Name: Why ARM CEO Rene Haas thinks his company will play a pivotal role in realizing artificial general intelligence
Uploaded: 2024-01-19
Duration: 15 min 6 s
Description: ARM specializes in designing and licensing microprocesser IP, which plays a key part in various devices such as smartphones, smartwatches, and data centers.

ARM specializes in designing and licensing microprocesser IP, which plays a key part in various devices such as smartphones, smartwatches, and data centers.

Why ARM CEO Rene Haas thinks his company will play a pivotal role in realizing artificial general intelligence

Conferences

Why ARM CEO Rene Haas thinks ARM will play a pivotal role in realizing artificial intelligence

I'm gonna start with a really simple question and that's just, are you an A I company now? Have you jumped on the bandwagon? Yeah, I, I don't know that any CEO can be, uh, a tech CEO these days without saying A I, uh, somewhere in their, uh, in their opening remarks. But for those of you who don't know, uh, who arm is or what we do, uh, I will assure you that probably every single person in this room has arm technology somewhere, uh, on your body or pocket now or your pocketbook. Uh, we are the world's leading provider of microprocessor IP. And what that basically means is we're in your iphone, we're in your galaxy phone. We're likely in your car. We're in your set top box at home. Uh, 70% of the world's population, uh, uses arm. So I think by definition, uh, you can't really run A I with, without arms. So, yes, we're, you're an A I company. Ok. So, um, a lot of your very important comp customers are also now A I companies, um, and video being one of them, which we'll, we'll get to later. Um, but do you, did you foresee this crazy boom happening? Were you surprised by it? Uh I don't think we saw it happening at the pace that, that it's happened um on one level. Uh A I is not a new thing. Uh When you think about Google assistant or the Amazon Alexa doing voice recognition, you know that that's, that's a I uh Google translate. That's A I, but I think when, when we have the chat GP T moment and just the capability of these large language models that could very quickly take something that sits on the internet and produce incredibly good language and text from it. That was uh a bit of a big bang moment. And I think now what it's initiated is a, a huge rush to where can we take the technology? And uh I do believe we are in incredibly early innings of this uh the the capabilities of chat GP T or Gemini, some of the, the models that are out there are, are truly amazing, but they're just scratching the surface in terms of what can really be done. So explain to us the, the sort of the, the pitch here when it comes to A I data centers, the compute power that's, that's needed to, to power all of this, even though we're just scratching the surface, um you've certainly dominated in the smartphone uh chip architecture market. Um Why do you think you can dominate in A I? Well, one of the things that makes the A I uh problem uh unique is simply the raw amount of computer that you need to, to run these models and algorithms. And, and today, what you're seeing with these um large language models is that the data centers required to, to power them are in the hundreds of megawatts. Uh And just to give you a level of context, the US itself is about five gigawatts the entire country in terms of what, what we use to get to A G I or artificial general intelligence. People think it's orders of magnitude in terms of the compute capacity required. And then you start to do the math and says, well, wait a minute, that's, that's gigawatts of, of power. How do we generate all that power? So one of the things that, and it's again, it's gonna take a lot of uh contributions to solve this, but one of the things that arms are really good at is power efficiency. Uh that is the genesis of the company. Uh It's why we've been so ubiquitous in smartphones and why we're growing into evs in the data center. So we think we play a central role in these next generation data centers that need to solve the A G I problem. And you're already seeing this reflected in your numbers, right? I mean growth last quarter was spurred on by some of this. We are definitely seeing it. Uh We are seeing uh a lot of companies looking to invest more uh in terms of A I capability. What does that mean? Devices that may have not had uh the requirement for as much computer need more uh areas that didn't need a computer may need one. things like security cameras or sensors or certainly uh autonomous vehicles, robotics, they all need far more compute capability. So what we've seen for, for arm has just been an increased demand for our technology. And I think that to your point, the numbers are reflected that and, and speaking of energy consumption, um we're seeing this uh trend starting to happen now. Uh spurred on partly by the Chips Act, of course, with more manufacturing, more investment in manufacturing from some of your biggest clients, some of your biggest customers. Um I know you don't manufacture, you license your technology. But um how do you think about some of the environmental implications? I know there's great opportunity for jobs and the economy and national security? Of course. Um Do you think about the environmental impact uh at your level on the chip design level and in other ways as well? Oh, we have to, you know, go back, go back to this uh five gigawatt problem uh in terms of sustainability. Uh We have to think of ways of being incredibly power efficient to make sure that we're not taking away fundamental resources from the planet. Uh We're not involved in areas uh like renewables and, and, and solar and things of that nature. But we work fairly closely with companies in that space relative to other things that we can do in terms of how to make our products as, as power efficient as possible. And that means not only being very efficient when they're running, but when they're not running, literally drawing no power, we, we have some products that go into uh devices uh for example, that batteries will last for years uh running on unarmed technology. And it's something that's very, very central to what we do. My, my watch, for example, I don't know if it's years on that watch or maybe your next watch it, right for a few years. Um So I, I know you were uh back to A I of course, um I know you were uh part of this A I uh safety summit just recently um in November, I think. And there were a lot of uh other notable names there including Elon Musk. Um Take us a little bit behind the scenes. What were the conversations like there? That was a fascinating event. Uh The UK did a great job. Uh It was at uh Bletchley Park uh for starters. So Bletchley Park was where uh the computer that basically cracked the enigma code uh during World War Two with, with uh with uh with Alan Turing and that's where they, where they held it. So that was cool in of itself. It was quite a mix of people. Uh There were lots of government folks. Uh In fact, one thing that was interesting was you had government officials from the US, uh China uh Africa, European continent. So the demographics were fantastic. A lot of researchers were there uh As you mentioned, uh Elon was there, Sam Altman was there. Uh Dennis Hes from uh from Google Deep Mine, Mustafa Silliman from inflection. A I quite a quite a mix of people. I felt very fortunate to be there as a chip guy. Uh There weren't any, an A I guy, an A I chip guy, there weren't many A I chip people there. So anything that kind of came out of my mouth was like gospel even though I, I don't think I was saying anything too uh earth shattering. Uh But it was what was interesting about that event was, I, I almost drew a parallel between maybe what the first climate change conferences looked like because the groups were trying to get together to figure out about A I ethics ta testing safety, how to get people's arms around this very, very complex problem relative to the hardware, the software, the regulations. And obviously we didn't solve anything in, in those, those sets of meetings. But there is another uh set of meetings coming up, I think six months plus November in Korea. There's another +16 months plus in Paris. And it is a great start to uh I think what is necessary because this is a, a really complex problem. We, we, on, on the heels of the open A I saga, we've seen these kind of two camps emerge and it was already starting to happen before. I think this just kind of highlighted it and it seems like uh like a lot of things. It's polarized people, right? You're either kind of for uh uh advancement of A I or, or you're against it because you're scared by some of the negative by product or what could happen. Um Do you find that there are nuanced voices when it comes to tech leadership? I mean, do, are there people who are more in the middle and are sort of trying to make that message of that? But uh the way I tend to think about it, Lara, you know, brought up very well in, in her opening remarks, this is coming and it's moving frankly at a pace faster than I think we had all realized. So there's a bit of get, get on board and start to figure out exactly how you're going to address some of these problems and issues because it's, it's happening at a pace much, much faster than I think any of us had realized. Now, I am a, I'm more of a for good person than for bad. I, I think there's phenomenal amount of good that can come out of the technology. Uh I think in terms of health care uh in terms of uh cancer research, just you know, one of the examples that has been given about researching all the technical papers that exist around a certain disease or a, a certain type of medical issue. Today's doctors have no capacity or ability to, to do that kind of research. A I can be a phenomenal tool to help in that area. So I, I believe there's a lot of good that can come from it. Now with any new powerful technology, there's gonna be bad actors, there's gonna be potential for uh for misuse. And that's what I think the industry needs to get, uh get together on making sure we understand how to address all that. Um I'm gonna come to you guys for, for questions in just a minute. Um But uh I wanna pivot a little bit to talking about Softbank and NVIDIA, my two favorite subjects. So, um you know, I mentioned the, the IP O of course. Um But, but what preceded that was uh what could have been a sale to NVIDIA where used to work? Um Can you talk a little bit about, you know, coming in as a CEO at a very pivotal time leading this company through a shift? That wasn't exactly the, the, the intended shift. Um And, and just what you've learned along the way as a leader. Uh That's a, that's a great question. Um Yeah, so I worked for NVIDIA for a for a lot of years and I've been with arm since, since 2013. And when uh when NVIDIA came up to uh to buy us in 2020 the idea of going back and working together with NVIDIA and Jensen again was, was exciting. Uh But then we ran into a, a tsunami of uh of regulatory issues around that and it became kind of obvious about a year and a year and a half into it that this was probably not going to come through and to start started thinking about kind of what the the future would lie. Uh The good news was a lot of the arm that is being successful today was really being put in place a number of years earlier when I was running the core business, uh what we called at that time that the IP products group. So I think when we started to put that strategy in place relative to diversification, the data center autonomous vehicles, uh IOT devices, that strategy is held together going forward. Uh One of the, the things that we're lucky about in the arm business is that uh our, we, we look like 5 to 10 years ahead in terms of where the, where the puck is going. So a lot of the investments that we needed to make were already in place from a strategy standpoint. So when the deal collapsed and I became CEO uh it was a bit of just continuing that, that level of, of, of work and effort. Now the learning piece was I realized that when you're the CEO uh people really listen to what you say uh for good or for bad. Uh And people will even look at your face on a zoom call. And you may have had a problem with your dry cleaning and people may think, oh my God, his face is frowning. He may be planning a layoff at some point in time. So I, you know, I started to learn that my, my behaviors and actions can, can really impact uh what people do. But I've been very lucky. We have a great business. We have a great, great team. Uh We were very fortunate to go through an IP O last year, as you mentioned, which could not be happier with how that all has turned out. So, yeah, and any learnings was, I think it's sticking to what your, your core beliefs are, uh has really worked out well and you're still friends with Jensen and I'm still friends with Jensen. Um Any questions, questions for Renee, please raise your hand. We'll bring a microphone over. B All right, if you do, please raise your hand later. Um So, uh I wanna ask you about, you know, we're here on the eve of ce s, right? There's gonna be all sorts of cool and crazy things on the convention floor. I'm sure for those of you who are are braving the crowds. Um, but, but what are you seeing, you know, you said so much of your business depends on your ability to, to look down the road. Um, what's, I think we talked about a, I quite a bit but, but more specifically what's really, what's the next smartphone? Yeah, you know, going back to so on the smartphone side, I think you're gonna see much more integration of A I capabilities. Uh what that tends to look like in terms of the application space and how it helps you, I think still needs to be figured out a little bit. Um But one of the things I actually, I am excited about the show floor is ce S has really evolved uh over the number of years now. It's a, a big auto show and there's a lot of exciting stuff going on with, with automobiles. Uh and, and I'm very curious to see where things are going from an au autonomous driving standpoint, but just going back to an example of, you know, A I and where A I can can help. Um I took a drive in a car. We were chatting earlier about this uh before the holidays in central London with a company that was working on autonomous driving. Now, what they're doing is they use these large language models, generative A I to create simulated video of traffic patterns. So for those of you who sees the, see these autonomous cars today, these Waymo or cruise vehicles that have tons of cameras and liars on top. And they're cap capturing images the way that works is they, they do these video rides, they capture the images and then they upload them to the computer to try to learn what's going on. This company does this all through generative A I, meaning that they will create a simulation of what Las Vegas Boulevard will look like at 725 on a Monday during ce S week and use that to teach the car how to drive. So what they do is they cut down all this time relative to capturing all this video. And if you think about the more simulations that you can create the better the car gets. And we took this drive through Central London, which is for those of you who have been there. It is not an easy place to drive and it did an amazing job and they only have 12 vehicles on the road and, and they've only been at this for 23 years and this is just AAA small example of where uh I think Gene of A I as a, as a vehicle. No pun intended for driving new innovation. It's gonna be amazing, amazing. It's actually kind of mind blowing when you really uh uh think about it. Um Renee, thank you so much for joining us tonight. Appreciate it. Thank you so much.