On this episode of The Six Five – On The Road, sponsored by Intel, hosts Daniel Newman and Patrick Moorhead welcome Intel’s CEO, Pat Gelsinger for a conversation on the “Siliconomy” and how Intel is “Getting the Geek Back” by reinforcing their belief that InnovatiON is for developers, as well as the opportunities created by the generational shift AI.
Their discussion covers:
- What priorities CEO, Pat Gelsinger has been razor-focused on during his three-year tenure at Intel
- How Intel’s AI approach goes where others aren’t yet–AI on the PC, at the Edge, in the Enterprise, as well as the cloud and data center with a broader range of hardware and software
- An expanded look at Pat’s keynote highlights, including the new coined phrase “Siliconomy”
- The main takeaways Intel hopes to deliver to developers over the course of the event
Be sure to subscribe to The Six Five Webcast, so you never miss an episode.
Watch the video here:
Or Listen to the full audio here:
Disclaimer: The Six Five webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Patrick Moorhead: The Six Five is live here in San Jose at Intel 2023 InnovatiON. It is great to be here. We are on the show floor, having fun, having some of the greatest discussions. We’re talking AI, we’re talking data, we’re talking security, we’re talking edge to cloud and everything in between. My co-host Daniel Newman. Dan, how much fun is this?
Daniel Newman: It’s great to be here on the floor. We are in the middle of the explosion of the sileconomy. And I won’t tease-
Patrick Moorhead: I heard that once.
Daniel Newman: Who may have said something about that, but I do think it’s been a really exciting week because you and I always talk about silicon will rule the world.
Patrick Moorhead: World. Exactly. I mean, hey, software’s important and we saw the software and it rules the world too, but it’s hard to rule the world with software without some amazing silicon on top of that.
Daniel Newman: Try to run it on air.
Patrick Moorhead: Exactly. It just doesn’t work.
Daniel Newman: It doesn’t work.
Patrick Moorhead: Without further ado, let’s introduce somebody that doesn’t actually need an introduction, Intel CEO, Pat Gelsinger. Pat, how are you?
Pat Gelsinger: Hey, always a pleasure to be on Six Five. Great to see you and the opportunity to have you join here at InnovatiON. And you guys always do a great job, just come in and bringing out the story associated with the tech. Thank you so much and great pleasure to join you.
Patrick Moorhead: Yeah, I appreciate that, Pat. And we do like to think that accuracy and facts matter. And I just want to say, I just want to appreciate and thank you for making us look so smart in terms of our prognostications over the last two to three years. Thank you.
Pat Gelsinger: Well, hey, you guys were advocating that hey, this turnaround was going to materialize. It’s really important and a lot of skeptics associated with it. And now as the green shoots are turning into trees and plants, like, wow, you guys were really smart.
Patrick Moorhead: Thank you for that.
Daniel Newman: There are those that root for the underdog. I come from Chicago, we’ve got Cubs.
Patrick Moorhead: I come from the Midwest. I’m a Midwestern.
Daniel Newman: You’re a Cleveland guy. Just our athletic backgrounds and the sports and our teams gave us a reason to root for the… But it’s funny because Intel’s really never been an underdog. It’s been a company that’s had some changes and it’s seen new competition and entrants.
And it really opens up a good conversation right here, Pat, is three years ago you came in the door. You came in ambitious, you were on tour, you were telling the story, the turnaround story. But now the time has flown by and you’ve been five in four. I mean, that’s been one of your big talking points. It’s the say, do, that stuck with me from day one, Pat. But talk a little bit about this past three years, those things you were really zeroing in on and how that’s going up to now.
Pat Gelsinger: Yeah. Yeah, and if you pick just a couple. Obviously you mentioned five nodes in four years. Generally a node took two years, so five. Okay, that’s eight or 10 years of work. Wow. And crunching it into four years. And nominally we’re on track in. Intel 7, done. Intel 4 with Meteor Lake or now the Core Ultra, done.
This is manufacturing ramp. We’re now ready to move it into our second factory out of the TD factory into Ireland. That’s done. The Intel 3. Hey, the Granite Rapids and Sierra Forest. And speaking of Sierra Forest, I can never show up without a prop. And what we showed today on stage was that this was 144 cores, what we showed before. What we didn’t tell you was we had two of them.
Patrick Moorhead: Surprise.
Pat Gelsinger: It’s 288 cores. And as I like to say, when I was working on the Nehalem, the four core part, it was like, wow, four cores on one die. 288 cores on one package, just mind-blowing. And Granite and Sierra are on Intel 3. That’s coming along. And we’ll be sending the production parts to the factory shortly.
20A, we showed the first Arrow Lake wafers, the first product on Intel 20A, which is the new transistor, the new backside power. That’s happening. And the test chips on 18A and we’ll soon be sending the first products onto 18A. Five nodes, four years, incredible and it’s coming together, so quite exciting.
Patrick Moorhead: Yeah, you’re surrounded by those wafers up there. It’s like a smorgasbord and it’s not even Thanksgiving yet. It was awesome. I could barely keep pace with boom, boom, boom, boom. I love that. 2023 actually, AI has been around a long time. Got really hot over the last five years, but starting last year and around November, it really heated up.
And you have a real diverse play of AI. All the way from CPU, GPU, NPU to ASICs, to FPGA, don’t want to forget that, and everything in between. And you scale from the client computer to the hyperscaler data center and pretty much everything in between. Yet you don’t get a lot of credit for being a part of this new and latest AI, let’s call it the generative AI game. What’s your response to that? I mean, you know what we’ve written in Dan’s, but I want to get your response to that. Why do you think that’s happening?
Pat Gelsinger: Well, first let’s characterize the market a little bit, and I call it the AI continuum. And the energy, the action has been on these very big systems that are creating these training environments. And why don’t we get credit? Well, because we’re not particularly represented there, and that’s where the hottest of the hot has been.
And then you sort of say, “Well, what do I do with those models?” Well, hey, I might put them into inferencing farms. When I start going up to my Bings or my Googles, I’m now getting a generative AI experience, so much more using those models, but then those models start to become more useful. People want to bring them on premise. Now we’re talking for more of our differentiation because now it needs to be my data with my use case.
Patrick Moorhead: Private AI.
Pat Gelsinger: On my premises. Or hey, maybe it’s hybrid AI. I have a trained model, I’m getting it updated, but that’s the foundation and I’m refining into my local data. Now we’re getting more into our sweet spot. But the thing that really excited me about InnovatiON this week was the AI PC because hey, if I have to go to the cloud for my AI, it’s expensive, the latency is high as I’m going up there and I break the sovereignty of my data to go do that.
This idea of pushing it downward and I call it the three laws of edge and AI. The laws of physics, wow, round tripping. Laws of economics, expensive resource in the cloud. And finally the laws of the land. And that’s going to push more and more of AI to the edge and to the client. And we showed off some pretty cool demonstrations today as Meteor Lake, our Core Ultra comes to market. And I think as that AI continuum takes shape, we’re going to get a lot more credit and we’re going to compete at the high end as well.
Patrick Moorhead: I love that.
Daniel Newman: It feels like the AI world wants to say AI’s been determined, like it’s done. And I think this is one of the things I took away from listening to you and just from our assessment of the overall industry, Pat, is that this is like, we’re still in warmups. The fans aren’t even in the stands. We’ll keep with these sports analogies a little bit longer.
The concessions are just heating up the dogs. And what I mean is we really are very early. And for instance, training is all the rage right now. Of course there’s some companies that are selling lots of GPUs for training and everyone’s like, that’s it, that’s the world. But when you talk about what you just talked about, Pat, so much of the AI is going to be inferencing. It’s going to be either done on the device, at the edge in an application. Most of that could be done on a CPU.
And by the way, I remember Intel telling this story five and six years ago talking about DL Boost and SAP workloads and all the things. It’s almost ironic that we’ve gone full circle, but people are still trying to make it such a one horse race.
Pat Gelsinger: And with that, I mean there’s a lot of energy there, a lot of excitement there. I understand that as well. But hey, you and I, we’ve been through many of these hype cycles in the industry.
Patrick Moorhead: Sure.
Pat Gelsinger: But I really believe that this idea of AI everywhere is so powerful where you truly start to see these use cases much more inference, retraining of models on local data. Also what we call nimble models. Hey, what do I do with a trillion parameter model running on the biggest supercomputer that’s ever been built?
I can’t do a lot with it, right? Hey, but if I can make it a billion parameters and put it on a couple hundred million PCs, now I can do a lot with it. And that idea of big models getting compressed and pushed out to the edge. The analogy I use is, how many of you generate weather models? How many of you use weather models?
And the answer is, hey, if you generate these super big things. But then it’s all a question of how do we use it? How do we bring it to every TV station, to every supply chain, to every delivery person, to every Uber driver? Okay, that’s what we’re talking about. That’s AI PC and that’s AI everywhere. And I think Intel will have a great opportunity here as we compete at the high end, but deliver it in volume, AI everywhere for every application on every device for everyone.
Patrick Moorhead: Pat, you coined a really catchy phrase or a catchy term up on stage, sileconomy. And in fact, not only did you attend and talk about sileconomy, you have the t-shirt as well.
Daniel Newman: When do we get those by the way?
Patrick Moorhead: No, exactly.
Pat Gelsinger: We’ll work on that.
Patrick Moorhead: And you talked about, hey, the introduction and the impact of generative AI on this next generation. And as Dan and I has always said, we always thought that silicon was eating the world. But what do you mean about sileconomy? Yeah. What are you trying to talk about here?
Pat Gelsinger: And I go back to the five superpowers is that everything will compute, every one and everything will be connected. We will have infrastructure from the cloud to the edge. We will have AI making sense of it, and we’ll increasingly enhance our senses as well in my AI enhanced hearing aids.
But if you think about those, literally the digital platforms are permeating every aspect of your existence, your financial, your social experience, your healthcare. We had some cool health demos as well. Everything’s becoming more digital. Literally your life is becoming a digital experience and everything digital runs on silicon.
Patrick Moorhead: Silicon.
Pat Gelsinger: And so that’s the silicon economy because everything… And as I say for the last five decades where oil reserves have been, have defined geopolitics for the world. Where technology supply chains and where silicon is will define geopolitics for the next five decades, it’s that important. It truly is a silicon-based economy.
And then the last piece of that is who rules? Is it political leaders? Is it business leaders? And I say none of the above. It is developers because they’re the ones that are unleashing that incredible power of Moore’s law and silicon for these next generation experiences. And AI has just been like a turbo-boost on that innovative cycle.
Daniel Newman: It’s interesting because I said rule the world, eat the world. One thing we can all agree upon though is that this layer of silicon is what’s changing. And everything that’s being built on top of it starts here. And it starts with companies like Intel that are really thinking about it from hardware to software.
And I think it’s often the company gets all the credit for hardware and none of the credit for software. But you’ve been doing software for a long time and you have a very loyal… I mean we sat and watched the keynote, a packed room full of developers. And you’ve made that comment, Pat, you made the comment and you said, “Developers will be the winners in this thing. They have the biggest opportunity to really drive this future.”
Talk a little bit about how you’ve come to that. Some people are saying AI will relieve programmers from needing to exist anymore. You obviously… I’m not saying we agree, I’m just saying we hear it. You clearly believe in this group, in this audience. And you stood up on the stage and you really said, “This is your time.”
Pat Gelsinger: Yeah. And with that, it really is the hardware now becomes this enabler. And I always say software defined, silicon enhanced because it is the software layer that is defining the user experience, the algorithms, et cetera. And the hardware guys are trying to catch up and how do we make it possible on that scale?
And I in my own personal way, I say, “Why did God have me run a software company for eight years before coming back to lead the silicon company, the most important company in technology?” Well, I think it’s for this understanding of how important that developer is to unleash those capabilities and not being chipheads, but software-defined, silicon enhanced.
And in this period of time, and again, I mean things like copilots and so on are just unleashing extraordinary productivity gains for developers and more natural language where essentially my low-code environment can be English going forward in these environments. But at the end of the day, the innovative cycle is driven by developers. They are doing the things that are not yet possible. And we showed some cool examples of it this week.
Patrick Moorhead: Yeah, you really did. And Pat, you brought out a lot of messages. I mean, you went GA on the Intel developer cloud, which just seemed like a very robust way, regardless of what type of compute you use to not only test out your software but also find the right accelerator. You’re not locked into whether it’s a cloud GPU, an FPGA, a CPU or an NPU.
It’s Hey, try all these out. And oh, by the way, as we know, these workloads scale from right to left on sophistication and ultimately, they wind up in a CPU. That’s happened for the last wheel of 30 years all the way back to the DX co-processor and accelerating instructions there. You did send out a ton of messages to a lot of developers. What are some of the key points if you have to solidify the messages over the last 24 hours that you want to send to developers? Because you really are bidding for their hearts and their minds and their time and their dollar investment.
Pat Gelsinger: And what we said is open, we’re going to be this open platform. Committed to open standards that they’re going to be able to build upon. Not closed, not proprietary, committed to open, that we’re going to give choice. We’re going to give choice of CPUs, GPUs, accelerators on the edge and so on for it. And we’re going to build trust.
And trust gets to be super important here. If I spent $20 million building my AI model, do I want it to get ripped off? Do I want it to be exfiltrated because it’s been trained with my proprietary data associated with that? I want to trust it.
And we’re making these commitments to our customers so that they can go develop on dev cloud. And we’re going to have CPUs, GPUs, accelerators, et cetera. We also showed and we announced stuff. For instance, OpenVINO is now being supported with Arm. And another people-
Patrick Moorhead: I heard Cheers by the way.
Pat Gelsinger: And everybody was like, “Huh, they’re working with Arm? Don’t you guys compete?” And it’s like, ah, in some domains we do, but in other domains this partnership is very important for that. And as we’re working to make our one API and enable Python programming at scale, how do we get high performance implementations? Because ultimately most of these app developers, hey, they’re saying, “Hey, I need more compute in the store. I never want to touch my app.”
And OpenVINO gets to say, “Got it.” We’re just taking advantage of the next generation accelerators. We’re downloading the updated models. And how do you make it that frictionless way that they can take advantage? And those are all the things that we rolled out this week, but fundamentally, open choice and trust, that’s our message.
Patrick Moorhead: Open choice and trust. Powerful.
Daniel Newman: All right, so bonus question. I think we went quick enough that we can get him for one more. I want to talk a little bit about the geopolitic environment, macro of all the things you’re doing from a foundry, from a developing and expanding capacity. We’ve had a crazy couple of years, Pat. And you highlighted it a little bit here, but we had the CHIPS act. You’ve made huge commitments to build out more capacity in, what, Oregon, New Mexico, Arizona, Ohio, and of course overseas too. How is that developing? It felt like it was this big seminal moment. Everyone focused it. We saw all the photo ops with President Biden. And then all of a sudden it’s like where’d this go.
Pat Gelsinger: It was Pat and Pat and Dan or Pat and Biden.
Daniel Newman: Yeah, I mean, and sorry, but yes. I mean Joe, Mr. President Biden was busy. But where’s that evolve to? Because it feels like it got quiet.
Pat Gelsinger: And first, the first thing I would say is if we were sitting here two and a half years ago as we began this journey and I says, “You know what we’re going to get done? We’re going to get the most seminal industrial policy legislation since World War II completed in the next two years. And we’re going to do it in US and in Europe.” You would have looked at me and says, “Are you an idiot?”
Patrick Moorhead: IDM 2.0.
Pat Gelsinger: And we did, for just the significance of that. Even as things have been a little bit more quiet. It was just a couple of weeks ago that the EU Commission approved the EU CHIPS Act, and it was a nailbiter vote, 587 to 10. Just the political processes, the approval processes take a while.
In the US We’ve now submitted our four applications to the CHIPS program office. They did the rule making and now we can apply against it. Those just are in their office right now. And I’ve said to them very clearly, “I have funds from this in my business plan for this year.” We’ve been working very closely. And as those come forward, that’ll be a very visible moment as well. We’ve launched all of the programs that we set our four sites in the US, Germany, Poland, and the completion of the Ireland site, Malaysia site. This is aggressive, crazy building cycle. And if you show up at those sites. And just if you think about one of these manufacturing locations, you have the largest construction project occurring in the world today at these sites.
The biggest cranes, the biggest trusses, more concrete, six football fields big, building the smallest things that have ever been built. The biggest construction doing the smallest things on earth. I mean, it really is spectacular. And that’s the journey we’re on.
And I’m thrilled with the progress you’re making, that the teams are making across the sites and the projects. And we’re building supply chains in the Americas, supply chains in Europe. It’s a compliment what we’re doing in Asia. This is profoundly important for the world and we are getting it done.
Patrick Moorhead: Yeah, I just couldn’t believe the IDM 2.0 announcement in what you were saying you were going to do. And it’s funny in retrospect what a good decision that was for so many reasons. And what some people can’t fully appreciate is how many years in advance that these decisions have to be made and how early the investment has to be made to have goodness down the line.
And it’s important for the country, it’s important for the world to have a protected supply, a competitive supply of microprocessors everywhere. And I think we all learned that over the past two or three years.
Pat, I want to thank you so much for the time. It’s always a treat when we can get you on The Six Five. Appreciate all the time you spend with us, not only in our capacity at The Six Five, but also as industry analysts. We sincerely appreciate that.
Pat Gelsinger: Hey, thank you both. Always a pleasure to be with you guys.
Patrick Moorhead: Thank you. So… this is Pat and Dan, little Pat, big Pat, Daniel Newman here for The Six Five at Intel InnovatiON 2023 live in San Jose. We want to invite you to tune into all of the videos that we shot here, all the amazing conversations and the insights. Hit that subscribe button and take care.