Intel's Crazy Plan for AI Chips IS WORKING! (Supercut)

hey Sam thanks so much for joining us of course and uh Hey a couple of quick questions you know the first one is 7 trillion what's this all about anyway look uh as as I'm sure you all know you shouldn't believe everything you read in the Press really we do think the the kernel of Truth is that we do think that investing a lot of money in AI compute energy data centers is going to be important to deliver the the amount of services people want and the tools that we all are going to really I think get a huge amount of value out of to help create better Futures mhm so there's a lot of investment to be made here I think so but it's across a lot of stuff it's not I mean it's a this is like a worldwide investment to help many many different efforts many different people and not just chips of course but sort of the whole infrastructure stack around it okay second question you and I are both Tech optimists a little bit so why should we be Tech optimists about AI I think this is going to be one of the greatest tools humans have yet invented if if you think about what we we can do with more intelligence available to us U more abundant intelligence cheaper prices things that humans on their own are not capable of uh we will all be able to invent new things for each other that astonish us that's how the future gets better making things more intelligent I like it when you say that uh third question our founder Andy Grove famously said the paranoid survive right paranoid about yourself your success but also paranoid about competition and things that could go wrong and there are those who are a little bit paranoid about AI today what do you say to them I think people are right to be cautious about AI I think this is going to be an extraordinarily powerful technology I think it's going to reshape significant parts of the way we live and work and that our economy runs and part of our strategy what we call iterative deployment is that we want to put these things out into the world early so that people can adapt and think and provide input and that uh we can be ready for the man of change that will come over time so I think that's great yeah getting them out there early people test them that's how things get better yeah making them itely better well Sam you've honored Us by joining Intel Foundry date right the partnership that we have with you is tremendous we're going to do great science together great AI machines together and uh just fundamentally make the world a better place thank you so much thank you for having [Music] me Moors law alive and well until the periodic table is exhausted we are not done with Mo's law we Bend physics we create molecules we do amazing things and Moore's Law is alive and well but we have to do it a different way we have to enable globally resilient sustainable and trusted Supply chains and at the end of our conference today I'm joined by Sam Altman and he's going to challenge the capacity needs of the industry and the system of chips needs new capabilities new tests and packaging capabilities and that's what a systems Foundry does and as we describe this Vision to become the number two Foundry of the world we realized there's only a few companies that can do this that have the capital capacity that have the R&D the longevity to go do this and now as we're in an economic downturn ah welcome to semiconductors and then the AI explosion and the cyclicality of the industry and we've seen the Geo instability and active Wars in Israel and Ukraine and the tensions and Taiwan on Straits this is anything but a resilient supply chain today and for that get to be a large meaningful the second largest Foundry but become the world's most sustainable Foundry the world's most resilient Foundry because that's exactly what the world requires but stunningly you know in 1990 80% of the semiconductors are built in us and Europe today 80% in a small concentrated area in Asia you know we've seen this long steady decline right in terms of our supply chains for the world nothing should be reliant on a single port to a single country a single place in the world we need resilience a resilient access to supply chains and capacity in the right regions at the right time and thus the choice the opportunity to drive systemic change in where and how we drive the most important aspect of our future where the technology Supply chains are what aspect of your life is not becoming more digital well everything is your Healthcare your financial you're social and with that we simply call it silicon ofy silicon and the economy becoming fused together in an inextricable way and as stewards of Moore's Law we see this Relentless pursuit of more efficient more capable more scalable Computing and for that we've been on this journey we also announced that we're we're going to get five nodes in four years we're going to do something unheard of in the industry to return Intel to process technology leadership and while we're not finished today we see the end is soon in front of us on that journey and Intel 7 shipping and ramping and volume Intel 4 with our cor ultr launch shipping and ramping and volume Intel 3 is production certified and we'll be with our server products launching in the first half of the Year going into volume production so with this we've gone on Incredible Journey but then it continues into what we call the angstrom era and for this Intel 20a and Intel 18a the adoption of ribbon fed a new transistor structure of power via power delivery technology the Embrace of the first major new transistor re architecting since 12 this is a Mona Lisa no no this is a rembrand no no no I think it's a Michelangelo right sculpted in Silicon right for these truly are works of art and I am thrilled for the progress and with that the finishes 18a we've already sent into Fab our first 18a products and this is a test tiip for Clearwater Forest so I'm thrilled this is what we call a family photo so kids come to Papa here we go five nodes in four years and I do want to announce describe and give a moment to our latest newest 18a customer my decades long friend Sacha Nadella speaking for Microsoft as the newest 18a customer let's hear from Sacha now thank you so much Pat it's great to join you at your launch event it's clear that we are in the midst of a very exciting platform shift that'll fundamentally transform productivity for every individual organization and the entire industry uh to achieve this Vision we will need a reliable supply of the most advanced high performance and highquality semiconductors and all of us at Microsoft are committed to supporting Intel's efforts to build a strong supply chain right here in the United States that's why we're so excited to work with Intel Foundry services and why we have chosen a chip design that we plan to produce on Intel's 18a process uh we look forward to sharing more details in the future and I can't wait to see all that we will be delivering together for our customers in the years ahead thank you so very much thank you Sacha so can I tell you more about what comes after 18a today we're announcing that we're extending these nodes we're adding major and minor uh nodes to it a combination of older and Leading Edge nodes to ensure our customers have access to the process technology they need today we are announcing Intel 14a for the first time you think about this like four 1.4 nanometer technology but Intel 14a venturing deeply into the angstrom era 14a first processing but we're also announcing that we're extending our nodes as you see on here adding P noes enhancements to those existing adding performance capabilities adding t- noes uh through silicon via new feature enhancements with e noes on the road map filling out that road map of capability today we're announcing for the first time Intel 16e enhancements to our Intel 16 technology as well so we're filling out that full set of nodes the road map that we have to go beyond it but as we've seen as we've gone through this period of time this AI era explosion Wafers are cool packaging has gotten to be really cool so Intel Foundry offers a broad set now of advanced assembly and test uh Technologies Intel 25 years ago drove the standardization of organic packages now we're driving the next generation of glass based packages and with that the ability to directly inter with Optics and wave guides directly into the package construct for the most advanced system capabilities as well you know and Intel Foundry has added a number of additional AI customers to our portfolio of packaging offerings as well AI era needs Advanced Wafers but it even needs more systems and packaging capabilities an area that Intel is the clear leader in and this now includes as customers some of the largest AI leaders in the world as I include my time on stage today is a day 3 years in the making and I couldn't be prouder of the team at Intel that is ralling behind this rebuilding of this iconic company and you're going to hear from a number of those leaders today bringing together the world's first system Foundry capabilities for the AI era and with that it's my pleasure to introduce to the stage a friend for decades now and the the zealous leader of our Foundry Services none other than Stu pan please welcome Stu to the St our people can sense and feel this momentum you can hear it in their voices you can see it in their actions they know that you don't need to be in Taiwan to build the world's most advanced semiconductors theyve realized this five NOS in four years this audacious goal that Pat laid out you can see the check marks across all the boxes here we're ready to earn the right to be your Foundry supplier nobody's going to give us that we want to earn it so that's one perspective I want to offer up a second perspective and this comes from Chris Miller the author of the book Chip War this is a New York Times bestseller fin Financial Times book of the Year took Chris 10 years to write it now Chris in a speech in October said this quote to the employees said quotequote Intel is the most important company of the last 50 years so I called Chris I said are you okay if I say this publicly it's yeah okay go ahead good publicity for the book I came back to Intel 3 months after pat did because I believe this quote to be true and I believed it long before Chris said it and what you're going to hear in the next 40 minutes or so is all the reasons why Chip War season 2 is about to begin so let's start from the beginning Gordon's original paper was called cramming more components on integrated circuits it was published in 1965 every time I read it I'm stunned by how pressing he was in the last section of the paper Gordon remarks we come to a Day of Reckoning quote unquote it may prove to be more economical to build large systems out of smaller functions which are separately packaged and interconnected and this is why I believe we're in the area of the systems Foundry you can no longer do just monolithic devices you have to break it up you have to adust for thermal profiles you have to address costs you have to have flexibility and this is all driving a new level of systems thinking Into The Foundry business in this systems era you not only have to have Open Standards between devices you have to have standards on the device think about the fact of to do a training model day requires 100,000 CPUs all running and concert all in the same data set the next round of training models will require a million CPUs that's something that requires standards everywhere from chips inside to chips outside this is what's necessary for our customers succeed and this is the fundamental reason why Intel is a systems company turning into a Foundry not the other way around so now that we described system thinking let's look at how the original Foundry model was created now since CC way has been kind enough to mention Us in his last few earnings calls I thought i' return the favor and talk about tsmc in my presentation now this was a presentation that Morris chain gave at MIT MIT last October and in it he describes The Foundry model he says it's research and development it's waer fabrication it's Advanced packaging and he says in the red line that's what tsmc does and said in the blue line Intel everybody else does everything else you might even call that an IDM 1.0 kind of manufacturer now tsmc has been incredibly successful with this model discipline execution discipline strategy consistent Innovation but to quote Bob Dylan The Times They are changing let me tell you why this there's an idea of Systems Technology co-optimization where you look at application workload Software System architecture memory interconnect all these different things what happens today is people focus on their layer and maybe the layer up on top of it in fact this is really what a classic Foundry does today but we're now in the realm of the exponential Sam alman's going to come out later on this afternoon and talk about he doesn't have enough capacity to do what he wants to do you have to look at all of these combinations you can get a couple of things right but to do a system that coordinates the activities of solving a training model across 100,000 CPUs requires you to get all the gear ratios right if you're mismatched in memory if you're mismatched in networking you wind up throwing away valuable cycles and valuable resources so we got to get 100x more out of what we're doing so we described the evolution now let's build out the strategy and talk about what's the revolution in all this we like to think about this in three basic layers first off we have to be a world-class foundary right Rowan Chen the COO of Qualcomm talks about the fact that silicon speaks and silicon speaks in four different ways performance power area and cost without that you're not in the business the next layer of the pyramid is all about resilient sustainable suppli this need to have capability around the world to build this and to build it in a sustainable fashion the top layer of the pyramid is this idea of systems of chips we're putting system inside a chip and we're working with our Foundry Partners to create systems of chips so let's walk you through what that means now I talked earlier about this Day of Reckoning what that means it's happening now we can no longer do designs at a monolithic level we are now at retical limited designs design sizes that are 800 square millim of silicon because we have to move Beyond Ral limits and thermal constraints and by the even cost constraints because when you're building these really big die sizes they're really expensive isn't there a better way to take advantage as Pat talked about to take smaller tiles on the more advanced notes get better yields out them package them together have more flexibility to do all that really requires if you will a system amount of chips so you can see in the animation here how we build it out and this is literally how we're going to build out in the factory the idea of a substrate the idea of Base dieses the idea of logic tiles the idea of IO tiles on the side why do this it gives our customers the ability to optimally trade for what they need for their design the things that you do for a training engine will be different than what you do for an inference engine and only by having all these levers to go pull can you get this done we learned a lot through a device called ponovo or as The Branding people call Intel data center GPU Max series it's ANC it's 100 billion transistor s so it's dozens of chiplet tiles 47 of them it's multiple suppliers by the way we coexist with tsmc in the same package right we develop testing techniques to go off and do that we do this idea of singulated die test what does that mean it means every single die that goes into that package is a known good die why is that important because you want every one of them to be good the assembly test shield on this device is 95% plus it is the Super Bowl of integrated design now what do you do with it well if you're argon National Labs and our partners at HP you build a really big supercomputer and they built a computer that was 66,000 Pacos 20,000 Sapphire Rapids and it looks like this okay what do you do with something like that well you solve some really hard science problems if you want to model the airf flow across the wing you can do that on a workstation if you want to model the air flow across a plane you do it on this if you want to model Fusion reactions which are pretty tricky things to model you do it on a device like this if you want to model if you want to model cancer curing drugs at the molecular level you do it like this it's 600 tons of compute it's four tennis courts it is the weight of an Airbus it has 300 miles of optical cable it takes 34,000 gallons a minute to Cool by the way your faucet at home it's a gallon a minute 34,000 gallons a minute so when we talk about how to design the stuff we have to find ways to make this more power efficient to make this more cost effective this today is the second fastest super computer in the world by the way at 100,000 CPUs roughly to handle the demands of AI we're going to go far beyond that and that's why this idea of scalability is so important now let's go through the top of the pyramid this idea of systems of Chip systems of Chip require as Pat mentioned great packaging technology we've created standard called UCI which allows chipto chip connectivity think of it as the PCI Express of what we did back in the mid 90s this idea that you can add mix and match by the way you can mix and match different Foundry suppliers we like that because we're sort of an underdog in all this some of our competitors not so much but our customers want this kind of flexibility so as you can see from the rendering what we're doing is taking systems on a chip and Building Systems of chips why do you need systems of chips because it's the same problem I mentioned earlier you need to train models with 100,000 CPUs now and a million potentially down the road and maybe up to 10 million when you're moving a coordinating data across all of these devices you need to have standards and connectivity and that's what we provide and really when you think about it what is what's it going to take to bring AI everywhere how do you make it cost effective how do you make it capital efficient yes yeah I know Sam's asking for trillions but we want to make sure he's spending all that money in the most costeffective manner possible so let's break that out you have data center chips doubling year on-ear but the efficiency needs are the things that are really eye-catching New York Times ran an article that AI could soon need as much electricity as an entire countries so I'm sure you're curious like which countries Sweden the Netherlands Argentina if you were to run all of the AI servers that market estimates have on djx2 dx100 those kind of devices you would take 85 to 134 terawatt hours terawatt hours by the way the great state of California its entire power generation capability to day is 30 tratt hours for the entire state so bringing AI everywhere is going to require us in our Foundry Partners to figure out how to do this cost effectively this is what this road map is going to take and this I think is what makes us different than other Foundry approaches you have to start off across first off with table sticks as I mentioned you have to start off with a great process add to that packaging which is we believe a unique differ for us why we build a lot of server parts and we're taking everything that we learn from our server business and offering up to our Foundry Partners these are table stakes and this is what it takes to be a Foundry player we have this planned out literally across all these dimensions for the next 5 years let's talk about cooling I mentioned that that argon needs 34,000 gallons a minute to cool it off the next wave of devices are going to have to be immersed and cooled and today Intel Zeon product line is the only product that offers an immersion cooling warranty immersion cooling allows us to deliver power much more effectively in a data center we're going to take what we've learned there and we're going to offer that up to our Foundry customers so that when we start looking at 2,000 watt devices 5 years from now we're going to have a way to cool those memory how do perhaps do we put memory on the device itself so it's more computationally cost effective and over the next five years you're going to hear us talk a lot about new technologies and new ways to increase memory bandwidth why decreasing the need for energy consumption a five-year kind of look interconnects interconnects between chips the idea of having high-speed series highspeed interconnect all the things that you expect from a Foundry provider is what we're going to give you and we're going to not plan out for just what's out there today keep in mind as a standards company we do this across all different standards and it's our job to make those standards available to all of you and last you know networking think about a net card that's capable of handling demands of AI modeling what you do with ethernet today isn't good enough so working closely with a number of ethernet standards Partners to figure out ways how do we make ethernet more capable to develop the idea of systems of chips and ethernet provides that lowcost high bandwidth potential to go off and do that add to that fonics add to that Technologies beyond that here again it's another 5-year road map this morning we're announcing a new partnership with arm merging business initiative how do we take advantage of all the programs that arm has to offer to bring design capability design education out to all of their customers we're doing this with arm we will make co-investments we'll do joint programs we'll provide shuttles at scale arm will provide IP at scale and this is how we're going to fuel this next wave of innovation and it's truly exciting so I'd like to now is invite R to come up on stage with me hello sir so good to have you here so Renee first off in what kind of universe would you have ever thought that you'd see arm and Intel standing together but you we rapidly came to the realization that 80% of the way for tsmc runs has an armed device in them there is no way you can be in The Foundry Business Without A partnership with arm and I think you'd like to probably tell us about a few of them starting with perhaps the announcements you made this morning yeah well thank you for having me as you said this is a bit of strange bedfellows I was trying to think of a parallel that I might give relative the story and only thing I could kind of think of is when Walt Mossberg asked Steve Jobs what it was like to see iTunes run on Windows and I think he said it was like an ice waterer in hell but I won't go that far I won't go that far the announcements that we made earlier today was around our neoverse product line which is the product line that we use for the data center which has just been exploding it was in a very very high growth trajectory prior to the AI wave and now it's become even stronger the Nee V3 that we announced today which is 50% faster than the neoverse N2 and then our N3 which is 20% faster but also much more efficient and when you think about these AI data centers which are pulling hundreds of megawatts and and more efficiency matters so yeah today was a big day for arm we're working on on on Cutting Edge technology when you talk about 18a and and system foundering the packaging this is uh the tip of the spear in terms of innovation and you guys have been terrific uh it has been an absolute joy to work with your groups the level of engineering engagement the depth of the technical discussions uh the information that we get uh we would not be able to announce the partnership that we've forged without it we talked a little bit about Faraday let's talk a little bit about what the world faray plays in neoverse and CSS and why what maybe to find for the a what is CSS and why is it so important for everybody here in the audience yeah so we announced uh a new strategy some months ago around what we call compute subsystems and the way to think about this is essentially rather than arm delivering blocks of Ip uh a mesh Network a memory controller the CPUs themselves we deliver a full subsystem fully verified completed that is if you want 64 cores 96 cores 128 cores for a CPU we deliver everything in terms of that system validated verified and it will work now one of the big benefits of this is simply the fact that time waits for no one the classic classic quote the amount of time it takes to design these s so's is really really hard and it's really really long if we're delivering final IP to a customer they still need to put all those pieces together if we can put all of that together for them prior at the same date they would have got the block of Ip that's a huge huge benefit and then when you add on to it the processing cycle times that are getting longer and longer I know you guys are doing your best but you know more euv steps means more complex time through the Fab and that just means that ultimately the processing times are long you have complex packaging you know that adds a lot of time so anything you can do on the front end to benefit the design time is is really beneficial so we announced this program called uh arm total design of Which Intel is a partner and in that model customers can come to their partners and work with those folks to get their design out and that's what Faraday is so Faraday is basically doing that they'll be able to put their IP together with our blocks and something that just end quote works and everyone wins the product's out faster it's compliant and it's going to work outstanding or Rene thank you for coming and joining us on today today pleasure totally appreciate the partnership and everything we're doing together everything we're going to do together thanks again thanks all [Music] so to finish out my talk I'd like to go back to where I started this is an ad rolled at the Super Bowl 1997 the year the Green Bay Packers beat the Patriots the halftime show was The Blues Brothers and James Brown and it was eight years after Taylor Swift was born I had to do it you can't associate superos without Taylor Swift and this commercial was iconic but you know things have changed a lot over the last several years and that change is only only going to accelerate as AI impacts your work your business your life how AI impacted the design cycle AI improves the capability for your engineers to do so much more with so much less effort so we decided to do with this commercial was take a generative AI view of it so this commercial no buy people were harmed we did 100% AI rendered it was done by very creative director Dave Clark out of Los Angeles and he's a big advocate of using AI tools not to take away creativity but to amplify creativity so as you leave our session today remember this that the only imitation to what we can do with AI is your imagination thank [Applause] you

Channel: Ticker Symbol: YOU

Views: 222,241

Keywords: openai, chatgpt, sora, sam altman, msft, msft stock, microsoft stock, arm stock, tsmc, tsm stock, intc, intc stock, intel stock, intel foundry, nvda, nvda stock, nvidia stock, amd stock, ai chips, semiconductor stocks, intel foundry direct connect, intel keynote, ai stocks, best ai stocks, top ai stocks

Id: Bik2nZksnh4

Length: 28min 18sec (1698 seconds)

Published: Thu Feb 29 2024