AI. AI is now productive and it's one of the most important times was in the beginning of the year when we so spoke about open claw which introduced the concept of an open-source agent. It really started what we all see across the world today. Agents
35:19think, reason, iterate, use tools, and as a result consume an enormous amount of tokens. But as a result, they become useful. So more agents are being developed. They're more useful and there and therefore they're used by more people and because they have to use a lot of tokens to think, the amount of token consumption has gone through the roof. GPU demand has really gone through the roof. Now what's processing it is
35:47this. This is an agent. These are the five elements of an agent. The first part are the language models. These are the tokens that are running on top of the GPUs and they're generating tokens that are consumed by the agent. What I I called the exoskeleton.
36:05That open claw harness sits inside a sandbox and that sandbox isolates it from the rest of the system. This is a very important concept and we're going to come back to there are five things. There's the model, there's the harness, the sandbox that the harness sits in and two other things, knowledge and tools.
36:27Models, sandbox, harness, knowledge and tools. These are the five elements of an agent. Now before I talk about how to make the agents useful, the most important thing is we have to make sure that the agent is secure. The agent is autonomous and it's autonomous system that has access to files, has access to resources, networks, access to tools and we need to make sure that the privileges
36:55that we provide this autonomous system is minimal and that we constantly monitor it. We introduced a fantastic new product, a new platform and it's called OpenShell. Open Shell came just in time. This is a secure runtime that provides for the things that I just mentioned and has an additional hardwarebased monitor. Before I explain it to you, let's watch a video.
37:29rises are preparing to deploy billions of AI agents to be useful. They'll reason, plan, use tools, [music] access data, and work together. Today, the only thing stopping an agent from misusing [music] all that is asking it nicely. Please don't delete anything. Please don't push [music] to me. Please don't touch my SSH keys. That's the please don't approach. But hope isn't security.
37:55[music] Safety isn't an aspiration. It's technology and engineering. NVIDIA [music] OpenS Shell and Sentry put that philosophy into practice. You write a policy for what each agent can touch and OpenShell [music] enforces it at the granularity of system calls, file access, and network connections. Formal verification checks [music] every policy change against your permission ceiling, so agents never gain more access than
38:22intended. It's like the early internet. Viruses [music] were everywhere until browsers enforced strict limits on what a web page could do. OpenShell does the same for agents. Enforcement happens [music] outside the agent, not just in LLM guardrails, bringing agent native security [music] into your infrastructure. Every agent runs inside its own isolated sandbox with no direct route [music] to the network. Every outbound network request goes through the supervisor outside the sandbox. It
38:54checks [music] the action against your policy. If it's not permitted, it's denied. Keys stay outside [music] with the supervisor and are added only on the way to an approved endpoint. A compromised [music] agent can't leak a key it never had. The gateway is the control plane. The supervisor makes [music] the decisions. The sandbox holds the agent and the policy prover checks proposed policies [music] against your permission ceilings. OpenShell combines Linux kernel isolation with runtime
39:26NVIDIA Sentry adds an optional layer of protection outside [music] the host itself away from the agent. Running on NVIDIA Bluefield 4, Sentry [music] monitors agent activity and enforces security policies in a hardware isolated trust domain. Because agent reasoning traffic passes [music] through Sentry, if the agent is planning to do something, it will update policies immediately and cut the connection to [music] the model, quarantining it in milliseconds. And it keeps tamper
39:56evident [music] records that operators can verify for detection, investigation, and response. OpenShell is open- [music] source under Apache 2.0, runs everywhere your agents do, edge to cloud, [music] and takes two commands to adopt. Install the CLI, create a sandbox, and [music] run your favorite agent. Agents are only as useful as the access you give them. Open Shell and Sentry put that access under your control. That's how [music] we advance AI capabilities and safety together.
40:30[applause] Agents are autonomous systems. We have to create extremely secure technology to make sure that we could develop these agents and deploy these agents safely. Now, it's incredible after announcing Open Shell, just about every major company in the world from the frontier labs to the operating system companies, cyber
40:59security companies, platform companies have come out to adopt Open Shell. It's really fantastic to see and the reason for that is because this technology has arrived at exactly the right time at a time when we want to deploy agents all over the world. Now remember this isn't just about the services using Open Shell as they deploy agents. This is really about all of the world's companies also adopting Open Shell so that you could receive these agents and have them run
41:26autonomously inside your company. every single level of capability, every single task that you give the agent, for every person that uses the agent, every one of those combinations requires a different level of privilege that you would provide to the agent. And every single time an agent is launched, the privileges are different. And so privileges, the policy enforcement, the monitoring, all has to be done
41:56dynamically. And so open shell is really revolutionary. I'm delighted to say that Enthropic and Nvidia are partnering to create cloud open shell uh agents. And so you can go to the website uh there's a really great write up about that. So the first part the first part is how do you use these harnesses open claw hermes um open hand uh you could create your own own uh harnesses from lang chain uh
42:24deep agents there's or just write it from python there's a whole bunch of different ways you could create agent harnesses that sits inside now an open shell secure harness okay and that's the sandbox that it sits in the secure sandbox bots. And then next, let's talk about models. Uh there you should use uh models of all kinds. In the future, it won't just be one model that we run everything on. We're going to run the
42:53we're going to use the best models whether they're closed, open, or custom. We're going to use all the models to do your work. Sometimes the work that you do is fairly generic. artificial general intelligence is going to do an incredibly good job. However, sometimes the work that you want to do captures too much of your proprietary knowledge, too much of your specialty, and you want to continuously enhance that intelligence.
43:21Every company in the future will use both closed models as well as open models that you customize. By the end of the session, you will see that in fact building your own AIS is not that hard. It starts with of course the model. We've been dedicated to advancing open models for to open up new opportunities and enable people to build their own custom AIs.
43:46Neotron is for languages. Cosmos is our world foundation model. We are the frontier.
43:55Groot for human robotics. We are the frontier. Alpio for autonomous navigation. We are the frontier. icing for quantum classical computing. We are the frontier. And of course, Earth 2 physics, AI, we are the frontier. So, in all of these different areas, we offer you all of the models and we're continuously working on it. We've been working on it for a decade and we'll continue to work on this forever. We
44:22offer you the model completely open, the data that goes along with it so that you know the providence of the model that we created as well as the scripts necessary for you to reproduce it. So not only do you have the open weights, you have from us the entire recipe and the data to recreate the open models so that you could know where it came from and have complete trust into it. Well, these models
44:50are not necessarily the frontier of frontiers, especially for the language model. However, with customization and with partners that the ecosystem that we work with, you could turn these models into frontier and as a result, turn your propri data into sovereign intelligence.
45:09Here's an example of A21 helping Alice turn Neimotron into a model that exceeds Frontier capabilities. Here's Base 10 helping Code Rabbit turning a Neotron model after fine-tuning on your data turned into Frontier Capability. Base 10 and Code Rabbit. Here's trajectory in Harvey again using Neimotron post-trained it and turned and created their own
45:39frontier level capability. Here's scale working with Sky Dance again doing the same thing. It goes on and on and on.
45:46The point being that you could start with an open model, something that's incredibly good like Neotron and with your own data, post- train it for your use case and achieve, of course, super intelligence. That is a very sensible thing to do. And all the technology exists today. Here's an example of Palunteer with Tampa General Hospital. The list goes on and on. We have a rich ecosystem of partners that can help you and they're incredible
46:15companies and they can work with you, you know, from around the world. Okay, so I we work it's just anyways I I'll mention one and I'll offend everybody else.
46:25I'll have to sit here and me mention all of them and I love them. I love them all. And so first agent harness running in a secure sandbox open shell. Second, Neotron, large language models. The first two components of what you need. But ultimately, what turns these super agents, these agents into super agents is that you give them access to your
46:52company's knowledge. And this is really special. This is called the agent knowledge foundation that we put together. libraries of models that are completely worldclass and allows you to turn your company's data the data could be in the assets could be in uh uh uh unstructured format in documents so that you need an AI to understand how to read these PDFs and read all these unstructured documents or it could be in structured data inside
47:22We have the libraries and the tools necessary to help you extract meaning and understanding from your company's data one at a time. Neotron parse completely worldclass top of leaderboard. Neotron embed taking a query and associating with the knowledge that's in inside the vector database top of leaderboard.
47:46Neotron rerank it might find many different sources where the query is associated with different pieces of data. It brings it together ranks ranks them based on which one's the most relevant uh to you. Uh kumo to be able to predict from your structured database what is likely to happen. And so this is for example if you were to receive um a notice that one of your suppliers
48:15will be uh short on supply or delay on supply that re that piece of information is likely uh unstructured data as a document. the information about the orders in your company are likely inside a structure database uh your SQL database for example and in order to predict what to re what's going to happen to all of these orders who are going to be short the the amount of delay you use kuma to make that
48:43prediction and then in order to uh figure out what to replace it with you might use vector database search and to go find a engineering document of a replacement component Okay, so all of those different uh libraries and data processing capabilities now makes for you your knowledge foundation. You've now armed your AI with proprietary data, proprietary knowledge. Now the last part that I'll talk about is ultimately tools. Tools of the trade.
49:16Nvidia has been working on CUDA X and for most of you in the last before AI all of our conversations have been about around how do we accelerate simulators and solvers data processing engines domain specific libraries of all kinds. They're used in scientific discovery in all the different fields of science.
49:36They're used in engineering. They're used in research. They're used in designing building creative arts. at its core are CUDA X libraries and these CUDA X libraries are algorithms and algorithms as you know are beautiful. Let's take a look at it. [music] [music] [music] Heat. Heat.
51:34[music] Heat up here.
52:14[music] [laughter] [applause] Math is beautiful. Math is beautiful.
52:32CUDA X libraries, these are the crown jewels of our company. And the reason for that is this. When you're building an AI assistant, it could be a a lab assistant, it could be a research assistant, it could be a co-designer, somebody who's co-creator with you building something with you. Ultimately, what makes those agents useful, what makes them able to do real work are the tools. And so we we now have
53:01models that allow these agents to reason, understand information, reason, to be able to plan and iteratively act and use tools until it could get the job done. But it still comes down to the tools. And these tools we've been working on for so many years. And there was one particular computer I showed you on the right hand side that runs every single one of these tools.
53:25And it could be fluid dynamics and particle physics. It could be quantum chemistry. It could be data processing. It could be ray tracing.
53:33All of these tools run on that one computer. And that one computer speeds up the processing of it compared to just running it natively on a CPU by 10, 50, 100x. And CUDA CUDA X tools is what give the agents the tools of the trade. And it's so cool to see how other people are using it. Let's take a look.
54:19[music] [laughter] So, [applause] So this is the magic magic behind the agents. Ultimately the agents have to use tools. Now of course in one of the first fields of agentic use is software programming and therefore a compiler or Python is all really it needs. But in the future as we have these tools do more and more and more and more useful work we're going to need to give them more tools. And here are some examples.
54:49So here's Lida Sciences. They're creating a lab assistant and they're using NVIDIA's Biono models that are designed for understanding proteins and understanding cells and generating proteins and chemicals in drug discovery. They're using Biono and CUDA X tools. This is Seaman's factory building a factory agent that uses NVIDIA Cosmos, our world foundation model that understands the physical world and CUDA X tools. And here's
55:16example of do creating a virtual companion that uses Nvidia's Neotron, our language model, and CUDA X tools. Let's see them in use. The most intelligent scientist still needs a laboratory. The best engineer still needs [music] simulators and solvers. An AI agent is no different.
55:41An agent brings together five building blocks. The model [music] perceives, reasons, and plans. The harness coordinates the work. Knowledge foundation gives the agent context from your company's [music] structured and unstructured data. Tools let the agent simulate, experiment, [music] and act. NVIDIA Open Shell controls what agents can access and execute. Start with a frontier [music] closed model or customize an open model.
56:09With NVIDIA's agent toolkit, any company can turn its knowledge and expertise into a customized [music] super agent. Let's look at three examples. For Laya Sciences, the goal is to design a therapy for B cell lymphoma. The Laya Iris agent built on NVIDIA Neotron turns that goal into an experimental [music] plant. In the computational dry lab, LILA agents use NVIDIA Bioneo agent toolkit to call models for protein design and RNA optimization, [music] running millions of virtual experiments.
56:44The most promising candidates move into Laya's AI [music] science factory, where autonomous robots and instruments developed with NVIDIA Omniverse and [music] Cosmos make and test them. The results return to Iris, shaping the next designs and experiments. The loop [music] continues at Seaman's Electronic Factory Erlangan.
57:04The [music] goal is to bring new products into production. Process preparation X agents use NVIDIA [music] Cosmos to generate virtual prototypes and synthetic data, then train and [music] validate inspection models before production starts. Intelligence Center X agents built with NVIDIA's factory operations blueprint use orders, supplies, and line setups to recommend [music] production plans. Digital Twin Composer uses NVIDIA Omniverse libraries
57:34to validate those plans with physics simulation and realtime [music] factory data before changes reach the factory floor.
57:42With the DASO systems platform, an engineer needs to adapt a Volkswagen Golf Door design for the Tiguan. Aura and LEO use NVIDIA Neatron to reason over engineering requirements. The harness [music] coordinates the specialized sub aents and preserves context across planning, design, and simulation. Aura uses Innovia to plan the [music] work around the vehicle's dimensions, weight limits, and other engineering requirements. Leo design
58:11agent selects [music] a compatible golf door design and uses KIA to reshape it for the TIG one. Leo simulation agent uses Simoleia accelerated by NVIDIA CUDA X libraries to test the [music] variance revealing a vibration problem. The results return to the design agent for refinement. The engineer reviews the refined designs and selects the final one.
58:35Different goals, different expertise, the same [music] building blocks. Open models, CUDA X accelerated tools, and the Open Shell secure runtime brought together with the NVIDIA [music] agent toolkit. Isn't it great? So, [applause] some people said that AI is here. It's going to be the end of software. It's going to be end of tools company. It's exactly the opposite. AI is here.
59:07There'll be more users of these tools that we've created over time than ever before. Every single tools company will now become an AI company. They'll be an agents company. Not only will they offer tools, but they'll offer you specially optimized agents that are incredibly good at using their tools. Cadence and Crowdstrike and Do Ivious and just about every enterprise company out there are all building agents and we're just delighted to be working with them.
59:36Building agents is not that hard. It's got four compon five components. It's got the model, the harness, both of them have closed and open versions, the open shell secure runtime, and then most importantly the knowledge and tools. And those knowledge and tools belong to you.
59:54That's your expertise.
59:56And so all of this is now possible to do. And I'll show you more in just a second. But before I do that, let's talk about physical AI. That's really the the next wave of AI. AI that understands the physical world, can reason about it because the physical world is diverse.
1:00:15It is complicated. It's got extremely long tails. and to reason about it and use a appropriate tools to interact with the physical world. Whenever people think about physical worlds, they think robotics. In fact, one of my favorite physical world applications is the radio cell phone cell towers. And the reason for that is this. It uses instead of cameras, instead of speakers and
1:00:43microphones, it uses radio waves to sense the world and to inter it reasons about it. It processes information, reasons about it and interacts with the world using beam forming. So in fact, this is an application, a technology that is just perfectly suited for AI. AI will revolutionize the way we do telecommunications. It will enhance our spectrum efficiency tremendously just as
1:01:12we've done so with ray tracing all of the graphics that you've seen completely 100% ray traced the amount of computation we thought necessary in the past and now that we apply AI to is a million to one difference and so we're going to bring to the telecommunications network an enormous step up in spectrum efficiency but beyond that we're also going to revolutionize all of these cell towers there's 100,000 sites around the world around the western world about 100
1:01:41gawatts of distributed data centers that will in the future become a distributed AI infrastructure. We're delighted to partner with many companies here in Europe. And just this last week after working on this for nearly eight years, Nvidia made the first commercial phone call on the on the T-Mobile network with Nokia. And so now CUDA makes phone calls the traditional way. Then once you can
1:02:10do that then we can layer AI on top of it. And so the revolution is going to begin. This is really exciting opportunity for us. The other physical AI application that I'm super excited about is essentially robotic radio therapy. Your t the tumor is changing in shape. The physical body is changing while you're in in therapy. The robot, this radio adaptive radiotherapy system has to perceive the changing
1:02:39environment, understand where the tumor is, segment it just perfectly and direct precisely direct the radiotherapy accordingly. Essentially robotic That's right. Go ahead. [applause] This is a form of robotic surgery. Now, of course, the highest volume, the most nearterm robotics application or physical AI application that everybody recognized has been something we've been working on for over a decade. And now we recognize that every single car company will want to make their cars part of the
1:03:19ride hailing fleet. Today we have ride ride hailing and as part of Uber and here in in Europe we have uh Bolt and we have Grab and Lyft and so on so forth and all these different ride hailing services. In the future there will be robo taxis that will be connected to those ride hailing services as well. But it won't be from just one brand. And so we recognized that building these systems are incredibly expensive particularly the software that goes on
1:03:48top of it. We decided to create one standard architecture. 14 different cameras, nine radar systems, three LAR, teen ultrasonics, this cocoon of a shell of sensors powered by dual NVIDIA GPU systems for the A AV computing system.
1:04:12We turned this into an industry standard and now it's being designed into one car fleet after another. On top of it, our AV software companies and we are in in fact one of them as well to build an autonomous rate robo taxi capability which then is connected into uh the ride hailing services. This ecosystem is being built out. The world travels about a trillion miles per year around the world. Many of those miles in the
1:04:41future, you know, a dollar or two per mile uh are going to be uh on robo taxis. So this is one giant opportunity, but it's incredibly fragmented. And so creating an industry standard, Hyperion, and working with an entire ecosystem, we can create essentially a giant virtual fleet um that everybody can connect into. Today, I'm super proud to announce that four more brands, four more cars
1:05:09are going to lines are going to come in. The VW van, I just can't imagine anything more iconic. uh the JLR brand new car um the uh Volvo fleet as well as Mercedes next generation are going to be also Hyperion compatible and be part of this network this ride hailing network robo taxi fleet. So welcome to Hyperion.
1:05:34[applause] The next major industry, the next major industry, the next great industry, and this one is likely to be extremely large, is human or robotics.
1:05:49There's a reason why robotics is hard to do because all the different work is rather diverse. And so building a special configuration for each one of the diverse tasks segments the robotics industry into all of these different configurations which prevents the technology flywheel to happen. The reason why the cell phone has advanced so quickly, the reason why personal computers have advanced so quickly, the reason why Nvidia's MVLink72 has advanced so quickly is
1:06:18because those standardized configurations allows many customers to use it. The volume to therefore be derived. That volume drives the technology flywheel and it gets faster, better, incredibly fast. And so we need a robotics uh configuration that is as standard as possible.
1:06:40Now human robotics obviously has the ability. It probably isn't perfect for everything, but it's likely to work for many things. As a result, this is likely going to be the largest volume robotics configuration. Now, if we could just create the AI that allows it to learn very quickly, then for small medium businesses who don't have an entire robotics team to teach and program the robot can now show the robot what needs
1:07:08to be done, the robot because it has general capabilities and general intelligence would be able to learn and adapt quickly. And so, this is likely to be the next major platform technology platform. We're working with so many different companies. You were recognized many of them here in Europe. This is your great opportunity.
1:07:29No country, no region has dedicated itself to making meatronics, electronics, mechanical systems at the level of precision and craft as the European region. So many companies here are incredibly good at manufacturing. This is going to be the next high volume industry. And you could see there's so many companies that are now part of it. and we're delighted to to be partnered with you.
1:07:57[applause] Now, in order to create in order to create a physical AI system, you need three computers. The first computer is the one we've been talking about and everybody knows quite well, which is training the AI model.
1:08:16The second computer is where we have the AI learn how to do its skills. You can't just have the AI program the AI and have the AI come out into the physical world and try to do whatever it wants to do. You do it inside a digital twin. We create a digital twin called Omniverse.
1:08:35And you've seen us talk about that repeatedly. And in fact, just now when when you saw all the CUDA X libraries, they're all being done inside a digital twin world. That digital twin world is called Omniverse. Omniverse is a library that's connected into all kinds of platforms and tools, Seammens and Doe and Cadence and so on so forth. And so you could see these digital twin systems being created. The second computer, the third computer is the robotics computer.
1:09:04That computer integrates in the case of the self-driving car, the robo taxi, we call it Hyperion. Hyperion consists of the computer itself, the functional safety system, the active safety system, a operating an operating system designed for physical AI safety. Well, we need to do exactly the same for robotic systems.
1:09:28So that every single time you start to develop research and start to work on your own robotic systems, you don't have to start from scratch to cobble up all of the software and computing systems and the sensors and all the mechanical systems. That alone would take several years. And so we decided to build a reference robot. This is the Nvidia Agile One in partnership with Agile Robotics. This robot is going,
1:09:57[applause] this reference robot will be available in the first quarter of next year and it will come pre-assembled with computer inside software integrated and it should be able to perform quite well right out of the bat. And so this is a great starting point and it'll be the flywheel by which we can accelerate development of robotics.
1:10:21Well, let me start let me end today's speech from where we started. Nvidia started in the PC revolution. It was in 1993 and the PC revolution was just about to start. Remember Windows 95 kicked it into turbocharge. We came since then invented the GPU programmable shading, real time, real-time ray tracing. We added CUDA to the computing model and helped ignite the modern AI revolution.
1:10:51And so four four years ago, we started to think what can we do to re-imagine the personal computer for this new era where the personal computer is not just your tool, your personal computer tool, but your personal agent or your personal assistant. And the way you interact with this computer would be essentially a companion, somebody who helps you do work. And so
1:11:20this computer needs to have exactly the five things that I mentioned today. It needs to have model. It could be running locally or it could be running in the cloud. It needs a harness. It needs a secure runtime open shell. And it needs access because it's available on your PC, it needs access to knowledge and tools. And the good news is because Nvidia has been accelerating tools for so long. If we could just create a
1:11:49computer that can accelerate OpenGL and DirectX and everything CUDA, everything Nvidia has ever worked on would run on this computer. And so this is it. After four years of partnership with Microsoft, we have created the RTX Spark and this is a brand new line of computers. This every literally thousands of years, thousands of engineering years went into this. We've been secretly working on it for four years. Created a chip called the N1X and
1:12:18then we just recently announced it with the entire PC industry joining forces. We created desktops, laptops, and workstations. This is one pedlop. One pedopl. Just to put that in perspective, in the year 2016, I delivered one pedlop and 128 gigabytes of memory. One pedlops and 128 gigabytes to a small company called OpenAI. This is one pedlop and 128 gigabytes.
1:12:52[applause] The the only difference of course is that I can hold this like this and DGX1 took two people to carry upstairs. And so this is the brand new RTX Spark PC.
1:13:10And if you would like to have a companion workstation that runs Katilla and everything do so and everything Autodesk and a you know Adobe and Cadence and Synopsis, you can always get yourself a DGX station that runs Windows and this is 20 pedlops and 754 gigabytes of memory. And so incredible.
1:13:36And of course there's also the desktop version. All of these are now uh these are now available and I think uh the DGX station with RTX spark uh with uh DGX spark DGX station with Windows will be available at the end of this year. Now the question is so far I just talked about the five components of an agent and now the question is how do you build it? How do you build an agent that runs let's say on one of those computers or runs in a
1:14:04cloud or runs on prem somewhere? Well, the answer is it's really easy and you do it this way. The way you build an agent is you ask an agent to build it for you.
1:14:19[applause] So, this this is the prompt that is going to run the next application. And what what you're about to see is this. You're about to see a robotics application. And while you're watching this demo, just remember there's an agent running on this personal computer. This demo runs across three computers.
1:14:47The first computer is to train. The second is a digital twin and the third is to put that AI into the computer itself. Okay. So two very two ideas. The first idea is agents. How do you build it and run it and what you can use it to do and the second is physical AI computing model which consists of three computers. Let's take a look. [music] Meet Micro Duck hugging faces robot.
1:15:28Let's teach it to see and walk with NVIDIA RTX Spark. I use Hermes as my agent harness, Neotron and OpenAI's GPT6 Astra as my models and [music] CUDA X accelerated tools Frecad Blender and Isaac SIM, Newton Physics Engine and Isaac Lab. I'll run everything on my RTX Spark except when I need to burst [music] larger workloads to cloud. First, I describe the course and what my MicroDuck needs to learn.
1:15:59On RTX [music] Spark, Hermes and Neotron plan the course, listing every object required. Astra takes it from there using FreeCAD to model 3D objects, Blender to apply [music] materials, and Omniverse sim ready libraries to add physical properties. In Isaac Sim, Astra composes the scene. Astra rigs Microduct's joints, actuators, and links, and simulates how Micro moves with Newton physics engine. [music] Now training begins.
1:16:46In the cloud, thousands of micro ducks train in parallel in Isaac lab. Every fall a lesson Improving the policy every time. Omniverse [music] libraries simulate MicroU's cameras, teaching it to see. With perception and physics working together, my Micro can avoid obstacles, navigate tight spaces, and adapt to its surroundings. I download the policy [music] to RTX Spark and use Neotron and Isaac SIM to evaluate my micro duct with multi-root orchestration.
1:17:40I deploy the validated policy to the real duck and he's on his way. NVIDIA RTX [music] Spark. Build and run AI agents and their accelerated tools right on your laptop.
1:17:56[applause] Hey guys.
1:18:08Oh yeah. Okay. Come on. Here. Come over here. Show people you can walk. And you have a camera.
1:18:27Can you guys sit down?
1:18:36[applause] Okay, stand up. Take a nap.
1:18:44You need to take a nap.
1:18:50Did you know that you were you were created in here? You were created on RTX Spark. I know. And trained you in the cloud on a whole bunch of Nvidia supercomputers and you learned your skills right here on RTX Spark in a digital twin, the Omniverse digital twin. Isn't that incredible?
1:19:11I know. You guys are so cute. And and uh your your uh your parents are called Pollen. P O L L E N. Is that right?
1:19:25Uh-huh. Pollen Robotics.
1:19:28And apparently everybody loves you guys so much. You're sold out. I know. So if you go to the website, everybody, everybody watching this, you know, there's going to be like 20 million viewers. Everybody watching this, go to the pollen website and place an order. The I think this startup company is like 10 people. They might blow up. Poof. It's going to be so exciting. Everybody should have one. You guys are so cute.
1:19:57And so we'll just have an RTX Spark and we'll teach you new skills. Isn't that right? Oi oi. Oh, look at that. That's incredible.
1:20:07Can you do a somersault?
1:20:17[laughter] That's so good. That's so good.
1:20:25Okay. Well, guess what? Okay. So, today, what did we talk about? We talked about Europe building its AI economy and it starts with infrastructure and we announced a brand new brand new system called DGX pod. It makes it easy for you to build AI factories and these AI factories don't just run models anymore. Pay attention. Stay focused. Come here. Come here. Come over here. Stay focused. And these AI factories don't just run. There you go.
1:21:00They they don't just run AI models anymore. They run entire agents. And these agents are consistent of five things. There's the large language model that does the perception, reasoning, planning, and acting. And it runs in conjunction with a harness that turns it. It's essentially exoskeleton. runs that harness sits inside runs inside a secure runtime we call open shell which has just so many people that have supported it and you
1:21:30give it access to knowledge and tools and this agent runs on top of these computers but we're also extending into the next generation which is where you guys come in physical AI so that's it that's what we talked about everybody can build agents every company in Europe will come will become an AI company.
1:21:54There's no question everybody will become an AI company and we're going to have a thriving AI economy here in Europe. So after 33 years, what an incredible journey and we made a really cool video. You guys want to watch it together? Okay, come here. Come over here. Just come over here just a little bit. We'll get out of the way.
1:22:17Top speed. Top speed.
1:22:20[cheering] How the heck are we supposed to follow that?
1:22:26Well, there is one thing that can make this crowd go even crazier. [cheering] What is it?
1:22:37You won't believe this. GTC has just gone full Euro Vision. [music] [singing] AI's next chapter begins. Agents working by your side. Europe's moment has arrived. But if you miss [music] what we said at GTC, we'll break it down for you.
1:23:03[music] Oh, this is a story all about how the world's building new infra now. First power plants let energy scale and cave data [music] centers. Now we've got mail. Now we're way past turning coal into wi [music] factories turn sand into move oil the S&P [music] to something more investable. Our GPU it works like an asset [music] producing green. Looks like a chip like a cash machine. Make
1:23:33revenue while you relax. The cash flows smoother than [music] [music] agents code and make decisions. Give them freedom but control their missions. [music] Open shell and bearers checking the list. No credentials.
1:24:02Access dismissed. Sentry on [music] blue field. Never clock. Protecting the sandbox like hassle. We advance AI [music] but limit risks. We're responsibly optimistic. RTX Spark keeps agent safe and bright. Our spark rates keep the price just [music] right. We do quantum Rexis gaming allinone cuz you can't spell fun without fun.
1:24:29Got [music] a body now. It's learning the rules of gravity, friction, and using real tools. With Isaac, Lab, and Cosmos, they train on their own. It's like college for [music] robots without student loans. DSX pods get AI in sync. Less power, more brains like Legos that think.
1:24:46QVS, [music] Neatron, and Kumo are here throwing knowledge faster than a cashier. Europe's full stack, it's all on its checklist. AI's [music] five layer cake, it eats it for breakfast. The Euro ecosystem is leading the race in love's open source. Just ask Hug and F. The continent is rising and the best is yet to [music] come. Come on all of Europe. Let's sing as well. We [music] are Europe. some robot [music] robot robot robot.
1:25:30We [music] are your We [music] are Europe. You and me. [music] Welcome all to Jesus. [music] [applause] Welcome to GTC.
1:26:04I want to thank all of you. I want to thank all of you. All the sponsors. You can see all the names of companies that are here. all the partners we've been working with for 33 years. Micro duck, my new friends. Um, let's work together to build Europe's AI economy and put AI to work. All right, give everybody. Have a great GTC.
1:26:31[applause] Hey. Yeah.
