WEBVTT - AMD's Chip Deal With OpenAI Triggers Explosive Rally 0:00:00.040 --> 0:00:13.480 Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is alive 0:00:13.520 --> 0:00:17.319 from coast to coast with Caroline Hide in New York 0:00:17.600 --> 0:00:20.080 and Eva Low in sentences. 0:00:19.520 --> 0:00:25.720 Go this is Bloomberg Tech coming up. AMD signs a 0:00:25.840 --> 0:00:28.720 historic deal with open Ai to roll out AI infrastructure, 0:00:28.760 --> 0:00:31.880 a packed the chipmakers said could generate tens of billions 0:00:31.880 --> 0:00:33.320 in dollars of new revenue. 0:00:33.440 --> 0:00:33.720 Plus. 0:00:33.760 --> 0:00:35.680 We speak with the White House AI and Crypto Zar 0:00:35.800 --> 0:00:38.280 David Sachs about relations between the US and China in 0:00:38.320 --> 0:00:40.000 the context of the AI race. 0:00:40.680 --> 0:00:43.760 And metas cmo joins us to discuss how AI is 0:00:43.840 --> 0:00:48.000 changing the advertising landscape what that means for small businesses. 0:00:48.080 --> 0:00:50.960 Here and now we turn our attention to the broader markets. 0:00:50.960 --> 0:00:52.760 Bitcoin at a record high. Then, now's that one. 0:00:52.720 --> 0:00:54.760 Hundred at a record high. We're up six tenths of 0:00:54.800 --> 0:00:58.160 a percent, the moon music around AIAI infrastructure, and the 0:00:58.200 --> 0:01:01.920 frenzy therein ed it continues. I'm looking at the socks 0:01:02.040 --> 0:01:05.000 up three point five percent, a new record high for 0:01:05.080 --> 0:01:07.640 the index that tracks the semiconductor industry. 0:01:07.920 --> 0:01:10.240 But what is leading a game? Of course, you know 0:01:10.280 --> 0:01:11.080 it's one key name. 0:01:11.240 --> 0:01:13.240 We look at AMD at one point hitting a record 0:01:13.280 --> 0:01:15.520 high We are up the most on this stock in 0:01:15.640 --> 0:01:18.319 nine years, twenty eight percent surge. At one point it 0:01:18.360 --> 0:01:21.000 was thirty seven percent higher. All as we hear that 0:01:21.120 --> 0:01:22.920 tens of billions are going to be added in terms 0:01:22.959 --> 0:01:27.080 of revenue because the open ai infrastructure pat a sign 0:01:27.520 --> 0:01:31.480 of real confidence in Lisa SU's AI accelerator and build 0:01:31.480 --> 0:01:34.440 out ed. We've got such an interesting conversation coming out 0:01:34.640 --> 0:01:35.840 about this new deal. 0:01:39.040 --> 0:01:42.360 Welcome to Bloomberg radio and TV audiences around the world. 0:01:42.400 --> 0:01:45.520 AMD has signed a definitive agreement with open Ai to 0:01:45.560 --> 0:01:49.760 deploy six gigawatts of amdgpus. AMD says it will equate 0:01:49.800 --> 0:01:52.720 to tens of billions of dollars in revenue. Open Ai 0:01:52.960 --> 0:01:55.200 will get up to one hundred and sixty million AMD 0:01:55.320 --> 0:02:00.000 shares in tranches and set against both operational and financial milestone. 0:02:00.320 --> 0:02:03.520 The focus is inference. Let's bring in a m D 0:02:03.640 --> 0:02:06.840 CEO Lisa Too and open Ai president Greg Brockman, both 0:02:06.840 --> 0:02:09.000 of whom join us on set here at Bloomberg Tech 0:02:09.120 --> 0:02:09.880 in San Francisco. 0:02:09.960 --> 0:02:12.680 Good morning, Good morning, It's great to see you here. 0:02:12.800 --> 0:02:15.760 Let's frame the opportunity. Lisa, you know that the market 0:02:15.800 --> 0:02:18.680 reaction is very clear, but for a m D and 0:02:18.760 --> 0:02:21.639 the AI industry at large. What do you think this represents. 0:02:22.080 --> 0:02:24.679 Well, look, this is a huge milestone for a m D. 0:02:24.919 --> 0:02:27.200 You know, we are so thrilled with the partnership with 0:02:27.520 --> 0:02:29.560 the open Ai team, and it's also you know a 0:02:29.639 --> 0:02:32.440 huge moment for the AI industry because you know, when 0:02:32.480 --> 0:02:34.400 you get to the right down to it, you need 0:02:34.440 --> 0:02:36.799 more AI compute. I mean, that's where we are today. 0:02:36.800 --> 0:02:39.240 Compute is a foundation for all of the intelligence we 0:02:39.280 --> 0:02:41.800 can get from AI. And you know, we are a 0:02:41.880 --> 0:02:45.079 compute provider. We have spent years on our roadmap, We've 0:02:45.120 --> 0:02:48.680 spent years working with open Ai and the team, and 0:02:48.840 --> 0:02:51.359 you know, together now we're embarking on you know, a 0:02:51.440 --> 0:02:55.320 massive build out of six gigawatts of AI compute, and 0:02:55.680 --> 0:02:58.320 it's it's a big deal for us, for our shareholders, 0:02:58.400 --> 0:03:01.239 for our teams, and for you know, the partnership and 0:03:01.560 --> 0:03:03.000 the overall AI ecosystem. 0:03:03.240 --> 0:03:05.720 Greg I say that the top the focus is inference. 0:03:05.840 --> 0:03:09.200 I think that's really important to be specific about what 0:03:09.240 --> 0:03:11.840 you will do with this capacity. So so literally explain 0:03:11.919 --> 0:03:14.000 that part. And I'm conscious that you know, in the 0:03:14.000 --> 0:03:17.080 first instance, the first target is one gigawatt and then 0:03:17.120 --> 0:03:19.880 eventually six gigawatts, but what will you use it for? 0:03:20.600 --> 0:03:23.960 Well, I think that the world continues to underestimate the 0:03:24.000 --> 0:03:27.880 amount of demand for AI compute. Right that we've seen 0:03:27.960 --> 0:03:30.920 this explosion of demand with things like chat GBT. You know, 0:03:30.919 --> 0:03:33.080 we're at eight hundred million weekly active users. Now this 0:03:33.080 --> 0:03:35.680 product didn't even exist three years ago, and we're in 0:03:35.720 --> 0:03:38.840 a position where we cannot launch futures, we cannot launch 0:03:38.920 --> 0:03:42.120 new products simply because of lack of computational power. And 0:03:42.200 --> 0:03:45.480 we see these models continuing to get exponentially better, and 0:03:45.560 --> 0:03:46.960 I think we're just heading to a world where so 0:03:47.040 --> 0:03:48.680 much of the economy is going to be lifted up 0:03:48.680 --> 0:03:51.040 and driven by progress and AI. And so we're very 0:03:51.120 --> 0:03:52.800 much heading to a world by default that I think 0:03:52.840 --> 0:03:54.960 looks like a compute desert, right that there's just not 0:03:55.080 --> 0:03:56.920 enough compute to go around, and so we're trying to 0:03:56.920 --> 0:03:59.440 build as much as possible, as quickly as possible. So 0:03:59.480 --> 0:04:01.600 we're starting with one gigawatt simply because you got to 0:04:01.600 --> 0:04:04.520 start somewhere. But honestly, we're building as fast as we 0:04:04.560 --> 0:04:07.440 possibly can and trying to bring as much computational power 0:04:07.880 --> 0:04:09.720 to bear for the economy and for the world. 0:04:10.480 --> 0:04:14.040 Lisa, this is such a big commitment to Instinct in 0:04:14.080 --> 0:04:17.240 particular as a customer. Does it make open ai the 0:04:17.320 --> 0:04:19.320 largest for that particular product. 0:04:20.160 --> 0:04:23.120 Well, this is certainly the largest deployment that we have 0:04:23.160 --> 0:04:26.400 announced by far. I mean, you know, six gigawatts of compute. 0:04:26.800 --> 0:04:28.400 As Greg said, we're going to start with the first 0:04:28.400 --> 0:04:31.080 gigawatt in the second half of twenty twenty six on 0:04:31.160 --> 0:04:34.920 our new next generation m I four to fifty chip. 0:04:35.400 --> 0:04:37.400 I think the thing to understand is, you know, these 0:04:37.440 --> 0:04:40.920 types of partnerships actually take you know, years to really 0:04:40.960 --> 0:04:43.359 get comfortable with the idea that we're going to you know, 0:04:43.400 --> 0:04:46.360 go all in together. And this isn't all in partnership 0:04:46.839 --> 0:04:50.000 in terms of building out you know, the AI compute 0:04:50.480 --> 0:04:53.520 that open ai needs for everything that they're offering to 0:04:53.520 --> 0:04:55.640 the world. So yes, it's a huge deal, and it 0:04:56.040 --> 0:04:58.640 also says a lot about you know, how much needs 0:04:58.640 --> 0:05:01.839 to come together for you know, this entire ecosystem to operate. 0:05:01.920 --> 0:05:03.800 So you know, we are setting up you know, certainly 0:05:03.839 --> 0:05:06.400 there's a lot of engineering work, but our teams are 0:05:06.400 --> 0:05:10.000 working together on hardware, software, We're ensuring the supply chain, 0:05:10.120 --> 0:05:12.800 all of those elements are set up and ready to 0:05:13.120 --> 0:05:15.480 deliver on this massive commitment. 0:05:16.040 --> 0:05:18.960 Greg, talk us through a little bit about the players 0:05:19.000 --> 0:05:20.600 that you need to also lean on. This has been 0:05:20.720 --> 0:05:22.880 years in the making, as you say, with AMD, but 0:05:23.320 --> 0:05:25.799 what other cloud providers were involved. How are you thinking 0:05:25.839 --> 0:05:28.800 about this working with an Oracle all others out there. 0:05:29.600 --> 0:05:32.200 Yeah, we really think of this as an industry wide effort, 0:05:32.480 --> 0:05:34.720 and in general, we think that compute is something that 0:05:35.080 --> 0:05:37.600 does require the entire supply chain to really wake up 0:05:37.600 --> 0:05:40.280 and to really to start building much more than people 0:05:40.279 --> 0:05:43.200 we're planning on. I think this starts from energy to 0:05:43.240 --> 0:05:46.200 try to get far more power to be built on. 0:05:46.400 --> 0:05:48.040 Things like nuclear I think are going to be very 0:05:48.040 --> 0:05:50.480 important to come online. The cloud providers are an important 0:05:50.480 --> 0:05:51.760 part of this as well. So we're going to be 0:05:51.800 --> 0:05:54.279 deploying AMD in our own data centers. We'll be deploying 0:05:54.320 --> 0:05:56.840 them together with cloud providers. You know, we have a 0:05:56.920 --> 0:05:59.960 deal with Oracle, lots of other cloud providers out there. 0:06:00.360 --> 0:06:02.359 You can really see that we're very much in the 0:06:02.680 --> 0:06:05.080 We just want compute as much compute as possible. We 0:06:05.080 --> 0:06:07.039 think this is important for the economy, We think this 0:06:07.080 --> 0:06:08.960 is important for the nation, we think this is important 0:06:09.000 --> 0:06:11.800 for humanity, and so really we're working with everyone in 0:06:11.839 --> 0:06:14.440 this whole industry in order to get as much compute 0:06:14.480 --> 0:06:16.160 power online as quickly as we can. 0:06:16.800 --> 0:06:20.960 Lisa, I'm sorry specifics where is this data center going 0:06:21.000 --> 0:06:25.440 to be? Is it one single site? Is it orical 0:06:26.320 --> 0:06:27.560 that we'll partner with you on this? 0:06:28.040 --> 0:06:31.200 Well, actually, what this really is is an announcement of 0:06:31.240 --> 0:06:32.720 what you know, A m D and open a are 0:06:32.720 --> 0:06:35.080 going to do together. You know, open ai has a 0:06:35.120 --> 0:06:37.800 lot of partners in terms of you know, where they deploy. 0:06:38.160 --> 0:06:40.719 I imagine a lot of it will be in cloud 0:06:40.720 --> 0:06:43.320 service providers. It's really up to you know open Ai 0:06:43.440 --> 0:06:46.640 and Greg and Sam and the team. But the way 0:06:46.680 --> 0:06:49.799 to think about it is, for this you know, amount 0:06:49.800 --> 0:06:51.359 of compute, it's going to have to be in a 0:06:51.360 --> 0:06:52.839 lot of different places. 0:06:52.880 --> 0:06:55.000 It's a massive amount, multiple locations. 0:06:54.640 --> 0:06:58.800 Multiple locations, I would imagine, you know, multiple providers to 0:06:58.880 --> 0:07:00.720 really get this online as fast as possible. 0:07:01.040 --> 0:07:05.200 Greg, there is a lot of focus on where open 0:07:05.240 --> 0:07:07.440 ai is going to get the money from to fund 0:07:07.440 --> 0:07:12.480 all of this. Sam Altman's big picture commitment is well documented, 0:07:12.560 --> 0:07:15.520 right and the numbers to his mind are in the trillions. 0:07:16.040 --> 0:07:19.520 But have you specifically thought about debt financing for this 0:07:19.600 --> 0:07:24.160 relationship with AMD? Have you thought about doing a specific 0:07:24.240 --> 0:07:28.160 equity raise. You are very committed across multiple projects. 0:07:28.400 --> 0:07:28.600 Yeah. 0:07:28.640 --> 0:07:30.040 Look, the way that I would the way that I 0:07:30.040 --> 0:07:33.160 would look at this is that AI revenue is growing 0:07:33.200 --> 0:07:36.880 faster than I think almost any product in history, and 0:07:37.120 --> 0:07:39.239 that ultimately, at the end of the day, the reason 0:07:39.240 --> 0:07:42.160 this compute power is so important, it's so worthwhile for 0:07:42.400 --> 0:07:45.880 everyone to build, is because the revenue ultimately will be there. 0:07:45.960 --> 0:07:46.200 Now. 0:07:46.480 --> 0:07:48.880 As a company that is trying to move as fast 0:07:48.880 --> 0:07:50.880 as we can, we look at everything, right, we look 0:07:50.880 --> 0:07:54.760 at equity debt, we look at trying to find creative 0:07:54.840 --> 0:07:58.000 ways of financing all of this. That's been actually a 0:07:58.040 --> 0:07:59.880 huge focus of us for the past couple of years, 0:08:00.000 --> 0:08:02.920 thinking about how can we possibly build the amount of 0:08:02.920 --> 0:08:05.680 compute that is required in order to really transform this 0:08:05.720 --> 0:08:08.640 whole economy into an aipowered economy. And so I think 0:08:08.680 --> 0:08:12.120 you'll see lots of creative ideas, but fundamentally, I think 0:08:12.120 --> 0:08:13.920 at the end of the day, it is because we believe. 0:08:14.320 --> 0:08:16.440 Sorry to jump in an interrupt and carriage, just forgive 0:08:16.480 --> 0:08:21.360 me on this one. The condition of AMD issuing the 0:08:21.440 --> 0:08:25.080 stock to open AI requires you to spend money basically 0:08:25.120 --> 0:08:29.720 because you have to deliver that gig awad of capacity. First, Lisa, 0:08:29.760 --> 0:08:32.560 I have to ask you if you have assurances that 0:08:32.640 --> 0:08:33.319 open ai. 0:08:33.240 --> 0:08:33.840 Is good for it. 0:08:34.160 --> 0:08:36.720 Well, let me be clear. I mean, this deal is 0:08:36.760 --> 0:08:39.680 a win for AMD, it's a win for open Ai, 0:08:39.880 --> 0:08:42.120 and it's a win for our shareholders. And that's kind 0:08:42.120 --> 0:08:44.680 of the way we put this together. I have full 0:08:44.720 --> 0:08:48.800 confidence in open Ai, Sam, Greg Sarah. I mean, this 0:08:48.880 --> 0:08:51.960 is a massive opportunity for us right now, right here. 0:08:52.120 --> 0:08:55.320 It's about who has the most compute and how fast 0:08:55.360 --> 0:08:57.720 can we get it online? And we're committing to doing 0:08:57.760 --> 0:09:00.920 this together. And the fact is as open ai ad 0:09:00.920 --> 0:09:03.760 buys chips, that's great for AMD. Our revenue goes up, 0:09:03.840 --> 0:09:06.600 our earnings go up. You know, we expect that it 0:09:06.640 --> 0:09:11.160 will also be very very accretive to our shareholders from 0:09:11.280 --> 0:09:13.760 day one. And as we do that, you know, we're 0:09:13.840 --> 0:09:15.960 very happy to have open Ai as a deep partner 0:09:16.000 --> 0:09:19.040 and we win together. So it's like a virtuous positive 0:09:19.080 --> 0:09:21.360 cycle in how we build out. You know, this big 0:09:21.440 --> 0:09:23.880 vision for having all this compute out there, right. 0:09:24.280 --> 0:09:27.319 And yet we still question, as you were just talking about, Greg, 0:09:27.600 --> 0:09:29.040 some of the other supply chain elements. 0:09:29.040 --> 0:09:30.760 You're talking about the need for nuclear for power. 0:09:31.200 --> 0:09:35.080 What's really interesting is we are you feeling confident enough 0:09:35.320 --> 0:09:37.839 about the rest of the compute the supply chain is there, 0:09:37.960 --> 0:09:41.080 is this going to be US manufactured? From your perspective, 0:09:41.120 --> 0:09:43.840 were you looking and also building out internationally with AMD. 0:09:44.840 --> 0:09:48.440 Yeah, we've been looking at really all options our preference 0:09:48.440 --> 0:09:50.400 and really the core thing that we try to do 0:09:50.440 --> 0:09:52.080 is build as much as possible in the US. And 0:09:52.120 --> 0:09:53.800 you can see the commitments that we've made over the 0:09:53.800 --> 0:09:56.480 past year, you know, five hundred billion dollars of investment 0:09:56.520 --> 0:09:59.520 in the US, and that's not stopping. We're continuing to build. 0:10:00.080 --> 0:10:03.240 Think the international that there it is also going to 0:10:03.280 --> 0:10:05.600 be important for the world to have compute. I think 0:10:05.600 --> 0:10:08.120 that computer is going to become this like national security 0:10:08.280 --> 0:10:11.600 strategic resource, and every country is going to need computational 0:10:11.679 --> 0:10:15.400 power and so that we are really not limiting our 0:10:15.559 --> 0:10:17.640 sort of sites in terms of where to build, but 0:10:17.720 --> 0:10:19.480 we do think it is important that the US leads 0:10:19.480 --> 0:10:22.360 in this technology, leads in computational power, and we're expanding 0:10:22.360 --> 0:10:24.240 the supply chain. But you can see that we've really 0:10:24.280 --> 0:10:26.520 been working with partners across the globe in order to 0:10:26.920 --> 0:10:28.960 actually meet the demand that we expect to becoming and 0:10:29.040 --> 0:10:29.720 upcoming years. 0:10:30.240 --> 0:10:33.360 LISTA the manufacturing of these chips. Will do you look 0:10:33.360 --> 0:10:34.520 to Intel at all for it? 0:10:34.520 --> 0:10:37.880 Do you think in the future, Well, as you know, 0:10:38.000 --> 0:10:40.880 the supply chain is something that we work on, you know, 0:10:41.080 --> 0:10:44.400 very very meticulously. I think we have a very strong 0:10:44.400 --> 0:10:48.199 supply chain. We're certainly deeply partnered with you know, TSMC 0:10:48.400 --> 0:10:51.760 across the supply chain. You know, just to that earlier question, 0:10:51.960 --> 0:10:55.559 we're absolutely prioritizing building in the United States because I 0:10:55.640 --> 0:10:59.720 think that's super important. This is the US AI stack. 0:11:00.080 --> 0:11:01.360 We want to have as much of it in the 0:11:01.440 --> 0:11:04.160 US as possible. And you know, we continue to really 0:11:04.160 --> 0:11:06.040 look at, you know, how do we ensure that there 0:11:06.080 --> 0:11:07.600 will be a strong supply chain, you know. 0:11:07.600 --> 0:11:08.160 Going forward. 0:11:08.440 --> 0:11:11.920 Greg Sam posted on x that this deal with AMD 0:11:12.080 --> 0:11:15.320 is incremental to what's already being done with Nvidia, but 0:11:15.679 --> 0:11:17.200 as least enough. So I spent quite a lot of 0:11:17.200 --> 0:11:19.959 time looking at the MII family and the newer generations 0:11:19.960 --> 0:11:23.199 of products to come. Is there a very clear specific 0:11:23.280 --> 0:11:27.960 benefit to using a MD technology for inference relative to 0:11:28.200 --> 0:11:31.000 the capabilities of Nvidia, or do you just see it 0:11:31.080 --> 0:11:34.360 broadly as some sort of diversifying factor. 0:11:34.760 --> 0:11:36.440 Well, I would look at it this way, that there's 0:11:36.480 --> 0:11:39.600 a huge fixed cost to getting AI models running on 0:11:39.640 --> 0:11:42.920 any platform, and so that when we look at what's 0:11:42.960 --> 0:11:45.760 out there, that actually getting AI training to work is 0:11:46.160 --> 0:11:48.680 a huge, huge amount of lift. That's something we've really 0:11:48.720 --> 0:11:51.200 only done the work for in VideA, but for inference 0:11:51.240 --> 0:11:55.199 that that's something that's much more that there's an easier 0:11:55.360 --> 0:11:57.600 barrier to entry there. And one thing we found is 0:11:57.640 --> 0:12:00.320 that I think that the work that Lisai team have 0:12:00.360 --> 0:12:02.640 been doing on the m I four fifty series, it's 0:12:02.679 --> 0:12:04.480 looking like it's going to be a really incredible chip. 0:12:04.720 --> 0:12:07.040 I think that that there's the way that these things 0:12:07.040 --> 0:12:10.360 work is there's niches for different balances of memory and 0:12:10.960 --> 0:12:14.480 computational power, and so as we have a diversity of workload, 0:12:14.520 --> 0:12:17.280 we're finding that having a diversity of chips also really 0:12:17.320 --> 0:12:18.600 accelerates what we're able to do. 0:12:18.920 --> 0:12:20.800 Lisa, at the beginning of this conversation, I said they 0:12:20.800 --> 0:12:24.160 were both operational and financial milestones to be met, and 0:12:24.400 --> 0:12:26.520 Greg explained, you've got to start somewhere. So in the 0:12:26.520 --> 0:12:29.600 first instance, one giga what, but would you just sort 0:12:29.600 --> 0:12:31.800 of draw out the pathway to that first giga what? 0:12:32.720 --> 0:12:35.560 You know, it seems like you're prepared to move quickly here. 0:12:35.720 --> 0:12:38.320 Yeah, absolutely, and maybe if I can just build on 0:12:38.440 --> 0:12:41.439 some things that Greg said. I think he's absolutely right. 0:12:41.520 --> 0:12:44.080 You know, we're a believer in there's a diversity of 0:12:44.080 --> 0:12:46.880 workloads and there will be a diversity of workloads across 0:12:47.240 --> 0:12:51.559 you know, customers, models, use cases, and from that standpoint, 0:12:52.000 --> 0:12:54.240 you know, we feel really good about how we're positioned. 0:12:54.280 --> 0:12:57.319 You know, we we love the work here because you know, Frankly, 0:12:57.840 --> 0:13:00.400 you know, open Ai is the ultimate power user of 0:13:00.400 --> 0:13:03.440 our chips and and test us in very good ways. 0:13:03.880 --> 0:13:06.800 So I think that's that's what gives us confidence that 0:13:06.880 --> 0:13:08.800 you know, the technology is there. And then to your 0:13:08.880 --> 0:13:12.600 point about milestones, yes, I mean this is clearly a 0:13:12.679 --> 0:13:16.480 case where we are tied to each other. The first 0:13:16.520 --> 0:13:19.560 gigawatt of deployment is super important. We're going to start that, 0:13:20.160 --> 0:13:22.080 you know, second half of next year, and we're going 0:13:22.160 --> 0:13:25.360 to build on from there. And it really is not 0:13:25.600 --> 0:13:29.480 just the technology, but your commercial milestones, adoption milestones, and 0:13:29.880 --> 0:13:32.600 just how we proliferate the capability going forward. But I'm 0:13:32.640 --> 0:13:34.840 looking forward to building this as fast as possible. You know, 0:13:34.840 --> 0:13:38.479 we're already working with a number of cloud service providers 0:13:38.679 --> 0:13:41.400 who are also very active on our technology, and I 0:13:41.400 --> 0:13:44.280 think this is a great catalyst to get the industry 0:13:44.360 --> 0:13:45.200 to build faster. 0:13:46.160 --> 0:13:49.160 Tied to each other is such an interesting ton of phrase, 0:13:49.240 --> 0:13:54.000 and Greg, you are seeing more AI users and chip 0:13:54.400 --> 0:13:57.400 makers and designers becoming more financially tied to each other. 0:13:57.880 --> 0:13:59.160 Is this going to continue? 0:13:59.360 --> 0:14:01.320 Is this the staf forward for how you see this 0:14:01.360 --> 0:14:02.400 financing going forward? 0:14:03.360 --> 0:14:06.920 Well, I really see the world transitioning to this AI 0:14:07.000 --> 0:14:09.719 powered economy. And the interesting thing is within open AI 0:14:10.040 --> 0:14:13.719 that we've really seen what it's like when your progress 0:14:14.040 --> 0:14:17.000 is limited and accelerated as two sides of the coin 0:14:17.880 --> 0:14:20.720 by computational power. Like teams within open Ai, that their 0:14:20.800 --> 0:14:23.560 ability to deliver really is tied to the amount of 0:14:23.560 --> 0:14:25.520 compute that they get. And I think we're heading to 0:14:25.560 --> 0:14:28.240 a world where that is how the whole economy will function. 0:14:28.480 --> 0:14:31.040 And we're starting to see it right that people having 0:14:31.080 --> 0:14:33.600 access to better AI tools. If you're a coder, you're 0:14:33.640 --> 0:14:36.040 able to do far more if you have access to 0:14:37.120 --> 0:14:39.080 better AI models. And we're heading to a world where 0:14:39.080 --> 0:14:41.560 if you can have ten times as much AI power 0:14:41.600 --> 0:14:44.600 behind you, you will probably be ten times more productive. 0:14:44.880 --> 0:14:46.360 And so I think that we're moving to a world 0:14:46.360 --> 0:14:48.680 where the whole industry is waking up to the fact 0:14:48.720 --> 0:14:51.560 that we have just not planned. We have not planned 0:14:51.560 --> 0:14:55.080 for this moment where this explosion in AI demand is happening. 0:14:55.240 --> 0:14:57.440 So it's happening all the way from the power to 0:14:57.520 --> 0:14:59.920 the silicon, and I think this whole industry has to 0:15:00.040 --> 0:15:02.920 to find a way to actually rise to meet the occasion. 0:15:03.480 --> 0:15:06.280 Lisa, you have given us a look into the future 0:15:06.320 --> 0:15:09.320 before about how you see the total addressable market the 0:15:09.400 --> 0:15:13.320 industry now that the ink is dry with open AI 0:15:13.440 --> 0:15:17.800 and Greg, are you rethinking either your bigger picture analysis 0:15:18.280 --> 0:15:21.240 of the market for AI accelerators and GPUs or do 0:15:21.320 --> 0:15:24.400 you see AMD now having an improved position in that 0:15:24.440 --> 0:15:27.280 market relative of course to your friends at Nvidia. 0:15:27.800 --> 0:15:30.880 Well, again, I think, and I've told you before, I 0:15:30.920 --> 0:15:33.960 believe that this is a huge market. You know, we 0:15:34.080 --> 0:15:38.640 have size just the AI accelerator TAM being you know, 0:15:38.720 --> 0:15:41.360 over five hundred billion dollars in TAM over the next 0:15:41.640 --> 0:15:44.040 few years. I think some might say, you know, maybe 0:15:44.080 --> 0:15:47.240 I was a little conservative in that TAM analysis, but 0:15:47.320 --> 0:15:49.840 the way to think about it is there's so much 0:15:49.880 --> 0:15:52.080 need from compute. I mean, you just heard it from Greg, 0:15:52.240 --> 0:15:55.120 so you know this is a huge pie and you're 0:15:55.120 --> 0:15:57.520 going to see the need for you know, more players 0:15:57.560 --> 0:16:00.920 coming into it, and you know, from my standpoint, this 0:16:01.000 --> 0:16:04.320 is a big validation of our technology and our capability. 0:16:04.800 --> 0:16:07.000 You know, as much as we love the work with OpenAI, 0:16:07.080 --> 0:16:09.400 we're working with a lot of other customers as well. 0:16:09.440 --> 0:16:11.440 There's a lot of excitement in the industry around m 0:16:11.480 --> 0:16:13.280 I four fifty, so we're ready for it. 0:16:14.040 --> 0:16:17.880 MDCO Lisa Sou Open Ai President Greg Brockman, it's been 0:16:17.920 --> 0:16:19.000 a joy having you on the show. 0:16:19.160 --> 0:16:20.280 Thank you both very much. 0:16:21.280 --> 0:16:24.080 Thank you, and coming up more on AMD and its 0:16:24.200 --> 0:16:26.200 impact on the broader tech markets. We've got a key 0:16:26.200 --> 0:16:28.480 investor for you, Tony one of t row Price. 0:16:28.800 --> 0:16:29.680 This is really metech. 0:16:36.480 --> 0:16:39.800 Let's stay on that AMD open Ai story, it's impact 0:16:39.800 --> 0:16:42.320 of the broader tech markets. Tony Huang t row Price, 0:16:42.400 --> 0:16:44.240 Science and Technology from Portfolio Manager. 0:16:44.400 --> 0:16:46.360 You have exposure to AMD. 0:16:46.600 --> 0:16:48.880 It's added more than seventeen billion dollars in market cap 0:16:48.920 --> 0:16:49.760 on one day alone. 0:16:49.760 --> 0:16:50.800 What does this deal signal? 0:16:52.200 --> 0:16:55.080 Yeah, Well, I think it's a really exciting deal here 0:16:55.120 --> 0:16:59.960 for AMD and validates their roadmap and their silica and software's. 0:17:00.560 --> 0:17:03.240 I think it gives them a really big lead customer 0:17:03.520 --> 0:17:06.639 to really scale and I think that this will probably 0:17:06.680 --> 0:17:10.960 attract other customers to their roadmap and continue to build 0:17:10.960 --> 0:17:11.400 out ROCK. 0:17:11.800 --> 0:17:13.639 And I've always had this like thesis. 0:17:13.280 --> 0:17:17.280 That the TAM is really big and if companies can 0:17:17.359 --> 0:17:21.159 deliver on that compute, that there's there's a lot of 0:17:21.400 --> 0:17:24.639 companies that can win in the overall compute. TAM so 0:17:24.880 --> 0:17:28.160 overall very exciting, and you know, I think it's got 0:17:28.200 --> 0:17:30.879 a good setup for the next few years for the 0:17:30.880 --> 0:17:31.720 company's runway. 0:17:31.840 --> 0:17:33.720 I mean, Lisa Sue just telling us that maybe she's 0:17:33.760 --> 0:17:36.000 being conservative when she puts the AI accelerate to TAM 0:17:36.040 --> 0:17:37.160 at five hundred billion dollars. 0:17:37.240 --> 0:17:38.400 Tony, I'm interested. 0:17:38.080 --> 0:17:40.920 Though, at opening eye telling us Greg saying that we're 0:17:40.920 --> 0:17:42.720 looking at all kinds of way of financing this. 0:17:43.800 --> 0:17:45.280 How are you feeling. 0:17:45.160 --> 0:17:50.840 About this knitting together of purchases and creators, designers of chips. 0:17:52.440 --> 0:17:54.440 Yeah, well, I think that taking your step back over 0:17:54.480 --> 0:17:58.200 the last kind of ten years, everybody's underestimated this TAM. 0:17:58.560 --> 0:17:59.480 And I think what's. 0:17:59.280 --> 0:18:02.360 Exciting about AI is that there are a lot of 0:18:02.440 --> 0:18:05.800 productivity use cases here and we're starting to see AI 0:18:05.920 --> 0:18:08.800 agents really form. And then I think physical AI, so 0:18:09.040 --> 0:18:11.280 I think as long as like the n use cases 0:18:11.359 --> 0:18:13.880 are delivering a lot of value, and there's promise there. 0:18:13.920 --> 0:18:16.280 I think that's what matters in terms of the ROI, 0:18:16.320 --> 0:18:18.400 and I think a lot of these kind of contracts, 0:18:18.400 --> 0:18:21.240 I mean they are backstop by really blue chip companies 0:18:21.280 --> 0:18:22.680 like Mirosov or Oracle. 0:18:23.600 --> 0:18:25.760 So in terms of you know, how I'm feeling. 0:18:25.520 --> 0:18:28.320 About the financing, I think it's it continues to be 0:18:28.359 --> 0:18:30.480 probably a good attractive way to do as long as 0:18:30.520 --> 0:18:32.080 the use cases are intact. 0:18:32.760 --> 0:18:36.120 Tony, the mechanics of the deal are the open AI 0:18:36.359 --> 0:18:38.600 has the right to buy up to one hundred and 0:18:38.680 --> 0:18:42.399 sixty million shares of AMD at a penny apiece, but 0:18:42.480 --> 0:18:46.840 that right contingent on the operational and financial milestones that 0:18:46.840 --> 0:18:51.080 we discussed. It is not circular financing, it's something different. 0:18:51.560 --> 0:18:55.200 But how comfortable are you with that mechanism. 0:18:55.320 --> 0:18:57.320 Well, I think it could make a lot of sense, right. 0:18:57.840 --> 0:19:00.639 I mean it puts the two companies of in co 0:19:00.840 --> 0:19:05.440 development and partnership, and I think there's shared economics and 0:19:05.440 --> 0:19:07.719 it's a win win for both companies. I think some 0:19:07.800 --> 0:19:10.040 of the prices the warrants are stricted at a pretty 0:19:11.600 --> 0:19:14.159