The Industrial AI Revolution: Siemens Keynote at CES 2026

[00:00:00] Welcome to the CES stage. [00:00:10] I want you to think back to the time before electricity. [00:00:18] The world moved at the pace of people. [00:00:24] Horses bridged our distances. [00:00:29] Steam powered our machines. [00:00:33] And ideas moved only as fast as a letter or a human voice. [00:00:41] Then electricity arrived. [00:00:45] This channel purpose technology became the foundation of modern life. [00:00:51] It turned night into day, amplified human capabilities, boosted productivity and powered progress that once felt out of reach. [00:01:03] At Siemens, we brought this world to life. [00:01:09] Pioneering telecommunications and electric grids. [00:01:15] Moving the machines that accelerated mass production. [00:01:19] Moving people and goods via electric trains. [00:01:23] Scaling the benefits of electricity all around the world. [00:01:29] Fast forward. [00:01:31] One and a half centuries later, another channel purpose technology has arrived. [00:01:37] And this time, it's not about energy. [00:01:41] It's about intelligence. [00:01:43] Artificial intelligence will be as transformative for this century as electricity was for the last. [00:01:53] It's redefining how we design and manufacture products. [00:01:55] From the phones in your hands to the cars you drive. [00:01:59] It's revolutionizing how we build and operate infrastructure, including buildings just like this one. [00:02:07] And it's allowing us to make entire systems, grid cities, economies more adaptive and more efficient. [00:02:17] A century ago, Siemens helped build the world in the light of electricity. [00:02:25] Now, we get to do it again in the age of intelligence. [00:02:31] We are powering the industrial AI revolution. [00:02:36] It has already begun. [00:02:38] And it's picking up steam faster than steam ever did. [00:02:42] In fact, steam took 60 years to transform society. [00:02:47] Electricity, 30. [00:02:49] Computers, 15. [00:02:51] For AI, we're looking at seven years or less before intelligence is embedded in the system we rely on every day. [00:03:03] And that changes something fundamentally. [00:03:06] Because when AI enters physical system, it stops being a feature. [00:03:15] It becomes a force. [00:03:16] A force with direct real world impact. [00:03:20] A force that transforms how we design and build. [00:03:24] How factories produce. [00:03:26] How infrastructure operates. [00:03:28] How the world powers itself. [00:03:31] Now, how do we do it? [00:03:34] How can we make this happen? [00:03:37] How can we do all of this at speed and scale? [00:03:41] And at the same time, in a way that it's reliable and safe. [00:03:47] And hallucination is not acceptable when AI is deployed in the industrial world. [00:03:54] Today, we are making it easy for you to scale industrial AI in the real world systems to create real world impact. [00:04:03] We are bringing together AI-powered technologies, industrial domain know-how, and the right partners. [00:04:13] And the whole thing is connected by data, valuable industrial data. [00:04:20] And all in one place, in our Siemens accelerator marketplace. [00:04:25] So let's have a look at these elements one by one. [00:04:29] And let's start first with the right technologies. [00:04:32] It is about software, hardware, compute, highly performant compute with GPUs. [00:04:41] And of course, it is about data. [00:04:44] Here comes the good news. [00:04:47] Many companies have already some important elements in place, but this does not make them AI ready yet. [00:04:57] Digital twins can simulate and software can solve problems, but they do not recommend what to do next in the real world. [00:05:06] Compute is CPU based with AI needs GPUs to deploy its full power and speed. [00:05:17] And too often data is fragmented, trapped in silos and unused. [00:05:22] That's why it is so important to have an end-to-end AI industrial stack to make impact in the real world. [00:05:33] And at Siemens, we are able to build this because we have been working with AI for industry for more than 50 years. [00:05:42] And today, Siemens has more than 1,500 AI experts and all our engineers live in both worlds, in the digital and in the real world. [00:05:52] And all our colleagues, over 250,000 of them, bring deep domain know-how in our 30 industrial verticals. [00:06:02] And that's because, for decades, Siemens has been helping customers in the United States and all over the world to automate manufacturing, [00:06:13] to run railway networks, to design and operate grids, or to operate buildings more efficiently. [00:06:20] In fact, today, in one out of three manufacturing machines worldwide, runs a Siemens controller. [00:06:31] With our software, you can build the most comprehensive digital twin of products and, of course, of a whole manufacturing site. [00:06:40] And now, we are adding AI. [00:06:45] Decades of experience in software, hardware, and AI. [00:06:49] And this experience means we have the domain know-how to identify which data matter and how to cluster it. [00:06:59] Which AI applications make sense and which decisions should be left to humans. [00:07:06] And finally, our partners. [00:07:10] For industrial AI to make use of all of this industrial design, manufacturing, and operating data, [00:07:16] so that you need a huge amount of computer, gigantic centralized AI factors that are run by our partners like AWS or Microsoft. [00:07:28] But also at the edge, very close to your machines, to your infrastructure. [00:07:35] With this massive GPU process, powered compute, digital twins no longer explore just thousands of options, but hundreds of thousands of them. [00:07:48] Complex industrial foundation models can get trained on unprecedented amounts of industrial data. [00:07:57] And a digital twin can control a whole plant in real time. [00:08:02] When these things come together, the right technologies, industry domain know-how, and the right partners. [00:08:10] That's when we stop just reporting issues and start anticipating them. [00:08:16] That's when humans stop reacting to errors because machines have started to act and adjust autonomously. [00:08:23] That's when companies can turn ideas into real-world impact with speed, quality, and efficiency. [00:08:33] It's the industrial AI revolution. [00:08:40] Now, I mentioned partners, the right partners, the best partners, including a very special friend of mine. [00:08:50] And we hit it off immediately. [00:08:52] He has a visionary-engineering mindset. [00:08:56] He thinks through problems from first principles. [00:09:00] He anticipates the next development step of AI technology. [00:09:05] And he has both bold vision and the ability to make things happen. [00:09:13] Please welcome Chancellor Nguyen. [00:09:17] Thank you. [00:09:18] Thank you. [00:09:18] Thank you. [00:09:19] Thank you. [00:09:20] Thank you. [00:09:21] Thank you. [00:09:22] Thank you. [00:09:23] Thank you. [00:09:24] Thank you. [00:09:25] Thank you. [00:09:26] Thank you. [00:09:27] Roland. [00:09:28] Good to have you. [00:09:29] Good to have you. [00:09:30] Happy New Year. [00:09:31] Happy New Year. [00:09:32] Happy New Year, everybody. [00:09:33] Happy New Year. [00:09:34] Please. [00:09:35] 200 years old. [00:09:40] You've seen a few of these industrial revolutions, actually. [00:09:44] We did. [00:09:45] Not quite 200 years, but we're heading for it. [00:09:47] 175. [00:09:48] Exactly. [00:09:49] Exactly. [00:09:50] And we came a long way with our partnership too, right? [00:09:53] I mean, we kicked it off a couple of years ago. [00:09:56] I remember that very well. [00:09:57] We have the same mission, industrial AI operating system. [00:10:00] That's what we want to build. [00:10:02] Huge opportunities for our customers. [00:10:04] So, what's with you? [00:10:06] Where are we on this journey now? [00:10:07] Yeah. [00:10:08] It's an incredible thing. [00:10:09] So, we did a press conference at Siemens headquarters. [00:10:14] In Munich. [00:10:15] Yeah. [00:10:16] How many years ago was that? [00:10:17] Three? [00:10:18] It was only... [00:10:19] 2022. [00:10:20] Okay. [00:10:21] Yeah. [00:10:22] It was three years ago. [00:10:23] And we spoke about this moment, in fact. [00:10:24] And Siemens is unquestionably at the foundation of every industry that we participate in, that the [00:10:33] the bills on. [00:10:34] And we had a vision that Siemens, and this was your vision, to turn Siemens into a software [00:10:41] defined company. [00:10:42] In a lot of ways, described it in a way very similar to a computer company. [00:10:46] Right. [00:10:47] And it would, of course, have the underlying infrastructure of computing, but mostly on [00:10:53] top, your infrastructure software, artificial intelligence, and it's going to revolutionize [00:10:58] the way things are designed, simulated, planned, and operated. [00:11:03] And that was the vision you described, and we spoke about three years ago, and here we [00:11:08] are. [00:11:09] You know, here we are at the beginning of that journey, and all the pieces are now finally [00:11:13] coming together. [00:11:14] It seems to. [00:11:15] And talking about what you achieved, I brought an example. [00:11:18] You know that. [00:11:19] HD Hyundai. [00:11:20] I mean, they built ships, huge ships, and shipyards, by the way. [00:11:25] And they use our technology. [00:11:27] That's right. [00:11:28] They're sitting on NVIDIA technology. [00:11:29] This is a digital twin of an entire ship. [00:11:31] Every nut and bolt is in there. [00:11:34] It's incredible. [00:11:35] I mean, this is the full CAD of the ship. [00:11:38] Visited. [00:11:39] The perfect, precise digital twin. [00:11:41] And you see also the people working on the ships. [00:11:45] You assimilate everything. [00:11:46] By the way, these ships are, they look alike. [00:11:48] So the body is the same, but each one of them is individual. [00:11:52] So you really have to be very, very clear how you design it. [00:11:54] You optimize in the digital world, and then you build it. [00:11:58] Not before that. [00:11:59] And that's where our whole stack comes together. [00:12:02] Complementary technology for the photo realistic representation. [00:12:07] We use omniverse. [00:12:08] And this is really quite a perfect, yeah. [00:12:10] This is quite a perfect example of the type of work that we do together [00:12:14] in this realization of this digital twin idea. [00:12:18] That you would design every aspect of engineering. [00:12:21] It's not just the CAD, but the computing, the electronics. [00:12:26] All of it would be integrated and built into a digital twin. [00:12:32] It would, of course, run all of the software in the digital twin. [00:12:36] And in the future, I hope, this ship, this digital twin of a ship, [00:12:40] will actually put it in the ocean, a virtual simulation of the ocean, [00:12:45] and see it completely operate. [00:12:47] Exactly. [00:12:48] Yeah. [00:12:49] And you have only one option to get it onto water. [00:12:51] If that sense, you have a problem, by the way. [00:12:53] So, well, we talked about bringing our partnership to the next level. [00:12:59] And the whole idea is, obviously, to take these examples [00:13:02] and scale them even faster. [00:13:04] And, of course, to make it AI-enabled. [00:13:07] I'm talking about industrial AI, which really hits the real world. [00:13:11] And we picked five areas of collaboration where we want to intensify. [00:13:15] And that is good news for our customers, too, I guess. [00:13:18] Let's start with the first one, take them one by one. [00:13:21] The first one is AI-native chip design. [00:13:25] So, if it comes to chip design, I don't need to tell you something. [00:13:28] Obviously, with each next generation, you're getting closer to the physical limits. [00:13:33] So, tell me what the challenges are and what can we do? [00:13:36] Well, let's use our latest generation that we just announced yesterday as an example. [00:13:41] Congratulations. [00:13:42] So, Vera Rubin, in order to build the latest generation GPU, [00:13:46] people think that our GPU is a chip. [00:13:48] It is a chip. [00:13:49] It consists of a chip. [00:13:51] But, ultimately, in order for these GPUs to scale up, [00:13:54] to process the level of AI that's necessary at the frontier, [00:13:57] these GPUs are essentially one giant rack. [00:14:01] And this entire rack is, in the case of Vera Rubin, 240 kilowatts. [00:14:07] It consists of, altogether, 220 trillion transistors. [00:14:12] We had to design six different unique chips. [00:14:15] CPU, GPU, networking, multiple types of networking switches. [00:14:20] One is for scale up. [00:14:21] One is for scale out. [00:14:23] And a smart NIC and a data processor for storage, for AI memory, basically. [00:14:29] All of this stuff came together. [00:14:31] It's 220 trillion transistors. [00:14:33] It's two tons. [00:14:35] And 150,000 engineering years came together to build this one system that we announced yesterday. [00:14:43] And in order to do this perfectly at a rhythm that our company runs at, [00:14:47] the level of chip design, system design, system integration, thermals, electricals, [00:14:54] all of that, in a lot of ways, our GPUs is a little bit like that HD Hyundai ship. [00:15:00] Yes. [00:15:01] And what we are hoping for, and the reason why we're partnering so closely together, [00:15:06] is so that we could build that Vera Rubin in the future as a digital twin. [00:15:12] The entire system as a digital twin, not just the chips, but the systems, the chips, everything together, cooling, thermals, [00:15:19] and run the thermal simulation directly through it as it is a digital twin. [00:15:24] And that's our vision. [00:15:25] And so we're accelerating the first part of our project together, our partnership, [00:15:30] is to accelerate with Siemens EDA, Siemens SimCenter, accelerate everything we can with GPUs. [00:15:38] And this way, we could scale up the simulations that we do and achieve the virtual digital twin that we hope for. [00:15:45] So what we're going to do is we will use Yakuda software to rewrite our EDA software so it can take advantage of GPUs. [00:15:54] That's right. [00:15:55] So working with GPUs and designing GPUs. [00:15:57] Imagine EDA software being a hundred times faster. [00:16:00] Exactly. [00:16:01] Or being able to scale up a million times more. [00:16:03] Right? [00:16:04] That's our hope and dreams. [00:16:05] But we don't stop there. [00:16:07] Still, we talk about AI native chip design. [00:16:11] What we want to do now, we want to train our technology, our models on a massive amount of data, [00:16:17] so that it's not only about validation of design, but really making a proposal of new designs in EDA. [00:16:23] This is breakthrough. [00:16:24] Exactly. [00:16:25] It's never done before. [00:16:26] That's what we are heading for. [00:16:28] The goal of an engineer is not to write Verilog. [00:16:32] That's not the goal of the engineer. [00:16:33] No. [00:16:34] The goal of an engineer, of course, is to solve problems, imagine great solutions. [00:16:38] And it would be incredible one day. [00:16:40] That would be agentic Siemens EDA designers that are sitting with our designers. [00:16:46] Exactly. [00:16:47] And they're exploring ideas together, trying iterations and exploring boundaries. [00:16:53] Number two, we talk about, same thing here, AI native simulation. [00:16:58] So, when it comes to simulation, obviously, this is a lot of number crunching. [00:17:03] Yeah. [00:17:04] And it very often still runs on CPUs. [00:17:06] This is the first step. [00:17:07] We want to bring this technology, the heavy number crunching elements of our simulation to GPUs. [00:17:13] That, again, accelerates six times, a hundred times, a thousand times. [00:17:18] And if you can, I mean, there's an example. [00:17:21] This is a car simulation, by the way. [00:17:23] It goes back to when we started off. [00:17:25] Our partnership is from BMW, where we simulate the airflow, the aerodynamics. [00:17:30] And it's a lot of consumption and it takes a lot of time. [00:17:34] So, if you speed it up a thousand times, you can do much, much more iterations. [00:17:38] Again, we do not stop there. [00:17:41] We want to have simulation not only to validate, but to create. [00:17:45] That's right. [00:17:46] Again, you need a lot of data. [00:17:47] You need our technologies. [00:17:49] And that's where NVIDIA has all. [00:17:51] This is our technology running in real time. [00:17:53] Our technology together. [00:17:54] Isn't it? [00:17:55] You know, simulations. [00:17:56] Physics is beautiful. [00:17:57] I mean, that's just the bottom line. [00:17:59] Numbers are beautiful. [00:18:01] We do that. [00:18:02] Only an engineer would say that. [00:18:07] Numbers are just so beautiful. [00:18:08] Beautiful. [00:18:09] And we do that for trains. [00:18:12] And again, we are using Physics Nemo from NVIDIA and the full Siemens simulation stack. [00:18:19] Siemens has the world's leading suite for industrial simulations. [00:18:26] And SimCenter and all of the suites of solvers in SimCenter, we're going to accelerate completely using CUDA and CUDAX. [00:18:37] We are also going to work together to create AI physics, to teach AI the laws of physics, so that it could emulate physics. [00:18:47] Not to simulate physics from first principles, but emulate the outcome of physics, essentially guessing where, you know, the physical properties are going to go at the next second or the next nanosecond, the next microsecond, the next second. [00:18:59] And by doing so, we could speed up physics simulations, physics experimentations, if you will, 10,000 times, 100,000 times, and essentially create a real-life digital twin of a wind tunnel. [00:19:16] We essentially put a real design into it, right, real digital twin design into a digital twin of a wind tunnel and see it work. [00:19:24] And so the opportunity to do all of these experimentation in real-time is really exciting. [00:19:29] Which brings me, talking about physics, the real world, physical AI, which brings me to the third area where we step up, which is AI-driven adaptive manufacturing. [00:19:41] So, and this always reminds me when we talk about our technology stack, what we want to bring to the plants, which reminds me what you always say. [00:19:50] I mean, actually, a plant is almost like a robot, isn't it? [00:19:53] It's a big robot. [00:19:54] Exactly. [00:19:55] Or a car. [00:19:57] You know, if you look at a car today, it's sensors connected to a computer. [00:20:04] That computer is running software, software-defined. [00:20:07] And today, the software that we run on top of it is largely AI. [00:20:11] Well, that car is going to be made inside a factory that has sensors running computers. [00:20:20] Siemens software stack and operating systems. [00:20:23] And all the AIs that you'll have running on top of that. [00:20:26] In a lot of ways, that factory, the car is an inside-out AI. [00:20:33] The factory is an outside-in AI. [00:20:35] And it's going to look at the cars that it's going to build and it's going to build it robotically. [00:20:40] So the entire factory is going to be one giant robot. [00:20:44] It's going to orchestrate a whole bunch of robots that themselves are robotic. [00:20:48] And they're going to make products like cars that are software-defined and robots. [00:20:53] So it's going to be robots orchestrating robots, building robots in the future. [00:20:59] And so this level of integration of technology has never been achieved before. [00:21:05] Exactly. [00:21:06] And so it's really the reason why we have to build the foundation of future manufacturing from the ground up. [00:21:12] And nobody could do it better than Siemens because you're already the operating system of manufacturing plants all over the world. [00:21:18] And so from that baseline, we're going to, number one, make it software-defined and then two, make it AI-driven. [00:21:25] And all of that, you know, it's so great to see you drive that vision and for Siemens to really make that possible for the world. [00:21:33] And we're going to use it ourselves. [00:21:35] Exactly. [00:21:36] We're going to start in Germany in 2026, a first fully AI-driven adaptive manufacturing site. [00:21:43] And we bring an AI brain on top of our software-defined automation and our operations software, which basically consumes that, what a digital twin produces. [00:21:54] And when it comes to beauty, we do that in real time. [00:21:57] That's right. [00:21:58] In real time. [00:21:59] So you can really control the manufacturing site. [00:22:01] So our partnership, we work together and run Siemens factories. [00:22:05] Meanwhile, we have our partner, Foxconn, working with us in the United States to run the factories [00:22:12] for building AI supercomputers. [00:22:14] And so this partnership is really incredible. [00:22:16] We've talked about design. [00:22:17] We've talked about simulation. [00:22:18] Now we're talking about manufacturing. [00:22:20] All powered by GPUs. [00:22:22] All powered by CUDA. [00:22:23] All powered by Siemens. [00:22:25] And, you know, all powered by AI. [00:22:27] And you see an example of how Foxconn is using the technology already. [00:22:32] And again, the AI layer on top, the AI brain, which connects to the real-world data and real-time data. [00:22:40] And it looks so real, but you know this is a digital twin. [00:22:43] It's a digital twin. [00:22:44] This looks so incredibly real. [00:22:46] And that's really the whole purpose of the digital twin. [00:22:49] It has to, in every physically important way and operationally important way, it has to be a perfect representation. [00:22:57] And so the ideal is that the digital twin and the real version of it, there is no ability for the computer to know the difference. [00:23:07] You know, it's to the extent that our AIs can't tell the difference between whether they're inside a digital twin or inside the physical world. [00:23:15] That's our goal. [00:23:16] Exactly. [00:23:17] Yeah. [00:23:18] There are other factories too. [00:23:20] They produce intelligence. [00:23:22] We call it AI factories. [00:23:24] So that's the next area of collaboration. [00:23:26] And I know that when you talk about Vera Rubin, you need obviously new AI factories. [00:23:33] They have to be signed differently. [00:23:35] And you're building one on your own as we speak. [00:23:38] Yeah. [00:23:39] So we might want to share about the challenges, what needs to be different. [00:23:42] Well, if you take a look at the reason why I don't call it a data center, a data center is a place where people store data. [00:23:50] These plants, in the case of a one gigawatt data center or one gigawatt AI factory, it's $50 billion. [00:24:02] The investment level is unheard of of any type of factory the world's ever built. [00:24:08] The amount of technology inside is incredible. [00:24:10] And so if you're building something that costs $50 billion, you want to make sure there is absolutely zero schedule delay. [00:24:18] You're not going to be able to tolerate design changes. [00:24:21] You're going to have to make sure that we create the digital twin. [00:24:24] We plan it up front. [00:24:25] We simulate everything up front. [00:24:27] We're pushing the power. [00:24:30] We're putting electricity. [00:24:31] We're going to push cooling all the way to the limit. [00:24:33] And so if we don't simulate this entire digital, this factory in a digital twin, we have no chance of success. [00:24:40] And then, of course, once you operate it, managing all of these computers, these AI supercomputers, the networking, the storage, and running it all at the precision we need to run it at, the control of it is incredible. [00:24:54] So you're controlling, of course, the management of the systems. [00:24:57] You're controlling the power. [00:24:58] You're controlling the cooling. [00:25:00] And this entire system is essentially one giant factory and is running all the time. [00:25:05] And so when you put that much capital at work, you better make sure that the uptime is perfect and always, and that you're going to make sure that your scheduled delay is basically zero. [00:25:18] And without us working together and building these together in a digital twin, the chance of success is extremely low. [00:25:26] Exactly. [00:25:27] We can build them faster. [00:25:28] This is no different. [00:25:29] You know, Roland, this is no different than NVIDIA today. [00:25:31] The idea that we would design chips without software is completely illogical. [00:25:36] Exactly. [00:25:37] The idea that you would design these electronic systems without design software and verification software and simulation software, come on, forget it. [00:25:44] And so in the future, the idea that we would build plants, manufacturing plants, whether it's a car plant or a chip plant or an AI factory plant, the idea that we would build any of them without first doing it in the digital twin is completely inconceivable. [00:26:02] And the idea that we would be able to operate these incredibly complex systems without artificial intelligence, completely inconceivable. [00:26:09] And so I think that we know that future is here and we know that the stakes are incredibly high. [00:26:16] And that's the reason why our partnership with Siemens is so important to me. [00:26:20] We built this blueprint designs for AI factories together. [00:26:23] We have automation technology, which is much faster. [00:26:26] It's industrial automation technology because the demand pattern of AI factories is so fluctuating. [00:26:32] We know how to connect them to grids because we know how to operate grids. [00:26:36] You've been doing it for 175 years. [00:26:38] Exactly. [00:26:39] We do that. [00:26:40] We know how to do it. [00:26:41] After 175 years, you're going to get good at something. [00:26:44] Well, and what we also do for more than 175 years is using our own technology. [00:26:49] That's right. [00:26:50] Which is the last area we want. [00:26:52] I mean, actually, the people said, don't use the saying, we drink our own champagne. [00:26:57] So now we modified it. [00:26:58] We drink our own beer or whatever. [00:27:00] That's right. [00:27:01] German company. [00:27:02] But we actually, we use each other technologies. [00:27:04] That's right. [00:27:05] We look deeper into it. [00:27:06] What can we do more and better? [00:27:07] And it starts, for example, with the plant in Germany I was talking about. [00:27:12] We talk about EDA software simulation and bringing that to the next level to speed up each other's development. [00:27:19] It's incredible. [00:27:20] We're going to work together to accelerate all of your EDA software, which we can then use to design our chips faster so that we could accelerate your EDA software faster. [00:27:29] We're going to accelerate all of your SimCenter software so that we could use it to design and simulate our AI factories faster so that we could create AIs that makes SimCenter faster. [00:27:44] And if you look at this, we're going to use Siemens so that we could automate our factories to build even more amazing supercomputers so that we could create AIs to make your AI factories even more productive. [00:28:01] Doesn't that sound like the perfect partnership of two companies coming together? [00:28:07] And that's why whenever Roland reaches out and he needs a bug fix, I fix it right away. [00:28:21] Because I fix his bug, which directly comes back to my company, and my engineers go faster. [00:28:26] And I tell you, this works in both ways. [00:28:28] They're scared when you have a call. [00:28:31] So, Jensen, thank you very much for the partnership. [00:28:34] Thank you. [00:28:35] We are at the beginning of a new industrial revolution. [00:28:39] And what perfect company for all of that capability to come together than the company at the center of every industrial revolution. [00:28:49] And so, it's a great privilege to partner with Siemens, and it's incredible to be friends with you and to see your vision of Siemens come together. [00:29:00] And for our journey of reinventing industry to achieve this milestone is incredible. [00:29:06] So, thank you. [00:29:07] Siemens and NVIDIA building the industrial revolution system. [00:29:09] Thank you. [00:29:10] Thank you. [00:29:11] Thank you. [00:29:11] Thank you. [00:29:12] Thank you. [00:29:13] Thank you. [00:29:14] Thank you. [00:29:15] Thank you. [00:29:16] Thank you. [00:29:17] Thank you. [00:29:18] Well, when I last spoke at CES, I shared a vision of the industrial metaverse, a vision for a virtual world that helps us make the real world better. [00:29:32] Thank you. [00:29:33] Today, we are making this vision a reality. [00:29:36] We are excited to launch our digital twin composer. [00:29:41] With the digital twin composer, you can create a virtual 3D model of any product, any plant, any process. [00:29:49] Bring it to life in a virtual realistic scene that you built and in real-time. [00:29:56] The digital twin composer is so powerful because you can connect the digital twin to the real world and to real-time data. [00:30:06] From engineering data to weather data, even time series data coming from your machines. [00:30:14] And when something worked well or something went wrong, you can go back in time and you find out why. [00:30:20] Or you can jump to the future to simulate designs before you build in the real world. [00:30:26] And if you want to unlock the full benefits of the digital twin composer, you connect it to our operations software and our hardware stack. [00:30:38] That way, you can make real-world changes from the virtual environment, optimize machine speed, temperature, pressure, or any parameter that determines the quality and the yield of your plant. [00:30:52] This is how we combine the real and the digital worlds. [00:30:57] This digital twin composer will be available on our Siemens Accelerator Marketplace in just a few moments. [00:31:08] PepsiCo. [00:31:10] PepsiCo is among the first companies to use it. [00:31:14] And this brings me to our next guest, who is leading the way when it comes to the digitalization of PepsiCo. [00:31:21] PepsiCo. [00:31:22] Making the companies smarter, more efficient, more sustainable, with industrial AI. [00:31:30] Please welcome Athena Kanyora. [00:31:33] Happy New Year. Happy New Year to everyone. [00:31:36] It's a great year. New Year, great year. [00:31:40] It is, it is, it is. So, tell us about the challenge PepsiCo is navigating through and these challenges are very much alike to what the whole food and beverage industry has. [00:32:05] Yeah, of course. And as Ronald said, I'm very privileged to be on stage with Roland. [00:32:10] You know, PepsiCo has been pioneering some of this innovation with you and Nvidia. [00:32:15] So, but from farm to shelf, PepsiCo has been operating in a very extraordinary and very complex environment. [00:32:22] Because we have operations everywhere. We serve billions of consumers. We have billions of touch points per day. [00:32:29] So, consumers expect the products to be everywhere, every time, in real time. And we want to be able to fulfill the supply. [00:32:39] So, imagine a world where you have demand spikes, where you have unusual events, when you have tornadoes, you have to be able to fulfill this demand in a very digital way where the physical footprint cannot be a showstopper. [00:32:54] Right. So, talking about the supply chain and delivering, let's take a look at your existing warehouses. [00:33:02] Yeah. So, I mean, many of you know, especially the ones that come from the manufacturing world, we have great warehouses that are fully modernized, but we also have warehouses that are very old. [00:33:14] So, in this specific case, this warehouse is half a century old. And that means in the current state, it really cannot serve and deliver the demand that might change depending on the events. [00:33:28] So, what is the answer? And the answer cannot be, I will just build more capacity. [00:33:34] The answer must be, I need to find a way to maximize the capacity that I have to drive more operational efficiency and therefore increase the throughput. [00:33:43] So, what we wanted to do is adopt a digital first design approach. And that means leverage digital twins and adopting AI at scale in all the parts of the operations where we were able to co-design and simulate and optimize the layout before any physical build begins. [00:34:04] So, and to do that, you decided to use our digital twin composer. Share your experience with us. [00:34:11] Yes. First of all, thank you. Thank you for the partnership. Thank you for the investment. Thank you for taking our feedback in the product innovation. [00:34:20] Today, we just had our press release. Yes. [00:34:22] We are very excited to help shape this technology in partnership with NVIDIA and, of course, Siemens. [00:34:29] This is the first of its kind for the CPG industry. It's how we build AI and digital twins into our operations across the board, globally. [00:34:38] When we say operations, that is manufacturing, warehousing, distribution centers, mixing centers, and every part of logistics. [00:34:47] So, the digital twin composer allows us to do everything in the virtual world, being able to simulate everything in a way where we don't have to spend a single dollar on the physical plan before we know the design is the final design. [00:35:02] So, therefore, we are able to pull now massive amounts of data that we are able to simulate in a unified immersive environment. [00:35:11] That means we can create a beautiful photorealistic way, which you saw before in the discussion. [00:35:17] And with the AI-powered simulation now, we can explore hundreds or thousands of potential layouts to find the most efficient options. [00:35:26] Therefore, with the digital twin, this type of task that could have normally taken months now takes days. [00:35:33] Well, and the question would be then, what does that do to your customers? [00:35:40] Yes. [00:35:41] Well, obviously for us, making sure we serve our customers in the most efficient way, where we provide the inventory and we have the intelligent network, [00:35:50] where both the plants and the warehouses and the network is able to anticipate the demand and fine-tune operations real-time is a must. [00:35:58] So, therefore, we have had significant impact even in the first three months. [00:36:02] So, one good example is in the Gatorade plant in the US. [00:36:06] We use the digital twin composer. [00:36:08] We were able to drive two very meaningful results. [00:36:11] One is to increase efficiency 20% just in three months. [00:36:15] Oh. [00:36:16] And obviously, if we were to look and we have estimated how these capabilities will benefit the company, [00:36:21] we estimate the capex reduction of 10% to 15% across all the operations. [00:36:26] So, for us, this is just the beginning. [00:36:28] We are very hopeful about what this technology can do when it comes to serving our customers, serving our consumers. [00:36:35] And I want really to thank you for the great partnership. [00:36:37] Thank you for the partnership. [00:36:38] And this is just the start. [00:36:40] We want to see more and we will see more. [00:36:42] Absolutely. [00:36:43] Athena, thank you very much. [00:36:44] Thank you so much. [00:36:45] Thank you. [00:36:53] You have seen a great impact the digital twin composer is having on PepsiCo's facilities. [00:36:59] Using this technology, even in the design phase of new plants, matters, especially in the United States, [00:37:06] because it accelerates the ramp up of new manufacturing sites and allows our customers to make them more productive and energy efficient. [00:37:14] Kion uses our technology to transform warehouses and entire supply chains too. [00:37:18] Let's have a look. [00:37:19] Kion, the supply chain solutions company, is scaling up physical AI and automation solutions to orchestrate customer supply chains. [00:37:28] Inside the four walls and up and down the supply chain with AI digital twins. [00:37:34] A new supply chain solution begins as a digital twin, compressing solutioning from years to months. [00:37:42] You can learn more about the digital twin. [00:37:43] You can learn more about the digital twin composer at our booth. [00:37:48] Also on the CES show floor when you will meet the team from PAVE 360. [00:37:54] PAVE 360. [00:37:55] Automotive is our digital twin for autonomous driving system and it's ready to use right off the shelf. [00:38:08] It replicates real vehicle hardware in the digital world. [00:38:13] Developers can simulate real world driving conditions. [00:38:18] They can test the autonomous system before a car even exists. [00:38:21] And this accelerates development. [00:38:24] PAVE 360 Automotive also replicates how the software in the car behaves. [00:38:32] And better yet, it integrates the latest automotive IP from our partner ARM for advanced driver assistance systems, infotainment and AI driven cockpit features. [00:38:45] It's a solution that will scale autonomous driving faster with the power of industrial AI. [00:38:53] Now imagine what happens when we apply industrial AI to something more personal, even more complex. [00:39:05] Your body, your health. [00:39:09] Life sciences. [00:39:12] Bringing new medicines to the market costs a lot of time and money. [00:39:17] Over the last 50 years, these costs have risen dramatically. [00:39:21] It often takes more than 10 years and costs are up to $2 billion. [00:39:28] Patients and pharmaceutical companies obviously want to have faster innovations. [00:39:34] That means tackling every step in the chain from early research to manufacturing. [00:39:41] And we bring AI to all these steps. [00:39:44] Let's take the example of a new cancer drug. [00:39:48] In research labs, scientists create billions of data points. [00:39:53] Today, this data is scattered across industries, instruments and files and around the world. [00:39:58] Through our platform, Luma, scientists use AI to bring this data together and structure it. [00:40:05] So you can ask questions in natural language. [00:40:09] Next step, scientists identify the most promising molecular structure of the cancer drug. [00:40:18] With AI-powered simulations, they can simulate the behavior of the molecule 2.5 million times more efficiently than ever before. [00:40:30] And this includes how molecules move and flex, how they interact with each other or how stable they are over time. [00:40:41] But then, then you need to go from making a small batch in the lab to producing at scale without the slightest deviation in the recipe. [00:40:52] And with exactly the same result. [00:40:56] Even for world-class production experts, this means a lot of trial and error. [00:41:01] A lot of experiments in a bioreactor. [00:41:04] In our digital twin of a bioreactor, you can run these experiments in the digital world first. [00:41:11] You simulate until you have the highest quality and the highest output. [00:41:14] And only then, and only then you start the production. [00:41:19] Siemens technology accelerates all steps from discovery to manufacturing. [00:41:26] Lifesaving therapies can make it to the market as much as 50% faster. [00:41:32] And at every stage, AI speeds things up. [00:41:38] This requires powerful AI infrastructure too. [00:41:43] The GPUs and the AI factories I talked about with Jensen. [00:41:49] But also the cloud infrastructure we get from partners like Microsoft. [00:41:55] On that, let's hear from my friend, Serena Tello. [00:42:02] Thanks so much, Roland. [00:42:03] It's fantastic to join all of you at CES. [00:42:06] In the industrial sector, we have an incredible opportunity to bring together digital intelligence [00:42:12] with physical operations to reinvent things that are designed, built, and run in this era of AI. [00:42:19] And we're already seeing the industrial AI drive better outcomes in both productivity as well as in safety. [00:42:25] That's why our partnership with Siemens is so important to us at Microsoft. [00:42:30] By combining your deep domain expertise with our trusted cloud and AI capabilities, [00:42:36] we are helping our customers tackle some of today's most pressing challenges. [00:42:40] From sustainable energy and transportation to smart cities and precision medicine. [00:42:45] I'm especially excited about the co-engineering work our teams are driving to build custom models and agents [00:42:52] that get trained with real-world industrial data. [00:42:55] And of course, we're just getting started. [00:42:57] I could not be more optimistic about what we do together and achieve together. [00:43:01] Thank you all so very much. [00:43:03] Thank you. [00:43:05] Thank you. [00:43:06] Thank you. [00:43:07] Thank you. [00:43:08] So, custom models, agents, and of course, co-pilots. [00:43:13] Let's talk more about it with Jay Parik. [00:43:19] Hi. [00:43:20] Hi. [00:43:21] Hi. [00:43:22] Good seeing you. [00:43:23] So, Jay, you are leading Microsoft's core AI team. [00:43:36] You put all AI experts together, which tells me, again, this is a foundational technology [00:43:41] which scales across the whole company. [00:43:43] So, where are you in the step of really making this AI journey, creating impact at your customers? [00:43:50] Yeah. [00:43:51] Well, first of all, thank you for the partnership. [00:43:53] Excited about what we're working on today and also going forward. [00:43:57] So, as what we're seeing today is really three waves with more waves coming in terms of AI. [00:44:02] The first wave was really folks using AI as a chatbot, right? [00:44:07] To ask it simple questions, maybe do code completions. [00:44:10] The second wave is much more focused on delegating a specific task to AI. [00:44:16] Maybe giving it a task to do and come up with a market strategy document or proposal. [00:44:22] To do these longer running, kind of more sophisticated tasks. [00:44:25] Now, what we're seeing emerging in this third wave is this set of agents working with and orchestrating [00:44:32] a set of different agents to be able to do asynchronous work, but much more complicated work, right? [00:44:39] Longer running tasks. [00:44:41] And it is this third wave that we're focused on building because you need multiple models. [00:44:47] You need to be able to train your own models, right? [00:44:49] You need to be able to bring in your enterprise data. [00:44:51] You need to be able to orchestrate these agents, but you also need to do it with security, with observability, with compliance, with all of that scale enterprise ready abstractions, right? [00:45:05] And it is this area in this third wave that I'm most excited for us to be working together because this third wave is how we bring this technology to specific use cases in industries. [00:45:17] But Roland, I'm really curious is like this third wave. [00:45:20] How does this show up as you lead industrials with AI? [00:45:24] How is this manifesting with your customers? [00:45:26] And I mean, there are many examples. [00:45:29] I brought one. [00:45:30] This is a very exciting one. [00:45:32] We talk about the Rolls-Royce. [00:45:33] They do turbines for airplanes. [00:45:36] And I mean, obviously, you want to have these turbines to work 100% reliable and of course, fuel efficient. [00:45:44] And within these turbines, they have a lot of design parts coming. [00:45:49] These are designed with Siemens software and technology from Microsoft. [00:45:53] And there's a particular one, which is a hydraulic pump. [00:45:56] And that makes this machine turning. [00:45:59] So this pump has now a digital twin of the pump itself, but it also has a digital twin of the machines machining it. [00:46:07] And we simulate the whole process in programming it. [00:46:11] And we use your technologies together with ours and simulate it with amazing outcomes. [00:46:17] These pumps are much stiffer and lighter. [00:46:20] But we use Microsoft as well here and AI technologies. [00:46:24] Yeah, and excited about the other things that we're working on with these joint AI co-pilots, right? [00:46:29] We're able to help reduce the programming time for CAM by 80%, drove up productivity in the factory by 30%. [00:46:38] And then also excited to work with your team to bring the GitHub platform to all of the developers to drive that innovation, that value for your customers faster. [00:46:47] Again, all built together on Microsoft's cloud and AI. [00:46:51] Exactly. [00:46:52] And there are many more examples. [00:46:54] The demand is huge and the benefits of not only shortening the development time, but also the production time is amazing. [00:47:03] We talked about technology, but we cannot stop there. [00:47:07] And we know that you're very passionate about also the other part, the cultural, the way how you transform. [00:47:12] Absolutely. [00:47:13] The people. [00:47:14] It's hard. [00:47:15] It's hard. [00:47:16] I know that. [00:47:17] Let's talk about it. [00:47:18] So, you know, I spent probably last year talking to about two or 300 customers. [00:47:24] And it was fascinating because somewhere between probably 70 and 80% of my conversations with customers started or ended or was dominated by this topic of change, about culture, about people, about organizations. [00:47:37] And it was such an interesting set of conversations. [00:47:40] And my kind of three current takeaways from those conversations were, one is like we're all learning together. [00:47:46] And we're sharing openly about what's happening in organizations, small startups to, you know, companies that are much, much, much bigger, multinational. [00:47:55] So this culture of change is something that we have to share openly and work together on. [00:47:59] The second thing that I noticed and comes up in conversations is actually many of the employees, many of the workers are using this technology in their personal lives, right? [00:48:09] They're using it. [00:48:10] They're actually really sophisticated with it for family, for personal growth, et cetera. [00:48:16] But there are institutional headwinds to adopting it, using it inside the company, whether it be compliance, whether it be risk, whether it be security, whether it be finance, whether it be just misaligned incentives, right? [00:48:29] And then, so it is our job to try to work and try to unlock this value of this creativity, this collaboration. [00:48:36] And then the third thing that I always encourage folks to do is really the potential intelligence of these models, the capabilities of the platforms that we're building here at Microsoft are so much more capable than what we realize today. [00:48:50] So my biggest encouragement to myself, to my teams, to our partners is to really raise our level of ambition of what we can do with AI to drive the world forward and to deliver more impact. [00:49:03] And I tell you, I can say from my perspective, we see a lot of these changes also in working with you, working with our customers, but also with you. [00:49:11] You come from a different environment, but we bring that together to our customers and they do appreciate it, I tell you. [00:49:16] Thank you very much for doing that. [00:49:18] And this is just the start. [00:49:19] Yes, thank you. [00:49:20] Thank you for the partnership. [00:49:21] Thank you. [00:49:22] So more powerful industrial co-pilots and now new co-pilots. [00:49:33] Today at CS, we are launching nine additional AI powered Siemens industrial co-pilots. [00:49:41] They bring intelligence along the entire industrial value chain, transforming how we design, engineer and operate. [00:49:51] And well, to help everyone access this kind of technology more easily, we are now putting it in the hands of our shop floor colleagues or to be more precise on their faces. [00:50:08] Together with Meta, we are making Ray-Ban Meta glasses for industrial AI. [00:50:15] Actually, we do not use the tainted ones, as you can imagine, maybe if the sun is shining, but they are also natural ones. [00:50:24] And they tell our colleagues to get real time audio guidance through the glasses, which button to press, which parameters to change. [00:50:33] Let's take a look at one of our factories when these classes will turn colleagues into connected experts so they can have more impact in the real world hands free. [00:50:46] Hi, I'm Sarah. [00:50:48] I just started working at this factory. [00:50:50] Honestly, there's a ton to learn. [00:50:53] It's great having AI from Siemens to help me get familiar with everything. [00:50:59] Good morning, Sarah. [00:51:00] Let's take a quick look at how the last shift went and what's coming up. [00:51:04] Today, you'll be operating the robotic assembly line. [00:51:10] I've noticed an issue with some sensors at cell four that could lead to assembly faults later. [00:51:16] A reset now can prevent that. [00:51:19] Should I guide you through the process? [00:51:21] What I love about my AI is how easy it is to use. [00:51:25] It brings together live machine data, knowledge from my colleagues, and all of our standard operating procedures. [00:51:31] Then explains everything in a way that makes sense to me. [00:51:34] At my pace, in my language. [00:51:37] And now select restart. [00:51:40] With AI from Siemens, I can solve problems efficiently, quickly, confidently. [00:51:50] The first tests show that colleagues are more confident in their work and they are also more productive. [00:51:56] That's impact in the real world by scaling industrial AI. [00:52:02] But we're heading for another huge bottleneck. [00:52:09] Power. [00:52:11] Electric power. [00:52:13] AI factories and data centers require gigawatts of electric power. [00:52:19] Now, what if, what if we had an energy source that was clean, safe, affordable, and practically limitless? [00:52:31] That's the promise of fusion power. [00:52:35] Which brings me to my next guest. [00:52:37] Commonwealth Fusion Systems is working on the world's first commercially viable machine to make fusion real. [00:52:46] Please welcome Bob Mungard. [00:52:58] Hello. [00:52:58] Hi Bob. [00:52:59] So, for the starters, what is fusion? [00:53:04] That's a good question. [00:53:05] So, fusion is the power of the universe. [00:53:08] It's the power of the stars. [00:53:09] And what we want to do is build machines here on Earth that create that power, that harness it. [00:53:14] And that would mean that we would go from an energy system that is about natural resource consumption to an energy system that's about technology. [00:53:22] Can you build these machines? [00:53:24] How fast can you build them? [00:53:25] How well do they run? [00:53:26] And we work on a type of machine called a tokamak, basically a bottle that holds a star inside of it. [00:53:32] So, these are massive machines, and there are for decades researchers, engineers trying to make it. [00:53:41] So, tell us, why should it work now? [00:53:43] Right. [00:53:44] So, we're at CES, where many of the technologies we see were ideas even just five years ago. [00:53:52] But the technical stack that has been developed from simulation tools, from the physical understanding has enabled a lot of these technologies. [00:54:00] And fusion is no different. [00:54:01] We have very large simulations where we understand the underlying physics of the process. [00:54:07] We can layer that into manufacturing these machines that are complex, but that perform these very delicate procedures. [00:54:15] And we can build all that up into a plant that makes energy. [00:54:20] That whole stack is enabling this entire industry. [00:54:24] So, let's go a little bit deeper into these challenges. [00:54:27] Because, I mean, putting a complex machine together is obviously very complicated, but you also have to produce it at the same time. [00:54:34] So, you have to design and manufacture. [00:54:36] And that's what you do also with the software from Siemens. [00:54:40] That's right. [00:54:41] In fusion, we're developing an entire new industry. [00:54:44] You can't just go and buy a fusion plant or a fusion part. [00:54:49] We actually have to take it from, you know, the first principles of how do these magnets work? [00:54:55] How do you manufacture them in a factory that can take new designs really quickly? [00:54:59] That can iterate really quickly? [00:55:01] That can compare test results to simulation tools and see where there's a mismatch? [00:55:06] What we don't understand. [00:55:07] And this produces a huge amount of data. [00:55:10] It produces a huge amount of insight. [00:55:12] And how do you harness that? [00:55:14] And that's where we use Siemens technology from the design of the actual parts [00:55:18] to how the factory runs, to how the plant itself is controlling, you know, very large amounts of current and cooling [00:55:25] to, you know, hold this star inside of it. [00:55:27] Right. [00:55:28] Actually, I was walking one of your plants where you built Spark, which is your first demonstrator, [00:55:35] where you really create more energy to put in, which is make a viability check of your technology. [00:55:41] And it's amazing what you do. [00:55:43] And imagine that the next one is even two times bigger. [00:55:47] It's hard to believe. [00:55:48] So tell me a little bit more about how Siemens technology played in that, designing it and, of course, [00:55:55] in supporting the manufacturing as well. [00:55:57] Right. [00:55:58] So we're building a plant in Massachusetts that is a machine that will make more power out from fusion than it takes to heat it up. [00:56:05] And to do that, Siemens was there from the beginning of how do we actually do the design work. [00:56:10] They're there in the factory with the PLCs that automate the factory. [00:56:14] And this is a factory that, you know, has engineers standing right next to the equipment that's running, that's being reconfigured all the time. [00:56:21] And then in the actual Spark facility, the brains that are actually running all the different processes, [00:56:27] it has to all come together to make the right conditions of a star inside of it. [00:56:31] Those machines are our Siemens machines, and they're running software that has been checked out in large computer simulations, [00:56:40] accelerated by NVIDIA, computation and DeepMind. [00:56:44] It's a very big stack all the way from idea to a plant. [00:56:49] So now here comes the key question. [00:56:53] When we will see real-world impact? [00:56:56] When will that commercially be viable? [00:56:59] So it's a great question. [00:57:01] It's a question I always get. [00:57:02] And the key part is that you can start to see things today. [00:57:05] So we have the plant in Massachusetts factory running, but the Spark facility is being put together. [00:57:12] And in fact, at CES, we have pieces of it that we've brought. [00:57:15] There's Siemens booth. [00:57:16] Go check them out. [00:57:17] But we also announced yesterday that we've put the first big magnet into place for Spark. [00:57:23] So you can see it coming together. [00:57:25] That's setting up for the next machine, which is going to be in Virginia. [00:57:29] That's what we call Arc. [00:57:30] And that's a commercial machine that will make about 400 megawatts of fusion power, electricity. [00:57:36] That can power a data center. [00:57:38] And in fact, Google has agreed to buy the power from Arc. [00:57:42] That is getting started here soon. [00:57:44] And so we can see out here on this technical stack, going from science all the way to energy, we can see a new age of abundance. [00:57:51] So, Bob, thank you very much. [00:57:53] There are a lot of customers counting on you to make that work, and we are here to support. [00:57:57] Thanks for the partnership. [00:57:58] Thank you. [00:57:59] Thank you. [00:58:00] Well, but even if we have abundant clean energy, there's another huge bottleneck: grid availability and grid stability. [00:58:15] Today, many electricity grids are on the verge of collapse. [00:58:20] Intermittent energy sources like solar and wind are increasing the stress on our grids. [00:58:26] We need to predict loads better and stabilize the grid and that in real time. [00:58:31] Our industrial AI-enabled technologies can do this, even for huge grids and autonomously. [00:58:40] We can simulate the impact of adding 10,000 electric vehicles to the neighborhood. [00:58:45] Buildings all over the city can act together and coordinate energy consumption so they could turn down their air condition for just half a minute to help stabilizing the grid. [00:58:58] And already today, we are using AI to maximize existing grid capacity by 20% without new infrastructure. [00:59:08] This is industrial AI creating impact in the real world. [00:59:15] So what does this all mean for you? [00:59:21] No matter whether you design, engineer, or operate, or all of the three, no matter what you make, food, cars, medicine, energy, no matter what type of infrastructure you manage, [00:59:37] you manage the water supply, a transportation system, or a grid. [00:59:43] We have the technology, the domain know-how, and the partners to help you do what you do, but faster, more efficiently, and more sustainably by scaling industrial AI in the real world. [00:59:58] Electricity. [01:00:04] Today, those of us in this room barely think about it. [01:00:09] Our lights turn on, our coffee machines run, our computers charge. [01:00:14] Electricity is a fact of life. [01:00:17] Soon, industrial AI will become a fact of life. [01:00:25] We will produce only what's needed, exactly what's needed, where it's needed. [01:00:32] Equipment failures and outages will be rare, unpredicted ones even rarer. [01:00:37] Medicine tailored to your unique biology will just be standard healthcare. [01:00:45] Autonomous vehicles will move with 100% reliability and zero emissions. [01:00:51] And knowledge will transfer through generations as naturally as current through a wire. [01:01:01] Energy will flow cleanly abundantly to infrastructure that repairs and optimizes itself and serves a world so defined by AI that we hardly notice it anymore. [01:01:15] When a breakthrough innovation becomes a real world essential and a general purpose technology becomes the invisible fabric of our lives. [01:01:29] That's when we'll know that we have scaled AI in the real world and realized its potentials. [01:01:37] And together with our customers, with our partners, with you, transformed the everyday for everyone. [01:01:46] Thank you.