CES 2026: AMD CEO Lisa Su delivers keynote on AI

[00:00:00] at ces 2026 innovators from every corner of the globe show up to the most powerful tech event in [00:00:09] the world we'll rally together imagining what's next and we'll make it real solving the greatest [00:00:17] challenges facing our world and proving that technology isn't just moving us forward it's [00:00:24] creating new possibilities here ai isn't an idea it's intelligence in motion the innovators [00:00:35] the storytellers and the game changers come together to redefine the landscape of content [00:00:41] creativity and culture pioneers will push healthcare beyond what we ever thought possible [00:00:50] from connected roads to autonomous flight from next-gen marine tech to innovations that feed [00:00:56] the planet this is what drives us forward in the halls of ces technology doesn't just compute [00:01:05] it collaborates quantum thinkers cyber security leaders fintech visionaries and robotics engineers [00:01:14] are rewriting the rules of enterprise itself because when the world's most determined minds [00:01:22] come together we don't just predict the future we build it this is where bold ideas meet a global force [00:01:33] where industries converge partnerships ignite and breakthroughs take center stage [00:01:41] innovation innovation isn't a solo act it's a shared pursuit ces 2026 innovators show up [00:01:56] you of the consumer technology association gary shapiro [00:02:09] thank you very much good evening everyone welcome to las vegas ces 2026 with our first keynote of the year [00:02:23] every year ces gives us a front row seat to the ideas and breakthroughs that shape the next decade [00:02:31] but it's actually the people behind those breakthroughs the leaders challenging what's possible who truly [00:02:39] define the moment tonight tonight we get a head start to ces 2026 with one such leader [00:02:47] you know ces has always been a platform for bold vision and ambition [00:02:54] it's where ideas become industries and where the next era of human progress first takes shape [00:03:01] and this year as ai accelerates change across every sector it's more important than ever that we hear [00:03:08] directly from the architects building the systems and breakthroughs it will define our very future [00:03:14] and that's why it is a tremendous honor to enter to introduce to you a leader quite frankly one of the [00:03:21] bastions of ces whose vision and impact can be felt across the entire technology landscape dr lisa sue [00:03:29] lisa is no stranger to this stage lisa is no stranger to this stage she has keynoted ces before [00:03:35] and each time she's done so she's helped set the tone for the industry and the year ahead [00:03:42] since becoming ceo in 2014 lisa has led the company through one of the most remarkable transformations [00:03:49] in modern technology under her leadership amd reinvented itself through relentless innovation [00:03:56] in high performance and anti-computing delivering products that now power ai training and inference [00:04:03] scientific research enterprise workloads cloud infrastructure and the devices and experiences [00:04:10] which millions of people rely on every single day today amd is a central force in the global ai [00:04:18] transformation its cpus gpus and adaptive computing solutions help unlock new capabilities across cloud [00:04:27] enterprise edge and pc and while the industry has spent years discussing what ai could become lisa has been [00:04:35] focused on building the computing foundation that makes ai real and accessible at scale but what truly [00:04:43] distinguishes lisa is her leadership she's analytical and deeply technical yet always grounded in purpose [00:04:52] she brings a rare combination of scientific brilliance strategic clarity and human-centered thinking [00:04:59] she believes in partnering deeply to engineer solutions that matter solutions that advance society [00:05:04] strengthen industries and expand opportunity that's the kind of leadership ces is designed to elevate [00:05:11] and that's the kind of leadership we all need as we navigate a world of accelerated innovation [00:05:17] tonight lisa will share amd's vision for how high performance computing and advanced ai architectures [00:05:25] will transform every part of our digital and physical world from research healthcare and space exploration [00:05:34] to education and productivity so speak to the extraordinary pace of ai the breakthroughs happening now [00:05:41] including the opportunities and responsibilities that come with building the future [00:06:02] hello hello hello and welcome to this unique moment in human history [00:06:09] a moment where what's possible might we forget what's impossible where any game you play [00:06:18] now has the power to play by your rules [00:06:28] where we're not only helps model the possibilities of what a city can be [00:06:34] but make sure our kids never forget what our cities used to be [00:06:42] a moment where no hope meets the treatment plans of ai mac genomes [00:06:50] a moment where a.i.i.i.i.s helping design a renewable energy source as powerful as the sun itself [00:07:16] and helping make travel time across the Atlantic just another puddle jumper [00:07:28] but as fast as everything's changing there's one thing that won't [00:07:34] we're working tirelessly to create a world where the most advanced ai capabilities end up in the right hands [00:07:45] and we're working in the right hand so we'll just keep walking even though [00:07:53] so ladies and gentlemen it is my privilege to welcome to the stage a globally respected technologist [00:08:00] an industry-defining ceo and a leader whose work continues to shape the very trajectory of modern computing [00:08:19] ladies and gentlemen please join me in welcoming to the stage chair and ceo of amd dr lisa [00:08:28] sue [00:08:37] thank you [00:08:44] all right what an audience how are you guys doing tonight [00:08:49] that sounds wonderful first of all thank you gary and welcome to everyone here in las vegas and [00:08:55] joining us online it's great to be here with all of you to kick off ces 2026 and i have to say [00:09:02] every year i love coming to ces to see all the latest and greatest tech and catch up with so many [00:09:09] friends and partners but this year i'm especially honored to be here with all of you to open ces now we have a completely packed show for you tonight and it will come as no surprise that tonight is all about ai [00:09:22] although the rate and pace of ai innovation has been incredible over the last few years [00:09:29] my theme for tonight is you ain't seen nothing yet [00:09:37] we are just starting to realize the power of ai [00:09:41] and tonight i'm going to show you a number of examples of where we're headed and i'll be joined [00:09:46] by some of the leading experts in the world from industry giants to breakthrough startups and together [00:09:52] we are working to bring ai everywhere and for everyone so let's get started [00:09:59] at amd our mission is to push the boundaries of high performance and ai computing to help solve the [00:10:06] world's most important challenges today i'm incredibly proud to say that amd technology touches the lives of [00:10:13] billions of billions of people every day from the largest cloud data centers to the world's fastest [00:10:18] supercomputers to 5g networks transportation and gaming every one of these areas is being transformed by ai [00:10:29] ai is the most important technology of the last 50 years and i can say it's absolutely our number one [00:10:35] priority at amd it's already touching every major industry whether you're going to talk about health [00:10:42] care or science or manufacturing or commerce and we're just scratching the surface ai is going to be [00:10:49] everywhere over the next few years and most importantly ai is for everyone it makes us smarter it makes us [00:10:58] more capable it enables each one of us to be a more productive version of ourselves and at amd [00:11:06] we're building the compute foundation to make that future real for every company and for every person [00:11:13] now since the launch of chat ppt a few years ago i'm sure we all remember the first time we tried it [00:11:19] we've gone from a million people using ai to now more than a billion active users this is just an [00:11:27] incredible ramp it look it took the internet decades to reach that same milestone now what we are projecting [00:11:35] is even more amazing we see the adoption of ai growing to over 5 billion active users [00:11:42] as ai truly becomes indispensable to every part of our lives just like the cell phone and the internet [00:11:49] of today now the foundation of ai is compute with all of that user growth we have seen a huge surge [00:11:58] in demand in the global compute infrastructure growing from about one zeta flop in 2022 to more [00:12:05] than a hundred zeta flops in 2025. now that sounds big that's actually a hundred times [00:12:12] but what you're going to hear tonight from everyone is we won't have we don't have nearly enough compute [00:12:21] for everything that we can possibly do we have incredible innovation happening models are becoming [00:12:28] much more capable they're thinking and reasoning they're making better decisions and that goes even [00:12:34] further when we extend that to agents overall so to enable ai everywhere we need to increase the world's [00:12:42] compute capacity another hundred times over the next few years to more than 10 yodaflops over the next [00:12:50] five years now let me take a survey how many of you know what a yodaflop is raise your hand please [00:12:59] a yodaflop is a one followed by 24 zeros [00:13:05] so 10 yodaflops is 10 000 times more compute than we had in 2022 there's just never ever been anything [00:13:14] like this in the history of computing and that's really because there's never been a technology like ai [00:13:23] now to enable this you need ai in every compute platform so what we're going to talk about tonight [00:13:28] is the whole gamut you know we're going to talk about the cloud where it runs continuously delivering [00:13:34] intelligence globally we're going to talk about pcs where it helps us work smarter and personalize [00:13:39] every experience that we have and we're going to talk about the edge where powers machines that can [00:13:45] make real-time decisions in the real world amd is the only company that has the full range of compute [00:13:52] engines to make this vision a reality you really need to have the right compute for each workload [00:13:59] and that means gpus that means cpus that means npus that means custom accelerators we have them all [00:14:07] and each of them can be tuned for the application to give you the best performance as well as the most [00:14:14] cost-effective solution so tonight we're going to go on a journey so you're going to go with me through [00:14:20] several chapters as we showcase the latest ai innovations across cloud pcs healthcare and much more [00:14:28] so let's go ahead and start with the first chapter which is the cloud [00:14:35] the cloud is really where the largest models are trained and where intelligence is delivered to [00:14:41] billions of users in real time for developers the cloud gives them instant access to massive compute [00:14:48] the latest tools and the ability to deploy and scale as use cases take off the cloud is also where most of [00:14:55] us experience ai today so whether you're using chat gpt or gemini or grok or you're coding with co-pilots [00:15:03] all of these powerful models are running in the cloud now today amd is powering ai at every level of [00:15:11] the cloud every major cloud provider runs on amd epic cpus and eight of the top 10 ai companies use instinct [00:15:19] accelerators to power their most advanced models and the demand for more compute is just continuing to go [00:15:26] up let me just show you a few graphs over the past decade the compute needed to train the leading ai models [00:15:34] has increased more than four times every year and that trend is just continuing that's how we're getting [00:15:41] today's models that are dramatically smarter and more useful at the same time as more people are using ai [00:15:49] we've seen an explosion over the last two years of inference growing the number of tokens a hundred times [00:15:56] really hitting an inflection point you can just see how much that inferences is really taking off [00:16:02] and to keep up with this commute compute demand you really need the entire ecosystem to come together [00:16:09] so what we like to say is the real challenge is how do we put ai infrastructure at yada scale [00:16:16] and that requires more than just raw performance it starts with leadership compute cpus gpus networking [00:16:25] coming together it takes an open modular rack design that can evolve over product generations it requires [00:16:33] high-speed networking to connect thousands of accelerators into a single unified system and it has to be [00:16:40] really easy to deploy so we want full turnkey solutions that's exactly why we built helios our next [00:16:49] generation rack scale platform for the yada scale ai error helios requires innovation at every single level [00:16:58] hardware software and systems it starts with our engineering teams who designed our next generation [00:17:05] instinct mi455 accelerators to deliver the largest generational performance increase we've ever achieved [00:17:12] and my 455 gpus are built using leading edge two nanometer and three nanometer process technologies [00:17:21] and advanced 3d chiplet packaging with ultra fast high bandwidth hbm4 memory this is integrated into a [00:17:30] compute tray with our epic cpus and pensano networking chips to create a tightly integrated platform each tray is then [00:17:38] connected with high speed ultra accelerator link protocol tunneled over ethernet which enables the 72 gpus in [00:17:45] the rack to function as a single compute unit and then from there we connect thousands of helios racks [00:17:53] to build powerful ai clusters using industry standard ultra ethernet nix and pensano programmable dpus [00:18:01] that can accelerate ai performance even more by offloading some of the tasks from the gpus now we are at ces [00:18:09] it is a little bit about show and tell so i am proud to show you helios right here in vegas [00:18:17] the world's best a iraq [00:18:31] you [00:18:31] you [00:18:31] you [00:18:47] now for those of you who have not seen iraq before let me tell you helios is a monster of iraq [00:19:07] this is no regular rack okay this is a double wide design based on the ocp open rack wide standard [00:19:15] developed in collaboration with meta and it weighs nearly 7 000 pounds [00:19:21] so gary it took us a bit to get it up here just so you know [00:19:26] but we wanted to show you what is really powering all this ai it is actually more than two compact cars [00:19:33] now the way we've designed helios it was really working closely with our lead customers and we chose [00:19:38] this design so that we could optimize serviceability manufacturability and reliability for next generation [00:19:45] ai data centers now let me show you a few other things at the center of helios is the compute tray [00:19:54] so let's take a closer look at what one of those trays look like [00:20:07] now i can tell you i probably cannot lift this compute tray [00:20:12] so it had to come out [00:20:14] but let me just describe it a little bit each helios compute tray includes four mi455 gpus [00:20:20] and they're paired with the next gen epic venice cpu and pensando networking chips and all of this [00:20:27] is liquid cooled so that we can maximize performance at the heart of helios is our next generation instinct [00:20:34] and you guys have seen me hold up a lot of chips in my career but today i can tell you i am genuinely excited to hold up this chip [00:20:44] so let me show you mi455x for the very first time [00:20:58] mi455 is the most advanced chip we've ever built it's pretty darn big it has 320 billion transistors [00:21:09] 70 percent more than mi355 it includes 12 2 nanometer and 3 nanometer compute in io chiplets [00:21:17] and 432 gigabytes of ultra fast hbm4 all connected with our next gen 3d chip stacking technology [00:21:26] so we put four of these into the compute trays up here [00:21:32] and then driving those gpus is our next generation epic cpu codename venice [00:21:39] venice extends our leadership across every dimension that matters in the data center [00:21:44] more performance better efficiency and lower total cost of ownership now let me show you venice for the [00:21:50] first time [00:21:54] i have to say this is another beautiful chip [00:22:02] i i do love our chips so i can say that for sure uh venice is built with two nanometer process [00:22:08] technology and features up to 256 of our newest high performance zen six cores [00:22:16] and the key here is we actually designed venice to be the best ai cpu we doubled the memory and gpu [00:22:24] bandwidth from our prior generation so venice can feed mi455 with data at full speed even at rack [00:22:30] scale so this is really about co-engineering and we tie it all together with our 800 gig ethernet [00:22:37] pensando volcano and selena networking chips delivering ultra high bandwidth as well as ultra low latency [00:22:44] so tens of thousands of helios racks can scale across the data center [00:22:50] now just to give you a little bit of the scale of what this means that means that each helios rack has [00:22:55] more than 18 000 cdna 5 gpu compute units and more than 4 600 zen6 cpu cores delivering up to 2.9 exaflops of [00:23:06] performance each rack also includes 31 terabytes of hbm4 memory an industry leading 260 terabytes per [00:23:14] second of scale-up bandwidth and 43 terabytes per second of aggregate scale-out bandwidth to move data in and [00:23:21] out incredibly fast suffice it to say those numbers are big when we launch helios later this year and [00:23:30] i'm happy to say helios is exactly on track to launch later this year we expect it will set the new [00:23:37] benchmark for ai performance and to just to put this performance in context just over six months ago [00:23:44] we launched mi355 and we delivered up to 3x more inference throughput versus the prior generation [00:23:51] and now with mi455 we're bending that curve further delivering up to 10 times more performance across [00:23:59] a wide range of models and workloads that is game changing mi55 455 allows developers to build larger [00:24:09] models more capable agents and more powerful applications and no one is pushing faster and [00:24:16] further in each one of these areas than open ai to talk about where ai is headed and the work that we're [00:24:23] doing together i'm extremely happy to welcome the president and co-founder of open ai greg brockman to [00:24:29] greg it is so great to have you here thank you for being here um you know open ai truly started all of [00:24:48] this with the release of chat gpt a few years ago and the progress you've made is just incredible we're [00:24:55] absolutely thrilled about our deep partnership can you just give us a picture of where are things [00:25:00] today what are you seeing and how are we working together well first of all it's great to be here [00:25:04] thank you for having me uh chat gpt is very much the overnight success that was seven years in the [00:25:11] making right that we started open ai back in 2015 with a vision that deep learning could lead to [00:25:18] artificial general intelligence to very powerful systems that could benefit everyone and we wanted to help [00:25:23] to actually realize that technology and bring it to the world and democratize it and we spent a long [00:25:29] time just making progress we're year over year that the benchmarks will look better and better but [00:25:35] the first time we had something that was so useful that many people around the world wanted to use it [00:25:39] was chat gpt and we were just blown away by the creativity and the ways in which people found how to [00:25:46] really leverage the models we had produced in their daily lives and so just out of curiosity how many people [00:25:51] in the room are chat gpt users that was pretty much the whole room i would thank you i'm glad to hear [00:25:59] it but very importantly how many of you have had an experience that was very key to your life or the [00:26:04] life of a loved one whether it's in healthcare in helping manage a newborn i in any other walk of your life [00:26:14] and to me that's the metric that we want to optimize and that seeing that number go up and to the right [00:26:19] has been something that has been really different over 2025 right that we really move from being just [00:26:24] a text box you ask a question you get an answer something very simple contained to people really [00:26:30] using it for very personal very important things in their lives and it's not just in personal lives for [00:26:35] healthcare and aspects like that it's also in the enterprise right and really starting to bring models like [00:26:41] codecs to be able to transform software engineering and i think that this year we're really going to see [00:26:47] enterprise agents really take off we're seeing scientific discovery start to be really accelerated [00:26:52] whether it's developing novel math proofs the first time we saw that was just a couple months ago [00:26:57] and the progress is continuing and it's really across every single endeavor of human knowledge work [00:27:02] where there's human intelligence that can be leveraged right you can amplify it we now have an assistant [00:27:08] right we now have a tool we have an advisor that is able to amplify what people want to do i i completely [00:27:14] agree with you greg i think we have seen just um an enormous acceleration of what we're using this tech [00:27:20] for now i would say i think every single time i see you you tell me you need more compute it's true it's [00:27:27] true it's almost like a broken record like you could just you know just play the greg wants more compute [00:27:34] um can you talk about just some of the things that you're seeing in the infrastructure some of the [00:27:38] bottlenecks and you know where do you think we should be focusing as an industry well the why [00:27:44] why doing more compute is the most important question right which is really when the models are not that [00:27:48] capable right and where we were in 2015 2016 2017 and so forth is that you basically just wanted to [00:27:54] train a model and evaluate it right maybe there'd be a very narrow task to be useful for but as we've made this [00:28:00] exponential progress on the models then there's actually exponential utility to them people want [00:28:07] to bring it into their lives in a very scalable way and i think that what we're seeing is as we move from [00:28:13] you ask a question you get an answer to agentic workflows where you ask the model to write some [00:28:18] software for you and it goes off for minutes or hours or soon even days and you're not just operating one [00:28:25] agent you're operating a fleet of agents right you can have 10 different work streams all going at [00:28:30] once on your behalf for a single developer right and that it should be the case that you wake up in [00:28:35] the morning and this is the kind of thing we are going to build by the way uh and chat gpt has taken [00:28:40] items off your to-do list at home and at work and that all of that that's going to require that you [00:28:46] know that big graph that you had of how much compute the world's going to need that's going to require far [00:28:50] more compute than we have right now like i would love to have a gpu running in the background for [00:28:55] every single person in the world because i think it can deliver value for them but that's billions [00:29:00] of gpus no one has a plan to build that kind of scale and so what we're really seeing is benefits in [00:29:06] people's lives right we're seeing for example some of my favorite applications and some of the ones i [00:29:10] think are the most important are in healthcare right that we actually see people's lives being saved [00:29:15] through chat gpt just over the holidays uh one of my co-workers uh that her husband had leg pain they [00:29:22] went to the hospital they went to the er and they got it x-rayed and the doctors are like ah it's a [00:29:28] pulled muscle just wait it out you'll be fine they went home it got a little bit worse type the symptoms [00:29:34] into chat gpt chat gpt said go back to the er this could be a blood clot and in fact it was it was deep [00:29:40] brain vein thrombosis in the uh in the leg in addition to two uh blood clots on on the lungs and [00:29:46] if they just waited it out that would have been likely fatal um it's not a unique story i fiji our [00:29:54] ceo of applications i work very closely with every day chat gpt literally saved her life too right she [00:29:59] was in the hospital um for a kidney stone and had an infection they're about to inject a uh antibiotic [00:30:06] and she said wait just a moment she asked chat gpt whether that one was safe for her chat gpt which [00:30:11] has all of her medical history said no no because you had this other infection two years ago that [00:30:16] could re-trigger it and that could actually be life-threatening as well and so she showed it to [00:30:20] the doctor the doctor's like wait what you had this condition i didn't know i only had five minutes to [00:30:26] review your medical history i i completely agree greg i mean that's one of the things that we can always [00:30:32] all of us can use a helper yes and that's really what we have here um look i mean i think you've [00:30:37] painted a vivid picture of why we need more compute of what we can do with ai um i think we feel [00:30:43] exactly the same way now we've also done an incredible amount of work with your engineering teams um [00:30:48] mi-455 and helios is actually a lot of it is through some of the the feedback from our engineering [00:30:54] teams working closely together you know can you talk a little bit about that infrastructure and and [00:30:59] what are you what are your customers wanting and how are you going to use mi-455 well one of the key [00:31:04] things with how ai is evolving is thinking about the balance of different resources on the gpu um [00:31:10] and so we have a slide to show i how how we've seen the evolution of this uh the balance of resources [00:31:16] across different mi uh generations uh so you see this slide that that i very painstakingly put together [00:31:21] actually i did not painstakingly put it together i asked chat to go put it to go create the slide [00:31:27] and so it literally did all the the research and you can see uh some sources at the bottom i i and it [00:31:33] actually went and read a bunch of different i you know amd materials created these charts put together [00:31:40] the title put together all these headers and produced not just an answer for me to then go do a bunch of [00:31:46] work it produced an artifact an artifact that i can show and this is just one simple example that you can [00:31:51] do today with chat gpt right that we are moving to a world where you are going to be able to have an [00:31:57] agent that does all this work for you and for that we're going to need to have hardware that is really [00:32:03] tuned to our applications what we have in mind is that we're moving to a world where human attention [00:32:09] human intent becomes the most precious resource and so that there should be very low latency interaction [00:32:15] anytime a human's involved but there should be an ocean of agentic compute that's constantly running [00:32:21] that's very high throughput and these two different regimes of low latency and high throughput yield a [00:32:26] bunch of different pressures on hardware manufacturers such as yourself so it's a pleasure to be working [00:32:31] together we we like building gpus for you that works well um look uh lastly greg let's talk a little bit [00:32:38] about the future you know paint a picture you know one of the things that we've talked about is there's [00:32:43] some people out there who are wondering you know is the demand really there can ai compute like do we [00:32:48] really need all of this ai compute and i know you and i have talked about it i think people don't have a [00:32:53] view of the future that you see i mean you have like a special seat so paint the world for what this [00:32:58] looks like in a few years well looking backwards we have been tripling our compute every single year for [00:33:05] the past couple years and we've also tripled our revenue and the thing that we find within open ai is [00:33:10] every time we want to release a new feature we want to produce a new model we want to bring this [00:33:14] technology to the world we have a big fight internally over compute because there are so many things we [00:33:20] want to launch and produce for all of you that we simply cannot because we are compute constrained [00:33:25] and i think we're moving to a world where gdp growth will itself be driven by the amount of [00:33:31] compute that is available in a particular country in a particular region and i think that we're starting [00:33:37] to see the first inklings of this and i think over the next couple years we'll see it start to hit in [00:33:42] a real way and i think that ai is something where i think data centers can actually be very beneficial to [00:33:47] local communities i think that's a really important thing for us to really prove to people but also the [00:33:51] ai technology you produce that is also something where in terms of scientific advances you think about [00:33:57] what has been the most fundamental driver of increase of quality of life right it really is about science [00:34:02] and every time we've gone into specific domains you just see how much limitation there is from how [00:34:07] things are done because that's just you know there's a particular discipline that's built with much of [00:34:11] expertise there are a small number of experts and that it's hard for them to propagate that to future [00:34:16] generations for example in biology we hooked up gpd5 to a wet lab setup and had i you know humans [00:34:24] described what the wet lab looked like the model said here are a couple of ideas to try the humans [00:34:29] would go try it and it actually produced a 79x almost 100x fold improvement in the efficiency of a [00:34:36] particular protocol and that's just one particular reaction that people have spent some time actually [00:34:40] optimizing but not like a ton and ton of time because it's just there's so much surface area available [00:34:46] in biology that no human can possibly get to all of it no human can be an expert across every single [00:34:51] subfield and i think what we're going to see is ai's that really bridge across disciplines [00:34:56] that humanity has been unable to bridge right you see this within healthcare where as humans learn [00:35:01] more we specialize more but we're going to have ai's going to amplify and so i think it'll be for hard [00:35:06] problems that ai will be brought to bear this will be true for enterprise every single application i think [00:35:12] we'll have an agent i that is accelerating what people want to do and i think the hardest problem for [00:35:18] humanity will be deciding how do we use the limited resources we have to get the most benefit for [00:35:23] everyone that is an incredible vision greg we are so excited to be working with you i think there's [00:35:30] no question in the world that we have the power to really change people's lives thank you for the [00:35:36] partnership and really look forward to it thank you so as you heard from greg compute is key and mi455 is [00:35:51] a game changer but with the mi400 series we've designed a full portfolio of solutions for cloud enterprise [00:35:59] super computing and sovereign ai at the top is helios that's built for the bleeding ed performance [00:36:05] hyperscale training and distributed inference at rack scale for enterprise ai deployments we have instinct [00:36:12] mi440x gpus that deliver leadership training and inference performance in a compact 8 gpu server [00:36:20] designed for easy use in today's existing data center infrastructure and for sovereign ai and super [00:36:26] computing where extreme accuracy matters the most we have the mi430x platform that delivers leadership [00:36:33] hybrid computing capabilities for both high precision scientific and ai data types this is something [00:36:39] unique that we at amd do because of our chiplet technology we can actually have the right compute [00:36:46] for the right application now hardware is only part of the story we believe an open ecosystem is essential [00:36:55] to the future of ai time and time again we've seen that innovation actually gets faster when the industry [00:37:02] comes together and aligns around an open infrastructure and shared technology standards and amd is the only [00:37:09] company delivering openness across the full stack that's hardware software and the broader solutions ecosystem [00:37:16] our software strategy starts with rockham rockham is the industry's highest performance open software stack [00:37:23] for ai we have day zero support for the most widely used frameworks tools and model hubs and it's also natively [00:37:31] supported by the top open source projects like pi torch vllm sg lang hugging face and others that are downloaded [00:37:40] more than a hundred million times a month and run out of the box on instinct making it easier than ever [00:37:46] for developers to build deploy and scale on amd one of the exciting ai companies using amd and rockham to power [00:37:54] their models is luma ai please join me in welcoming the luma ai ceo and co-founder amd jain to the stage [00:38:10] hello how are you it's great to have you here with us um you're doing some incredible work in video [00:38:18] generation and multimodal models can you tell us a little bit about luma and what you're doing absolutely [00:38:23] lisa thank you so much for having me here of course luma's mission is to build multimodal and general [00:38:28] intelligence so ai can understand our world and help us simulate and and improve it most ai video [00:38:35] and image models today uh they're in early early stages and they're they're used to generate pixels [00:38:40] they're used to produce you know pretty pictures what is needed in the world are more intelligent models [00:38:46] that combine audio video language image all together so at luma we are training these systems that [00:38:52] simulate physics causality are able to go out do research call tools and then finally render out [00:39:00] the results in audio video image text whatever is appropriate for the information that you're trying [00:39:05] to work with in short we are modeling and generating worlds so as an example let me show you some of our [00:39:13] some results from our latest model ray three uh by the way ray three is the world's first reasoning video [00:39:21] model so what it means is it actually is able to think first in pixels and latence and and and decide [00:39:26] whether what it's about to generate is good and it's also the world's first model that can generate in 4k [00:39:31] and hdr so uh please take a look [00:39:40] when you close your eyes [00:39:44] what do you see [00:39:48] when we talk about reality [00:39:52] how do you know what is real [00:39:56] or what is simply [00:40:00] your imagination [00:40:16] now open your eyes [00:40:30] i can say amit that looked pretty incredible [00:40:47] so tell us how are customers using ray3 today [00:40:51] so we are working with very large enterprises as well as individual creators across the spectrum [00:40:57] and we work with them in advertising media entertainment and and industries where like you [00:41:02] know you want to tell your story um 2025 was the year when they started to deploy our models and [00:41:10] experiment with them and and towards the end of it we have we are seeing large-scale deployments [00:41:14] where people are using our models for as much as actually making a 90-minute feature-length movie [00:41:20] what customers are also asking us a lot for well as they're using it more and more is control and [00:41:25] precision how can they get their particular vision out onto the screen and what we have realized and [00:41:33] and to our research is that control comes from intelligence not just better prompts you can't [00:41:37] keep like you know typing in again and again and actually do those things so we have built [00:41:41] a whole new model on top of ray3 called ray3 modify that allows you to edit the world so [00:41:49] let me show you actually what that looks like this won't have audio and i'm going to tell you a [00:41:52] little bit about what you're seeing so what's playing on the screen is a demo of ray3's world editing capabilities [00:41:58] it can take any real or ai footage so footage from cameras or footage that you generated [00:42:03] and change it as little or as much as you want to realize the creative goals [00:42:07] it's a powerful system that we have developed for our most ambitious customers who are most demanding [00:42:14] and and they they spread the gamus across entertainment advertising uh and and this [00:42:19] has allowed us to enable a new era of hybrid human ai productions the human becomes the prompt through [00:42:26] motion timing and direction like you know you act it out and and then the model can produce it what that [00:42:32] means in practice is that filmmakers and creators can create entire cinematic universes now without [00:42:38] elaborate sets and then edit and modify anything to get to the result they want this has never been [00:42:45] possible before but in 2026 we are focused on actually going much further 2026 will be the year of agents [00:42:53] where ai will be able to help you to accomplish more of the task or hopefully the full end-to-end of the [00:42:59] task rather than doing some patchwork so our teams have been working diligently building the world's most [00:43:06] powerful multi-modal agent models using luma will suddenly feel like you have a large team of capable [00:43:13] creatives who are working with you in your creative pursuit i want to show you a brief demo what that would [00:43:18] feel like so what you're seeing here is a new multimodal agent uh that can take a whole script [00:43:27] from of ideas with like you know characters and everything and start imagining that in front of you now [00:43:33] this is not script to movie this is human ai interaction and our next generation of models [00:43:39] provide the ability to analyze multiple frames long form video and make selection and maintaining the fidelity of the [00:43:46] characters scenes and story and only editing when it's needed here you're seeing human and ai collaborate [00:43:52] in designing characters environment shots and the whole world and with our agents we believe that [00:43:58] creatives will be able to make entire stories what used to take a large production before again this has [00:44:04] never been possible before and and we have been using it heavily internally and and we couldn't be more [00:44:08] excited individual creatives or small teams will suddenly have the power of doing what entire hollywood [00:44:15] studios do that's um pretty amazing it's really nice to see how these luma agents come together [00:44:21] and make this happen now i know you have a lot of choices in compute and when we first started [00:44:26] talking actually you called me and said you needed compute and i said i thought i could help [00:44:32] can you tell us some a bit about why you chose amd and and what has your experience been [00:44:37] yeah we bet on amd very early on that call was in 2024 early 2024 and since then our partnership has [00:44:44] grown into a large-scale uh collaboration between our teams so much so that today 60 percent of luma's [00:44:53] rapidly growing inference workloads actually run on amd cards uh today we also so you know initially when we [00:44:59] started out uh we used to do a bunch of engineering but today we are at a point where most any operators most any [00:45:05] workloads that we can imagine run out of the box on amd and this is huge props to your software teams and [00:45:11] and the diligent work that is going into the rockam ecosystem we are building multi-model models and [00:45:17] actually these workloads are very complex compared to text models uh one example of that is these [00:45:22] consumes hundreds of times thousands of times more tokens a video that you you know you saw a 10 second [00:45:29] video it's about 100 000 tokens easily compare that to a response from an llm it's about 200 to 300 tokens [00:45:37] so when we are working with this much information tco and inference economy is is absolutely critical to [00:45:45] our business otherwise there's no way to serve all the demand that is coming our way through our [00:45:50] collaboration with the amd team we have been able to achieve some of the best tco tcos total cost of [00:45:56] ownership that we have ever seen in our stack and we believe as we build these more complex models that [00:46:02] are able to do autoregressive diffusion and that are able to do text and image and audio and video [00:46:06] all at the same time this collaboration will allow us to significantly differentiate [00:46:12] on cost and efficiency which as you know in ai is a big deal so through this collaboration we have [00:46:19] developed such degree of confidence that in in 2026 we are expanding our partnership to a tune of about 10 [00:46:26] times what we have done before and these 455 cards uh and i cannot be more excited for mi455x because [00:46:38] this uh the rack scale solution and the memory and the infrastructure that you're building is essential [00:46:43] for us to be able to build these world simulation models well we love hearing that amit and look our goal [00:46:49] goal is to deliver you more powerful hardware your goal is to make it do amazing things so just give [00:46:55] us a little brief view of what do you see customers doing over the next few years that just isn't even [00:47:00] possible today right so uh as greg was mentioning early on in llm land right like you know in 2022 2023 [00:47:09] they were great for writing copies small emails things like that we could have never imagined that we would [00:47:14] actually put these models uh uh you know into into real-time systems into into uh healthcare and these [00:47:22] kind of things through accuracy and architecture and scaling llms have now gotten to that point video [00:47:28] models are currently in that early stage today they're great for generating video and pretty pictures but [00:47:33] soon by scaling these models up by by improving the accuracy and data we would end up with a place where [00:47:40] these models will help us simulate real physical processes in the world like cad architecture fluid [00:47:47] flows help us design entire rocket engines plant cities and this is not outrageous this is what we do [00:47:52] today manually with big giant teams in simulation environments these models will allow us to do that [00:47:59] and automate that to a great degree and as they become more and more accurate multi-model models is what we need [00:48:07] for the backbone of general purpose robotics your home robot uh you know will run hundreds of simulations [00:48:14] in its head in image and video and then try out like okay how how do i do this how do i solve this [00:48:20] so that it's able to do a lot more than current generation of llm and vlm robots are able to do [00:48:25] this is how the human brain works humans are natively multi-model our ai systems will be as well [00:48:30] that sounds wonderful amit look thank you so much for being here today thank you for the partnership and we [00:48:35] really look forward to all that you're going to do next thank you so much thank you [00:48:46] so you've heard from greg and amit what they said is they need more compute to build and run their [00:48:51] next-gen models and it is the same across every single customer that we have which is why the demand [00:48:57] for compute is growing faster than ever now meeting that demand means continuing to push the envelope on [00:49:03] performance far beyond where we are today mi400 series was a major inflection point in terms of delivering [00:49:11] leadership training across of all workloads inference scientific computing but we are not stopping there [00:49:18] development of our next-gen mi500 series is already well underway with mi500 we take another major leap in [00:49:27] performance it's built on our next-gen cdna6 architecture manufactured on two nanometer process technology [00:49:35] and uses higher speed hbm4e memory and with the launch of mi500 in 2027 we're on track to deliver [00:49:43] one thousand times one thousand x increase in ai performance over the last four years making more [00:49:50] powerful ai accessible to all so with that let's thank you [00:50:03] so with that now let's shift from the cloud to the devices that make ai more personal pcs [00:50:20] so for decades the pc has been about a powerful device helping us be more productive whether at work [00:50:26] or at school but with ai the pc has become not just a tool but it's a powerful essential part of our lives [00:50:34] as an active partner it learns how you work and it adapts to your habits and it can help you do things [00:50:40] faster than you've ever expected even when you're offline ai pcs are starting to deliver real value [00:50:48] across a wide range of everyday tasks from content creation and productivity to intelligent personal [00:50:54] assistance let's just take a look out of a few of the ai pc applications today starting with content [00:51:01] creation these videos were created from simple text prompts on a ryzen ai max pc so not in the cloud [00:51:09] but in a local environment anyone can generate professional quality photos and videos in minutes [00:51:15] with no design expertise microsoft has been a key enabler of aipcs helping bring next generation [00:51:22] capabilities directly into our productivity tools for example managing your meetings summarizing meetings [00:51:29] summarizing emails quickly finding files that you need using real-time translation on video conferences [00:51:37] and with microsoft copilot advanced ai capabilities are being built directly into the windows experience [00:51:43] to complete tasks faster you just describe what you need and the pc takes it from there [00:51:49] now at amd we saw the ai pc wave early and we invested that's why we've led every inflection point [00:51:58] we were the first to integrate a dedicated on-chip ai engine in 2023 and the first to deliver copilot plus [00:52:06] x86 pcs in 2024 and with ryzen ai max we created the first single chip x86 platform that could run a 200 [00:52:16] billion parameter model locally and now we're extending that leadership again with our next-gen ryzen ai [00:52:23] notebook and desktop processors so today i'm proud to announce the new ryzen ai 400 series the industry's [00:52:31] is broadest and most advanced family of ai tc processors ryzen ai 400 combines up to 12 high-performing [00:52:54] now powering the next generation [00:53:24] and of ai pc experiences takes more than just hardware it takes smarter software with models [00:53:32] that are lighter faster and can run directly on device these are different than what you're seeing [00:53:37] in the cloud so to talk more about this next wave of model innovation please welcome ramin hasani [00:53:44] co-founder and ceo of liquid ai [00:53:57] for me it's great to have you here um i'm very excited about the work that you guys are doing at [00:54:02] liquid you were really taking a different approach to models can you talk a little bit to the audience [00:54:07] about what liquid is doing and you know why it's different from others absolutely lisa it is great [00:54:14] to be here we're a foundation model company spun out of uh mit two and a half years ago [00:54:37] we're going to talk a little bit later and we're going to talk a little bit about which we're really going to talk a little bit to talk a little bit [00:54:42] about that we're going to talk a little bit about how we're going to talk a little bit about [00:54:43] how we're going to talk a little bit about how we're going to talk a little bit about [00:54:44] that we're going to talk a little bit about how we're going to talk a little bit about how we're going to talk a little bit about [00:54:48] the computational cost of intelligence from first principles without sacrificing quality [00:54:55] that means liquid models deliver frontier model quality right on a device device could be a phone [00:55:02] could be a laptop could be a robot could be a coffee machine and it could be an airplane [00:55:08] basically anywhere compute exists with three value propositions privacy speed and continuity [00:55:17] it can work seamlessly across online and offline workloads [00:55:22] ramin you know our teams have been working really closely on bringing more capable models to ai pcs [00:55:29] can you share a bit about that work absolutely today i've got two new product announcements one [00:55:37] we are excited to announce liquid foundation models 2.5 the most advanced tiny class of models on the on the [00:55:46] market at only 1.2 billion parameters the model performs best on instruction following capabilities [00:55:55] between its class and models that are larger in its class lfm 2.5 instances are the building blocks of [00:56:02] reliable ai agents on any device to put this in perspective for you this model delivers instruction [00:56:08] following capabilities better than the uh you know deep seek models and gemini pro kind of models gemini 2.5 pro [00:56:17] right on the device we are releasing five model instances a chat model an instruct model a japanese enhanced [00:56:26] language model a vision language model and a lightweight audio model audio language model basically [00:56:33] these are highly optimized for amd ryzen ai cpus gpus and npus and um today they are available for [00:56:41] download on hugging face and on our own platform leap you can enjoy them that's pretty cool [00:56:53] so we can stack this lfm 2.5 instances together to build agentic workflows but then it would be really [00:57:00] amazing if we can bring in all these modalities into one place so that brings me to my second [00:57:06] announcement lfm3 lfm3 it is designed natively multimodal to process text vision and audio as input and [00:57:16] deliver audio and text as an output in 10 different languages with sub 100 millisecond latency for [00:57:23] audio visual data you will get lfm3 later in the year all right that's fantastic so [00:57:33] now ramin help our audience understand like why should they be so excited about lfm3 like what can we [00:57:39] do with these models on an ai pc absolutely so most assistants ai assistants co-pilots today [00:57:46] are reactive agents you open an app then you ask a question it responds but when the ai is running [00:57:54] fast on the device and is always on it can be working you know on the tasks proactively for you [00:58:02] the task can be done in the background so let me show you a quick demo a reference design to inspire [00:58:08] what is possible to build on pcs with lfm instances let's jump in imagine you're a sales leader working [00:58:16] on your amd ryzen laptop with lfm3 backbone proactive agents activated you're in full focus mode working [00:58:23] on a spreadsheet notifications start piling up you get a calendar notification for a sales meeting [00:58:30] but want to continue your work in deep focus a liquid proactive agents notices the meeting and offers to [00:58:37] join on your behalf you allow the agent to join and while you focus on your data analysis task in the [00:58:44] the background the meeting is in progress with your agent representing you are you sure we can trust [00:58:49] this agent i think i'm a little worried there remain this system can actually transcribe [00:58:57] more than transcribing your systems you know and really [00:59:11] so with the deep research functionality you can analyze every email and draft the response for you [00:59:16] again everything under your own control you know you're not going to go uh this is not going to go rogue [00:59:22] everything is offline locally on the device so this system can deliver you know a summary it can do the [00:59:30] jobs better than what you have expected what you have seen from reactive agents i think this year is [00:59:35] going to be the year of um you know proactive agents and i'm very excited to announce that uh we're bringing [00:59:43] we're working we're collaborating with zoom to bring these features to the zoom platform actually that's fantastic [00:59:55] ramin we're really excited about what you're doing i think you've just given [00:59:59] people just a glimpse of what we can do when we bring you know true ai capability to our pc so thank you [01:00:07] we're excited and we look forward to all we're going to do together thank you so much thank you for having me [01:00:19] so now you've seen a little bit about what's possible with local ai but the latest pcs aren't [01:00:24] just running ai apps they're actually building them that's why we created ryzen ai max the ultimate pc [01:00:32] processor for creators gamers and ai developers it's the most powerful ai pc platform in the world with 16 [01:00:40] high performance zen 5 cpu cores and 40 rdna 3.5 gpu compute units and an xdna 2 npu delivering up to 50 [01:00:51] tops of ai performance all connected by a unified memory architecture that supports up to 128 gigabytes of [01:00:59] shared memory between the cpu and gpu in premium laptops ryzen ai max is significantly faster in both [01:01:07] ai and content creation applications compared to the latest macbook pro in small form factor workstations [01:01:14] ryzen ai max delivers comparable performance to at much lower price than nvidia's dgx spark generating up to [01:01:22] 1.7 times more tokens per second per dollar when running the latest gpt oss models and because [01:01:29] ryzen ai max supports both windows and linux natively developers maintain full access to their preferred [01:01:35] software environment tools and workflows now there are more than 30 ryzen ai max systems in market today [01:01:43] with new laptops all-in-ones and compact workstations launching at ces and rolling out throughout the year [01:01:50] but our mission is to advance ai everywhere for everyone the truth is there are ai developers many of you in [01:01:58] this room who want access to platforms that enable you to develop on the fly so we took this one step [01:02:04] further today i'm excited to announce the amd ryzen ai halo a new reference platform for local ai deployment [01:02:20] now i would say this is pretty beautiful do you guys agree so let me tell you what it is this is the [01:02:30] smallest ai development system in the world capable of running models with up to 200 billion parameters [01:02:37] locally not connected to anything it's powered by our highest end ryzen ai max processor with 128 gigabytes of [01:02:45] high speed unified memory that is shared by cpu gpu and npu this architecture accelerates system performance [01:02:54] and makes it possible to efficiently run large ai models on a compact desktop pc that fits in your hand [01:03:02] thank you halo supports multiple operating systems natively ships with our latest rockham software [01:03:14] stack comes pre-loaded with the leading open source developer tools and runs hundreds of models out of [01:03:20] the box and this really gives developers everything you need to build tests and deploy local agents and ai [01:03:27] applications directly on the pc now for all of you who are wondering halo is launching in the second quarter [01:03:33] of this year and we can't wait for folks to get their hands on them [01:03:43] so now let's turn to the world of gaming and content creation [01:03:48] a few gamers out there [01:03:51] i think there are a lot of gamers out there um look every day gamers and creators rely on amd [01:03:57] across ryzen and radeon pcs threadripper workstations and consoles from sony and microsoft [01:04:03] to deliver tens of billions of frames and while the visual quality of those frames has advanced [01:04:08] dramatically over the years the way we build those worlds really hasn't it still takes teams months or [01:04:15] even years to bring a 3d experience to life now ai is really starting to change that to show what's next [01:04:23] in 3d world creation i'm honored to introduce one of the most influential figures in ai known as the godmother [01:04:30] of ai her work has transformed how machines see and understand the world please welcome the co-founder [01:04:38] and ceo of world labs dr feifei lee [01:04:52] feifei we are so excited to have you here you know you've been one of the leaders shaping ai for [01:04:58] decades can you just give us a little bit of your perspective where are we today and why did you start [01:05:04] world labs yeah first of all thank you lisa for inviting me to be here congratulations to all the [01:05:10] announcement i can't wait to use some of them so it's true that there has uh truly been great breakthroughs [01:05:18] in uh ai progress in the past few years and as you said i've been around the block for a while for more [01:05:24] than two decades and i really cannot be more excited than now by where things are going so in the past few [01:05:32] years language based intelligence in ai technology really has taken the world by storm we're seeing [01:05:40] the proliferation of all kinds of capabilities and applications but the truth is there's a lot more [01:05:49] than just language intelligence even for us humans there's more than passively looking at life in the world [01:05:56] we are incredible spatial intelligent animals and we have profound capabilities that use our own spatial [01:06:05] intelligence that connects perception with action think about think about all of you being here how you [01:06:12] brave through airports this morning i'm one of them or woke up in your hotel room and get to the nice coffee shop [01:06:20] or find your way in this maze in vegas to be here all this requires spatial intelligence [01:06:29] so what excites me is that there's now a new wave of gen ai technology for both embody ai and generative ai [01:06:39] that we can finally give machines something closer to the human level spatial intelligence [01:06:48] it's the ability to not only perceive but create 3d or even 4d worlds reason about objects and people [01:06:57] and imagine entirely new environments that still obey the laws of physics and dynamics in worlds virtual or real [01:07:07] so that's why i started world labs i really want to bring spatial intelligence to life and deliver value [01:07:14] to people i i remember the first time i talked to you about your concept for world labs and your passion [01:07:21] about um what this could bring uh tell us a little bit about what your models do so the audience gets [01:07:27] a feel for what does this really mean yeah well um i heard that there are gamers out there so this is very [01:07:34] very exciting so traditionally building 3d scenes requires laser scanners or calibrated cameras or [01:07:42] hand-built models using pretty sophisticated and complicated software but at world labs we're creating [01:07:49] a new generation of models that can use the recent gen ai technology to learn the structure not only just flat [01:07:58] pixel structure i'm talking about 3d 4d structure of the world directly from data a lot of data so give [01:08:07] the model a few images and even one image it the model itself can fill in the missing details predicts [01:08:17] what's behind objects and generate rich consistent permanent navigable 3d worlds so what you're seeing [01:08:27] here on screen is um uh is a hobby world that's created by our uh world labs model called marble we just give it a [01:08:37] handful of images and they created these uh 3d scenes that are persistent and you can navigate you can even see a top view [01:08:44] and that our system transformed a few visual inputs into a fully navigable expensive 3d world and it shows how [01:08:55] these models not just reconstruct the environment they really imagine cohesive worlds wondrous worlds [01:09:04] and once these worlds exist they flow together and allowing effortless transition from one environment to [01:09:13] the next and scaling into something much larger and this is much closer to how humans piece together a place [01:09:22] from a few glances well it looks um you know pretty amazing that you can do that with such little input now can [01:09:30] you just show us a little bit about how the technology works yes definitely let's just ground it in a real [01:09:37] world a bit more from uh the the hobbit world uh let's let's do something that you're very familiar with [01:09:43] so over the break uh our team went to amd's silicon valley office i i hope they got your permission and [01:09:51] and they did not but that's okay okay well now here we are we just use some regular phone cameras there's [01:10:00] no special equipment just uh just phones to capture a few images and then we put them into world labs uh [01:10:07] a generative 3d generative model called marble and then our model that can use amd's mi325x chip and the [01:10:18] rock cam uh stack software stack can create a 3d version of that environment and including windows [01:10:28] doors furniture size and and sense uh sense of depth and scale and keep in mind you're not looking at [01:10:36] photos you're not looking at videos you're looking at truly 3d consistent worlds and then our team started to have a [01:10:45] little more fun and decided to be you decided to remodel exactly for free for you for different design styles [01:10:54] right i don't know which one you guys like the most i personally really like the egyptian one but uh [01:11:03] um maybe that's because i'm going there in a few months and uh while this transformation is keeping the [01:11:10] geometric consistency and and the 3d uh inputs so um you can imagine this can be such powerful [01:11:19] tools for many use cases whether you're doing robotic simulation or game development or design this can [01:11:28] what would traditionally take months to do in a typical workload we really could do it in minutes now [01:11:35] and we can even navigate into an entirely different world like actually the venetian hotel and we just [01:11:44] did that yesterday by tooki taking a picture and then put it in the model and then did have a little fun [01:11:51] and then it turned this whole place into a 3d imaginative uh imaginative space [01:12:03] and now i'm sure you guys can take pictures and send it to marvel and experience this yourself [01:12:09] but what you don't see here behind the scene is how much computation is happening and why inference speed [01:12:17] really matters the faster we can run these models the more responsive the world becomes instant camera [01:12:26] moves instant edits and a scene that stays coherent as you actually navigate and explore and that's [01:12:34] what's really important you know um feifei i think you're going to have a few people going out to your [01:12:39] website to try marvel um we'll keep the server up but um look that looked uh really amazing can you just [01:12:47] share a little bit about your experience working with amd and our work on instinct and rockam yeah of course [01:12:53] and uh even though we are old friends our partnership is relatively new and i gotta be honest where i'm [01:13:01] very impressed by how quickly this came together our part of our model is a real-time frame generative frame [01:13:09] model it was running on mi325x in under just a week and then with amd instinct the rock m our teams [01:13:19] were able to iterate really rapidly over a course of a few weeks to improve performance by more than four [01:13:26] fold and that was really impressive that matters because spatial intelligence is fundamentally different [01:13:34] from what came before teaching ai to understand and navigate 3d structure have motion understand physics [01:13:43] requires enormous memory massive parallelism and very fast inference [01:13:50] and i was seeing your announcement i can't wait to see platforms like mi450 continue to scale [01:13:59] and they will give us the ability to train larger world models and just as importantly [01:14:06] to run them fast enough that these environments can feel alive react instantly as the user or agents [01:14:16] move explore interact and create now that's um that's wonderful thank you for those comments your team has [01:14:23] been fantastic working together thank you um so feifei with all the compute performance that we're [01:14:29] going to give you and all of the innovation in your models give the audience a view of what to expect [01:14:34] over the next few years yeah i know and uh as you know me i will i don't like to hype i think the world [01:14:42] this is called under hype they say so no i think we should just share what it is it is gonna be a changing [01:14:50] world a lot of workflows a lot of things that were difficult to do will actually go through a revolution [01:14:59] because of the incredible technology so for example creators can now experience uh and create real world scenes [01:15:08] uh scenes in real time shaping what's in their mind's eye experimenting with the space the light the [01:15:17] movement as if they are sketching in inside a living world and the intelligent agents whether it's robots [01:15:26] or vehicles or vehicles or even tools can learn inside very rich physics aware digital worlds before they [01:15:35] even need to be deployed into the real one making them much safer make the development much faster more [01:15:43] capable and more helpful to people and designers for example architects can walk through ideas before anything [01:15:52] is built exploring form flow materials and navigate spaces rather than just looking at staring at static plans [01:16:05] so what excites me most is that this represents a shift in how ai shows up in our lives we're moving from [01:16:15] to systems that understand words and images passively to systems that not only understand but can help us to interact [01:16:24] with the world so lisa what we are sharing today um which is turning a handful of images or [01:16:32] or photos into coherent explorable world in real time is not a glimpse of the distant future anymore [01:16:41] it is really the beginning of the next chapter so you and i talk about this even offline we know as [01:16:49] powerful as ai technology is it's also our responsibility to deploy and develop it in ways that reflect true human [01:16:59] values that augment [01:17:07] that augment human creativity productivity and our care for each other while keeping people firmly at the [01:17:16] center of this story however powerful technologies are and i'm very excited to partner with amd and with [01:17:24] you on this journey [01:17:26] feifei i think i speak it for everyone um you are uh you know really an inspiration to the ai world [01:17:33] congratulations on all the great progress and thank you for joining us tonight thank you lisa [01:17:44] okay next shift now let's turn to the world of health care [01:18:03] of all the ways ai is advancing the world health care impacts us all and amd technology is enabling the [01:18:11] incredible to become possible cancer detection is happening earlier with supercomputers analyzing data [01:18:18] at massive scale patients are receiving therapy sooner with compute modeling complex biological systems [01:18:27] promising treatments are moving forward faster through molecular simulation [01:18:33] medicine is becoming more personalized through genome research [01:18:39] and patient outcomes are improving through robot assisted surgeries [01:18:44] our partners are using ai to accelerate science and better human health [01:18:50] advanced by amd [01:19:03] so look as you saw in that video amd technology is already at work across health care this is one [01:19:10] of the most meaningful applications you've already heard about some of the stories tonight [01:19:15] of how high performance computing and ai is one of the areas that i am most personally passionate [01:19:21] about is how you can bring health care there there's nothing more important in our lives than our [01:19:26] health and the health of our loved ones and using technology to improve health care outcomes means we measure [01:19:32] progress in terms of lives saved i'm very happy to be joined tonight by three experts who are leading the [01:19:39] way in applying ai to real world health care challenges please join me in welcoming sean mcclain ceo of [01:19:46] alimina and ula angkus head of molecular ai at astrazeneca [01:20:07] all right guys thank you so much for being here you can see there's a lot of excitement about health care thank you for the tremendous partnership you know sean at [01:20:14] absci you're using generative models and synthetic biology to design new drugs from scratch can you walk [01:20:21] through a little bit of how that works yeah thank you so much for having us here today biology is hard it's [01:20:28] complex it's messy drug discovery and development is this archaic way of going about discovering drugs [01:20:38] ultimately it's this trial and error process where you ultimately are searching for a needle in the haystack [01:20:45] but with generative ai and what we're doing at absci you're actually able to start creating that needle and [01:20:52] being able to actually engineer in the biology that you want being able to go after the diseases that have [01:20:59] large unmet medical need and being able to have the manufacturability the the developability that you want in the drug [01:21:09] we're actually able to start having precision engineering now because of ai with biology [01:21:16] just like apple is engineering an iphone or you all are engineering the 455s we're able to start [01:21:24] engineering biology and what is that actually doing it's allowing us to start tackling some of the [01:21:31] hardest most challenging diseases that still exist that have high unmet medical need where standard of [01:21:37] care is poor and at absci this is exactly what we want to tackle we want to tackle these hard challenging [01:21:45] diseases two of them that we're focused on at absci is intergenic alopecia so think common baldness [01:21:53] we actually have the opportunity in the not too distant future to have ai cure baldness wouldn't that be [01:22:00] incredible and not only that be able to focus in areas that have been neglected women's health for far [01:22:14] too long women's health has been pushed aside and we have a drug that we are developing for endometriosis [01:22:21] that affects one in ten women with the opportunity to potentially deliver a disease modifying therapy [01:22:29] for these women this is what ai and drug discovery is all about [01:22:40] and this wouldn't be possible without the compute partnership that we've had with amd [01:22:46] lisa you and mark papermaster invested in absci roughly a year ago and within that year we've been [01:22:53] able to scale the inference and be able to in a single day screen over a million [01:22:59] drugs in one single day that's incredible and additionally we're getting on to the 355s [01:23:05] and the memory there is going to allow us to contextualize the biology in a way that we [01:23:10] haven't been able to before and ultimately create better models for drug discovery the the future is [01:23:16] really bright in in ai and drug discovery that's fantastic sean well look um thank you for the [01:23:21] partnership we're really excited about all the work we're doing together now jacob illumina is really a leader in reading and [01:23:28] understanding the human genome to improve health how is ai helping in your work and you know [01:23:34] talk a little bit about you know what the impact is for the future of precision medicine yeah absolutely [01:23:38] is and i'm super excited to be here and we definitely share a deep passion for impacting health so looking [01:23:44] forward to everything we can do the two companies together both what we have done and what we're gonna do [01:23:48] um but uh and of course sean i'm rooting for that drug again so let me talk a little bit about illumina uh we [01:23:57] are the world leader in dna sequencing and the dna as you know is the blueprint of life which makes all [01:24:05] of us unique and therefore it's essential to be able to measure that for prevent diagnose and treat diseases [01:24:11] and in a simplified way you can think about the human genome at three billion letters um so that actually [01:24:20] is like a book with two hundred thousand pages in and that is in each of our cells now if there's just [01:24:27] one mistake spelling mistake in that book that can actually mean the difference between a long and [01:24:33] healthy life and a short and terrible life so accurate dna sequencing is extremely important but it's super [01:24:42] data and compute intensive in fact we are generating in our sequences more data than is generated on youtube [01:24:49] every day and therefore the relationship uh with amd is super important we are using your fpda and epic [01:24:57] processes in our sequences every day and that's the only way we can compute all that and translate [01:25:02] that into insights over the past decade our technology has already now been used as we talked about in [01:25:09] in drug discovery but also impacting healthcare today it's used for profiling terrible diseases like [01:25:16] cancer and inherited diseases and it's really uh very important to make sure and we are impacting a lot [01:25:23] of people's health out there and have saved a lot of millions of people's lives but we're just getting [01:25:26] started but biology is super complex so now and and our brain can't really [01:25:32] comprehend all that but the combination of using generative ai genome proteomics together is [01:25:40] pushed to completely change our understanding of biology over next period of time it will impact drug [01:25:45] discovery but it will also impact on how we prevent and treat early early diseases so really it will [01:25:53] change our way we think about longevity and healthier life and we can only do that with the collaboration between [01:25:59] us and all of us and all of us in the states and the whole ecosystem so i'm really excited about that [01:26:03] that's fantastic jacob and ola at astrazeneca you're scaling ai across one of the largest drug discovery [01:26:10] pipelines there is talk about how ai is changing the way you develop new medicines okay thanks lisa and [01:26:16] also thanks for the invitation and also say to at astrazeneca we really apply ai end to end from early drug discovery [01:26:24] to manufacturing to healthcare delivery and for us ai is not only about productivity it's a lot about innovation [01:26:33] how can we work in a different way how can we do new things with ai and one area that i'm personally very passionate about [01:26:41] is how can we deliver candidate drugs quicker with the help of generative ai so how we're working there [01:26:49] is that we train our generative ai model on all our experimental data that we have generated over [01:26:56] several decades and then we use those models to assess virtually assess in the computer which [01:27:05] hypothesis which ideas of candidate drugs might work or not and then we can assess millions of different [01:27:12] potential candidate drugs and then we take the best only the one that we think is really good [01:27:20] into the experimental lab and really validate the hypothesis there so we use our generative ai model [01:27:28] to to generate candidate drugs to modify them to optimize to really reduce the number of experiments [01:27:34] we need to do in the laboratory and we are applying a new way of working through the whole astrazeneca [01:27:41] small molecule pipeline and we see that we can deliver candidate drugs [01:27:47] 50 percent faster with a new way of working and also improve clinical success later and we can't do that [01:27:54] alone we need to do that in a collaboration so we collaborate with academia with ai startup and with [01:28:02] companies like amd and one very important area for us is hyperscaling because we have a lot of great data [01:28:10] and we really want to create the most optimal best models we can and there we work in a collaboration [01:28:17] with amd to basically scale our drug discovery and in similar flow so you can handle this large new data [01:28:26] set so basically we optimize the whole workflow with the help of amd that's fantastic ola look all of your [01:28:33] stories are really amazing and we're thrilled to be working with you to bring these things to life [01:28:38] now let's wrap up and think about you know what's the one thing each of you are most excited [01:28:43] about when it comes to how ai will improve healthcare and maybe jacob we'll start with you [01:28:47] yeah i'm just excited about the the time we are in this is the first time that you have technology can [01:28:52] create massive amount of data the first time you have the compute power and the generative [01:28:57] uh air models that will truly change our understanding on as i mentioned before biology [01:29:02] that will be translated into huge impact on health care ola so i think with ai we can really [01:29:11] transform the understanding of biology so we can go to not only to treat diseases but we should have [01:29:18] have the ambition as a community that in the future it can prevent chronical diseases that's fantastic sean [01:29:26] bring us home absolutely so to riff a little bit on what ola said i want to live in a world where we can [01:29:34] interact with people before they get sick where we can provide drugs and treatments to allow them to [01:29:43] continue to live their healthy life where they're metabolically healthy they have a full head of hair [01:29:53] and they have that vitality that we all look for being able to go from sick care [01:30:01] to preventative care to ultimately regenerative biology and medicine where aging no longer is linear [01:30:13] that's the world that i want to live in that ai is going to help us create it's an exciting time [01:30:28] i think we can all say sean we are super inspired i mean look this is what i heard we should expect [01:30:34] ai should help us predict sickness prevent sickness and personalize treatments such that we can get um [01:30:42] really extend lives and you guys are really at the forefront of it so it is our honor to be your [01:30:48] partner thank you each for joining us today and we look forward to really you know moving this frontier [01:30:54] forward over the next few years together thank you thank you thank you thank you [01:31:07] all right now we're entering the world of physical ai this is where ai enters the real world powered by [01:31:15] high performance cpus and leadership adaptive computing that enables machines to understand their [01:31:21] surroundings and take action to achieve complex goals at amd we've spent more than two decades building the [01:31:28] foundation of physical ai today amd processors power factory robots with micron level precision [01:31:35] guide systems that inspect infrastructure as it's being built and enables less invasive surgical [01:31:40] procedures that speed recovery times and we're doing it together with a broad ecosystem of partners [01:31:47] physical ai is one of the toughest challenges in technology it requires building machines that seamlessly [01:31:53] integrate multiple types of processing to understand their environment make real-time [01:31:58] decisions and take precise action without any human input and all of this is happening with no margin for error [01:32:06] delivering that kind of intelligence takes a full-stack approach high performance cpus for motion control and coordination [01:32:13] dedicated accelerators to process real-time vision and environmental data and with an open software [01:32:19] ecosystem developers can move fast and seamlessly across platforms and applications now seeing is believing [01:32:27] so to show how some of this work is unlocking the next generation of robotics please welcome ceo and co-founder of [01:32:34] generative bionics daniela pucci to the stage [01:32:47] hello danny it's great to have you your team is doing some amazing work can you just give us some background [01:32:54] about what you're doing lisa generative bionics is this is the industrial spin out of more than 20 years of [01:33:01] research in physical ai and biomechanics at the italian east of technology but when we look back [01:33:08] actually everything started from a simple but profound question if an artificial agent needs to understand [01:33:17] the human world doesn't it need a human-like body to experience it to answer this we built some of the most [01:33:24] advanced humanized platforms in the world icab for cognitive research then ergocab for safe industrial [01:33:32] collaboration and then we built ironcab the only jet-powered flying humanoid robot in the world [01:33:40] throughout the process of building these robots however lisa there has been one belief that has never [01:33:45] changed the real working technology is that one that amplifies human potential and that is built around [01:33:53] people not the other way around now this belief has become the mission of generative bionics but to make [01:34:00] it real we need compute that is fast deterministic and local a humanist for instance touch balances safety loops [01:34:06] cannot wait for the cloud that's why our collaboration with amd is for is so fundamental amd in fact gives [01:34:13] us a unified continuum continuum from embedded edge platforms such as reason ai embedded and [01:34:21] versatile ai edge running physical eye on the robot to amd cpus and gpus powering simulation training and [01:34:29] large-scale development so lisa one computer architecture from one partner end to end i like that i like that [01:34:37] a lot now look let's talk a little bit about your philosophy and approach to you know how are you building [01:34:44] these things and and what are your use cases we think lisa that now humanoid robots have to be elevated to [01:34:50] another level so our approach to physical ai is to build a platform around the humanoid robot and then the [01:34:58] platform is designed to achieve human level intelligence safe physical human robot interaction [01:35:04] and engineered into real products now let's start from the robot here we are really inspired by by [01:35:10] biomechanics in fact if you look at human movements they rely on fast reflexes we walk by falling forward [01:35:19] and our nervous system basically exploits our biomechanics so we are exploiting the same principles [01:35:25] into our humanoid robots then humans basically learn also through touch which is a primary source of [01:35:33] intelligence so we believe that human robots really need the sense of the sense of touch and finally [01:35:39] let's talk about the platform so we are developing an open platform around the humanoid robot to enable the [01:35:46] the next generation of humanoid robots so just to give an example the same tactile sensors that we [01:35:53] used for the humanoid robot we are basically used also into a sensorized shoe that is being used in health care to help [01:36:03] patients recover better and faster but more importantly the shoe acts as another robot sensor so that the robot has the feeling of whether or not and how to help the patient [01:36:14] the patient so lisa we are not building a robot we are not only building a robot we are not only building a [01:36:21] product we are building basically a platform to close the loop between humans and humanoid robots and enabling [01:36:29] what we call human-centric physical ai that's super cool denny now we are at ces and you know people like to see [01:36:36] things so what exciting news do you have for us so lisa we focus on a new product identity and our first human drop design [01:36:47] basically that defines our dna in terms of products gene one and i'm really happy to say that gene one today [01:36:55] has been ready to be released right now [01:37:06] so [01:37:06] so [01:37:16] so [01:37:18] so [01:37:21] so [01:37:25] so [01:37:29] so [01:37:31] so [01:37:31] so [01:37:33] so [01:37:37] so [01:37:39] so [01:37:41] is [01:37:48] is this gorgeous [01:37:48] is this gorgeous or what [01:37:53] danny tell us about gene one [01:37:56] so our vision is a future where humans remain in the center supported by technology [01:38:02] that's why we focus on building human robots that people can trust and accept [01:38:08] for us acceptability means beauty grace and safety gene one is italian by design [01:38:15] is it really italian yeah [01:38:19] and but basically what really says uh says gene gene one apart is touch [01:38:25] a distributed tight skin across the robot body allowed gene one to feel pressure contact intention [01:38:30] making and intention making touch a primary source of intelligence just to give you examples why this is so [01:38:36] important in factories that touch makes possible basically to [01:38:42] allow human robot collaboration and in healthcare that is going to be pivotal [01:38:48] basically a patient can hold the robot and he can feel how to help the patient in the best way [01:38:55] so this enables safer decisions and more natural interaction in the real world powered by md [01:39:01] compute platforms our first commercial humanoid will be manufactured in the second half of [01:39:07] 2026 and we are already working with industrial partners including a leading steel manufacturers to deploy [01:39:14] these robots in safety clear critical environments [01:39:17] lisa this is not science fiction and we are making happen thanks to you [01:39:22] thank you so much danny this is uh truly exciting we are super excited about what gene one can do thank you for being here thank you [01:39:36] okay now let's turn to one more demanding environment for robotics and automation and that is space [01:39:48] so [01:39:59] so [01:39:59] so [01:39:59] amd technology is powering critical space missions today from delivering satellite internet connectivity to remote communities to enabling autonomous autonomous [01:40:11] exploration of mars the moons of jupiter and beyond [01:40:24] on mars amd adaptive computing enables the perseverance rover to operate autonomously [01:40:39] that same technology is also powering robotic systems at nasa jpl and both the european and indian space agencies [01:40:47] delivering reliable compute in some of the harshest and most unforgiving environments [01:40:52] one of the leaders in space exploration and a company using amd technology to help build the next [01:40:57] generation of spacecraft and lunar infrastructure is blue origin please welcome john coluris [01:41:03] senior vice president of lunar permanence at blue origin to the stage [01:41:16] hi lisa john thank you so much for being here you know blue origin is doing just some amazing things [01:41:22] talk to us a little bit about your mission and what you're working on yes thank you everyone for having [01:41:27] me here i'm very excited to tell you about what we're doing um jeff bezos our founder likes to say [01:41:33] that earth is the best planet in the solar system it's sustained life for millions of years [01:41:40] and as we explore the solar system earth will be the origin of that life that pale blue dot that is earth and [01:41:49] that's why our company is named blue origin to protect that planet we want to move heavy industry [01:41:57] eventually off the earth as we look to build things such as solar power satellites in low earth orbit [01:42:04] settle the moon settle mars explore the asteroid belt we'll move on and we'll build that infrastructure [01:42:12] so that eventually millions of people will be living and working in space for the benefit of earth and [01:42:20] that starts with one person originally we had yuri gagarin and alan shepard explore then the apollo [01:42:28] astronauts then just recently the international space station celebrated 25 years of continuous human [01:42:35] space in space the next step for us is lunar permanence and our business unit that i'm lucky [01:42:44] enough to be a part of is named lunar permanence specifically so that everyone knows immediately what [01:42:49] we're trying to do is establish a permanent presence of humanity on the moon and that requires reliable [01:42:57] repeatable low-repeatable low-cost operations and reliable and repeatable low-cost equipment and [01:43:02] vehicles and amd is a critical partner of ours to make that happen well thank you so much for that [01:43:09] john and look talk a little bit about why high performance computing is so important in your work [01:43:15] and especially as your missions are getting more complicated certainly so space is the ultimate edge [01:43:22] environment the flight computers that we build are the heart and soul of our vehicles that compute stack [01:43:29] needs to be reliable deterministic and resilient it needs to survive the environment of space and what [01:43:37] that means is we have mass constraints power constraints radiation considerations and the amd embedded [01:43:47] architecture allows us to reduce mass save power on these vehicles and tolerate the demanding radiation [01:43:56] environment of deep space and i think when we you know think about all of this looking ahead you know [01:44:03] talk a little bit about how ai is playing a bigger role in your future missions certainly so ai's impact [01:44:10] on earth-based systems is well known in fact i've got to say lisa and i kind of surprised you earlier today [01:44:18] amd has been a phenomenal partner of blue origin we only a few months ago started to talk to amd about [01:44:26] using the versal 2 in our flight computer stack and within a few months the amd team and the blue origin [01:44:33] team worked tirelessly and were able to get ship us units that we were then able to incorporate into [01:44:40] development flight computers we've now in a couple of months built the development flight computers that are [01:44:46] flying in our vehicle test bed and those will eventually power our mark ii lander that mark ii lander will [01:44:54] land astronauts on the moon as early as 2028. in fact it was so impressive we had a team of blue origin [01:45:02] engineers working over the holidays and we took the entire flight computer stack and we're able to [01:45:07] successfully simulate a landing on the moon this has saved months and months of schedule now you take that [01:45:15] to the ai use and how important it is for us right now blue origin ai use on earth is critical every [01:45:24] employee at blue origin has access to ai tools whether it be for design for analysis for just basic pinging back [01:45:32] and forth ai has sped our development process so quickly that we're now looking how do we bring this [01:45:39] to space flight and so what that means for space flight for us that's the next great step where ai becomes [01:45:48] a complement to the astronaut a co-pilot if we will identifying landing sites looking out for hazards being [01:45:55] able to do that level of compute in a real-time environment is critically important to us for me [01:46:01] personally though uh you think of edge ai what is really interesting is as we go to explore the solar [01:46:12] system radio astronomy has been a passion of mine and what radio astronomy is is you're looking for weak [01:46:19] signal radio frequencies that are being emitted throughout the universe the problem we have is [01:46:26] that earth is a great emitter of radio frequency noise and interference so it's hard to identify that [01:46:32] the far side of the moon provides a natural shelter a barrier to that noise so if we could land a mark [01:46:40] one vehicle on the far side of the moon we could start to explore this untapped radio frequency environment [01:46:50] if we have edge ai we can now utilize that to do the deep exploration to actually identify where [01:46:58] we should be looking next because relay back to comms doing this latency really hurts our ability to [01:47:04] explore so by having it on the far side of the moon the mark one vehicle with edge ai will tell us land the next [01:47:12] vehicle here to optimize your exploration that's really what excites me i mean that's super super cool [01:47:20] look john um we're honored to work with you i think it is an incredible mission that you have [01:47:25] thank you so much for the partnership and we look forward to what you do next absolutely thank you very much [01:47:30] thank you okay guys now let's turn to our last chapter of the night science and the supercomputers used for [01:47:43] the most advanced scientific research we're incredibly we are incredibly proud of our leadership in high [01:47:59] performance computing and we have continued to push the bleeding edge of performance here [01:48:04] we're actually seeing a convergence between traditional high performance computing systems and ai [01:48:09] as we bring together the best of both of these worlds today amd powers the two fastest supercomputers [01:48:15] in the world and more than half of the 50 most energy efficient systems these systems are using massive [01:48:22] amounts of compute to solve previously impossible problems in finland the lumi supercomputer has cut [01:48:29] climate model update times by more than 85 percent enabling earlier warnings and better preparation [01:48:35] for extreme weather events energy giant eni is using an amd powered supercomputer to develop longer lasting [01:48:42] batteries and cleaner fuels at oakridge national labs the world's first exascale supercomputer is running [01:48:49] orbit 2 a hyper resolution global model that allows us to predict unprecedented forecasting detail with [01:48:57] nearly 99 percent accuracy and at lawrence livermore national labs the world's fastest supercomputer el capitan [01:49:05] is modeling how viruses might mutate and evolve enabling scientists design to design [01:49:11] more resilient antibody treatments and respond faster to future pandemics going forward there's a lot more [01:49:18] that we can do to power the future of scientific discovery we are actually working very closely with the [01:49:24] u.s department of energy and america's national labs as part of the genesis mission [01:49:29] genesis is a national program launched late last year to accelerate the convergence of ai [01:49:34] supercomputing and quantum computing together with oakridge national labs we recently announced two new [01:49:41] supercomputers that are part of the genesis mission the lux computer that is the first [01:49:45] dedicated us ai factory for science that will come online early this year and discovery [01:49:51] the next flagship supercomputer planned in 2028 genesis is the most ambitious public private technology [01:49:59] initiative in decades leading this historic effort is michael kratios who has shaped national policy at [01:50:06] the highest levels as a former u.s cto and under secretary of defense for research and engineering please [01:50:13] join me in welcoming the president's chief science and technology policy advisor michael kratios to the stage [01:50:21] Thank you. [01:50:31] Michael, thank you so much for being here. I know just how busy you are. You've described [01:50:36] this Genesis mission as a real moonshot with the largest mobilization of federal scientific [01:50:41] resources in decades. Can you talk a bit about why Genesis is so important? [01:50:47] The Genesis mission is a great example of how President Trump has moved fast in less than [01:50:52] a year in order for the U.S. to lean in and win the AI race. Genesis is the largest marshalling [01:51:01] of federal scientific resources in recent history and at a scale and an urgency at the Apollo [01:51:08] mission or even the Manhattan Project. We are bringing together the unmatched power of our [01:51:13] national laboratories, supercomputers and the nation's top scientific and innovative minds [01:51:19] with the goal of doubling the productivity and impact of American science within a decade. [01:51:26] This whole of government approach represents a historic mobilization of resources, tasking [01:51:32] the Department of Energy to integrate its world class supercomputers and data sets into a unified, [01:51:38] closed loop AI platform. Integrating this data, the Genesis mission leverages [01:51:44] the power of AI to automate experiment design, to accelerate simulations and generate predictive [01:51:50] models that accelerate federal R&D productivity. Priority areas of focus include the greatest scientific [01:51:57] challenges of our time that can dramatically improve our nation's economic and national security. [01:52:03] These span biotechnology, critical minerals, nuclear energy, space exploration, quantum, [01:52:09] semiconductors, microelectronics. And a few weeks from now, a few weeks ago, we announced the [01:52:15] first wave of industry partnerships to the Genesis mission and that included AMD. So thank you for that. [01:52:20] As a next step, we're working towards bringing even more federal resources into the Genesis mission. [01:52:25] And this is going to include a variety of agencies, including the National Science Foundation, [01:52:29] National Institutes for Health, and the National Institute for Standards and Technology. [01:52:33] Well, look, thank you, Michael. We are very proud to be part of Genesis, super excited. If you look back, [01:52:41] you know, so many of the technologies that we have today really started with long-term public and private [01:52:45] partnerships. So where do you see Genesis actually making the biggest impact beyond science? [01:52:51] Well, through Genesis, we will create the world's largest and highest quality scientific data sets to [01:52:59] train the next generation of AI systems, pushing them beyond their current mastery of language and code [01:53:06] into the realm of science. Now, as you can imagine, this will lead to tremendous spillover effects [01:53:11] across healthcare, drug discovery, energy, and manufacturing. Fundamentally, we are seeing a massive [01:53:18] shift in America's science and technology enterprise. We are now at a place where the U.S. government, [01:53:25] private sector, and universities together are investing over a trillion dollars in R&D every year, [01:53:31] with the private sector leading the way by carrying out two-thirds of that R&D alone. [01:53:35] Now, the Genesis mission understands this and leverages the full strength of that entire ecosystem. [01:53:40] That's wonderful, Michael. Now, we have talked a lot about how important it is for the U.S. to lead in AI. [01:53:47] Can you talk about what are the biggest things that we must do to get right such that we lead in AI? [01:53:53] Absolutely. There are three strategic priorities the U.S. needs to get right, [01:53:58] as laid out in President Trump's AI action plan. The first is we need to remove barriers to innovation [01:54:05] and accelerate research and development. We are already at work looking for regulatory roadblocks [01:54:11] innovation and seeing where we can update or remove them entirely. This effort will ensure the U.S. is a home for [01:54:17] the next great technologies to be created and to be commercialized. Next, we need to get AI [01:54:24] infrastructure and energy production right. We've taken significant actions to streamline [01:54:29] permitting to data center construction and support all forms of energy, including advanced nuclear reactors. [01:54:35] Looking beyond our borders, it's all about AI diplomacy and exporting American technologies to the world. [01:54:42] The U.S. government is underway in establishing the American AI export program to bring American [01:54:49] innovators and innovations to our partners and allies around the world. The Department of Commerce will be [01:54:55] issuing an RFP this month seeking proposals to create a turnkey AI stack, including everything from [01:55:01] infrastructure and chips to models and applications. Last but not least, another strategic priority for [01:55:08] President Trump and First Lady Melania Trump is AI and education. I talked earlier about winning the [01:55:14] AI race. Focusing on AI and education is about truly winning our AI future today. It starts by helping [01:55:22] parents and teachers and students navigate AI opportunities and challenges in the classroom. [01:55:28] I am thrilled that in a matter of months, we are seeing tremendous participation with other 5,000 [01:55:34] students and 1,000 educators across all 50 states signing up for the Presidential AI Challenge. [01:55:41] Now, submissions close on January 20th, so please visit AI.gov to participate. And also look for these regional [01:55:48] competitions that are ultimately going to culminate a championship at the White House this summer. [01:55:54] We've also secured over 200 pledges from leaders like AMD for free AI educational resources, [01:56:00] including apprenticeships, access to AI top models, and curricula for so many teachers around the country. [01:56:06] Michael, look, we're incredibly proud to support the AI education pledge and really to help expand [01:56:12] access to AI education with more hands-on opportunities for students to learn and build. [01:56:18] That's why we've actually committed $150 million to programs that bring AI into more classrooms and [01:56:24] communities across the country. We're investing in the next generation of AI research and talent. [01:56:33] We're building research collaborations with more than 800 educational institutions around the world, [01:56:40] including many of the top engineering and computer programs. And we're also committing to developing [01:56:45] coursework to promote our open ecosystem. So we're offering free online AI courses to reach over 150,000 [01:56:52] students this year. So, Michael, I want to say thank you for your leadership on this topic and the First Lady's [01:56:59] leadership. I can tell you that it certainly is making a difference in galvanizing the industry. [01:57:05] Now, before you go, I have a very fun thing for us to do. It is really a moment for us to highlight [01:57:11] some really amazing work as a direct outcome of the AI education pledge. So a little bit of background. [01:57:18] We recently partnered with Hack Club on a nationwide AI and robotics campaign. More than 15,000 high [01:57:25] school students signed up with the top teams coming together in Silicon Valley last month for an in-person [01:57:30] hackathon to bring their designs to life. It's actually incredible to see what these students were [01:57:36] able to build in just one weekend. You can imagine it was a little bit competitive at this hackathon. [01:57:42] And as part of the recognition, we invited the top three teams to be here at CES, right here in the front row, [01:57:48] so they could experience the biggest tech event of the year firsthand, and we could congratulate them in person. [01:57:54] So let's give them a big round of applause. [01:58:06] And to tell us a little bit more about their project, I'd like to invite the hackathon gold medal winners, [01:58:11] Emmy McDonald, Ruzana Gaboyan, and Afia Ava of Team Arm Tender to the stage. [01:58:28] Thank you. All right, you guys are amazing. Congratulations on the incredible work. Now, [01:58:35] before you talk about the project, can you just share a little bit about yourselves? Like, where are you from, [01:58:40] and when did you start coding? Sure. I'm Emmy, I'm 17, and I'm from Chapel Hill, North Carolina. [01:58:47] I started coding when I was around 12, and I joined Hack Club when I was 16. [01:58:54] Good evening, everyone. It's great to see you all. My name is Ruzana Gaboyan, [01:58:58] and I am a 17-year-old student from Cleveland, Ohio. I started coding when I was 12, [01:59:03] and I joined Hack Club as soon as I turned 16. Hi, everyone. My name is Afia. I'm 18, [01:59:10] from Beaver Dam, Wisconsin. I started coding about two years ago through Hack Club. [01:59:15] That is fantastic. Now, tell us a little bit about your project. [01:59:18] Of course. Together with Ruzana and Afia, my Hack Club teammates, we built an AI robot barista. [01:59:25] It's a robotic arm that autonomously serves beverages using a motorized wheel that spins to select a soft [01:59:29] drink. We trained a single unified vision language model to multitask using the AMD developer cloud with [01:59:36] MI300X GPUs. The robotic arm runs entirely on an AMD Ryzen AI laptop using three cameras, [01:59:43] and we came to the hackathon with no previous AI training experience. [01:59:46] Now, can you believe that? No previous AI training experience? And this is what they did? [01:59:56] You guys are clearly doing great things. Just give me an idea. Like, what are you most excited about [02:00:01] working on next? Thanks so much, Lisa. So my mom actually works at our local fire department. She's in [02:00:07] the audience. Hi, mom. And she is a former firefighter. One thing that we noticed when training Arm Tender is [02:00:12] that it was able to capture complex human behaviors. Like, it was able to try again to grab a can after it [02:00:18] missed it without us programming that specifically. I want to build a robot that can be used in firefighting to go [02:00:24] into building fires and case the building before firefighters go in. And the complexity of motion [02:00:28] that we saw AI exhibiting could make that possible in a way that wasn't before. I love that. What do you guys think? [02:00:40] Well, look, we are all about encouraging and inspiring young people to pursue their dreams. And tonight, [02:00:46] we actually have a special surprise for you guys. So AMD is awarding each of you a $20,000 [02:00:52] educational grant to invest in your future as innovators to help you keep building. Thank you so much. [02:00:59] Thank you so much. [02:01:04] Look, you guys are just a great example of what the AI education pledge is all about. Because [02:01:09] what you've created, it's clear that there's so much we can do. So congratulations to all of you. [02:01:15] And Michael, thank you for being here and really helping really bridge all of this together. We [02:01:20] appreciate everything you're doing for the country and for the industry. Thank you so much. Thank you. [02:01:26] Thank you, guys. Congratulations. [02:01:34] Look, it's been fantastic being with you tonight, but it's time to wrap this up. So [02:01:39] I hope you all saw tonight what I see every single day. This moment in tech not only feels different, [02:01:47] AI is different. AI is the most powerful technology that has ever been created. And it can be [02:01:54] everywhere for everyone. We're entering this era of yada scale computing, where the deployment of more [02:02:01] powerful models everywhere will require a massive increase in the amount of compute in the world. [02:02:06] Meeting that demand will take a broad portfolio of solutions from the largest systems in the cloud, [02:02:12] to AI PCs, to embedded computing. And just as important, it takes an open ecosystem built on [02:02:18] industry standards. That's what you saw on stage tonight. We wanted to bring you the entire spectrum [02:02:25] from amazing technology to deep co-innovation with industry leaders across the ecosystem to very strong [02:02:33] public-private partnerships. All of us are working together to bring AI everywhere for everyone. [02:02:39] On behalf of the 30,000 AMDers around the world, we're proud to be building the future together with all of [02:02:46] you, because the world's most important challenges can only be solved by bringing the industry ecosystem [02:02:52] together. Thank you for joining us tonight and enjoy the rest of CES 2026. [02:03:08] Thank you. [02:03:38] Thank you. [02:04:08] Thank you. [02:04:38] Thank you. [02:05:08] Thank you.