WEBVTT - AMD CEO Lisa Su & OpenAI President Greg Brockman Talk New Partnership 0:00:02.520 --> 0:00:07.000 Bloomberg Audio Studios, podcasts, radio news. 0:00:07.880 --> 0:00:11.000 AMD has signed a definitive agreement with open ai to 0:00:11.080 --> 0:00:14.920 deploy six gigawatts of AMD GPUs. AMD says it will 0:00:14.920 --> 0:00:17.920 equate to tens of billions of dollars in revenue. Open 0:00:17.960 --> 0:00:20.280 Ai will get up to one hundred and sixty million 0:00:20.320 --> 0:00:24.240 am D shares in tranches and set against both operational 0:00:24.440 --> 0:00:28.640 and financial milestones, the focus is inference. Let's bring in 0:00:28.720 --> 0:00:31.840 AMD CO Lisa Sou and Open Ai president Greg Brockman. 0:00:32.120 --> 0:00:34.199 Both of them join us on set here at Bloomberg 0:00:34.240 --> 0:00:36.840 Tech in San Francisco. Good morning, Good morning, It's great 0:00:36.880 --> 0:00:40.280 to see you here. Let's frame the opportunity. Lisa, you 0:00:40.320 --> 0:00:43.040 know that the market reaction is very clear, But for 0:00:43.159 --> 0:00:46.080 AMD and the AI industry at large, what do you 0:00:46.080 --> 0:00:47.120 think this represents. 0:00:47.560 --> 0:00:50.600 Well, look, this is a huge milestone for AMD. You know, 0:00:50.640 --> 0:00:53.360 we are so thrilled with the partnership with the open 0:00:53.400 --> 0:00:55.760 Ai team, and it's also you know, a huge moment 0:00:55.800 --> 0:00:58.200 for the AI industry because you know, when you get 0:00:58.200 --> 0:01:01.040 to the breakdown to it, you need more AI compute. 0:01:01.080 --> 0:01:02.920 I mean, that's where we are today. Compute is a 0:01:02.960 --> 0:01:05.240 foundation for all of the intelligence we can get from 0:01:05.240 --> 0:01:08.280 AI and you know, we are a compute provider. We 0:01:08.360 --> 0:01:11.840 have spent years on our roadmap. We've spent years working 0:01:11.880 --> 0:01:14.920 with open Ai and the team and you know, together 0:01:15.080 --> 0:01:17.880 now we're embarking on you know, a massive build out 0:01:18.240 --> 0:01:21.679 of six gigawatts of AI compute, and it's it's a 0:01:21.680 --> 0:01:24.640 big deal for us, for our shareholders, for our teams, 0:01:25.000 --> 0:01:27.560 and for really you know, the partnership and the overall 0:01:27.600 --> 0:01:28.520 AI ecosystem. 0:01:28.760 --> 0:01:31.200 Greg I say that the top the focus is inference. 0:01:31.319 --> 0:01:34.680 I think that's really important to be specific about what 0:01:34.720 --> 0:01:37.319 you will do with this capacity. So so literally explain 0:01:37.440 --> 0:01:39.480 that part. And I'm conscious that you know, in the 0:01:39.480 --> 0:01:42.560 first instance, the first target is one gigawatt and then 0:01:42.600 --> 0:01:45.360 eventually six gigawatts. But what will you use it for? 0:01:46.080 --> 0:01:49.440 Well, I think that the world continues to underestimate the 0:01:49.480 --> 0:01:53.080 amount of demand for AI compute, right that just we've 0:01:53.120 --> 0:01:55.960 seen this explosion of demand with things like chat GBT. 0:01:56.280 --> 0:01:58.320 You know, we're at eight hundred million weekly active users. 0:01:58.320 --> 0:02:00.800 Now this probably didn't even exist three years go, and 0:02:00.840 --> 0:02:03.720 we're in a position where we cannot launch futures. We 0:02:03.760 --> 0:02:06.840 cannot launch new products simply because of lack of computational 0:02:06.880 --> 0:02:10.400 power and we see these models continuing to get exponentially better, 0:02:10.840 --> 0:02:12.359 and I think we're just heading to a world where 0:02:12.360 --> 0:02:13.960 so much of the economy is going to be lifted 0:02:14.040 --> 0:02:16.359 up and driven by progress and AI. And so we're 0:02:16.440 --> 0:02:18.120 very much heading to a world by default that I 0:02:18.120 --> 0:02:20.320 think looks like a compute desert, right that there's just 0:02:20.360 --> 0:02:22.360 not enough compute to go around, and so we're trying 0:02:22.360 --> 0:02:24.560 to build as much as possible, as quickly as possible. 0:02:24.840 --> 0:02:27.000 So we're starting with one gigawatt simply because you've got 0:02:27.040 --> 0:02:29.920 to start somewhere, But honestly, we're building as fast as 0:02:29.919 --> 0:02:32.600 we possibly can and trying to bring as much computational 0:02:32.600 --> 0:02:35.240 power to bear for the economy and for the world. 0:02:35.960 --> 0:02:39.520 Lisa, this is such a big commitment to Instinct in 0:02:39.560 --> 0:02:42.720 particular as a customer. Does it make open AI the 0:02:42.800 --> 0:02:44.880 largest for that particular product. 0:02:45.680 --> 0:02:48.640 Well, this is certainly the largest deployment that we have 0:02:48.639 --> 0:02:51.880 announced by far. I mean, you know, six gigawatts of compute. 0:02:52.280 --> 0:02:53.839 As Greg said, we're going to start with the first 0:02:53.880 --> 0:02:56.560 gigawatt in the second half of twenty twenty six on 0:02:56.639 --> 0:03:01.240 our new next generation four fifty chip. I think the 0:03:01.280 --> 0:03:04.120 thing to understand is, you know, these types of partnerships 0:03:04.200 --> 0:03:07.480 actually take you know, years to really get comfortable with 0:03:07.560 --> 0:03:09.320 the idea that we're going to you know, go all 0:03:09.360 --> 0:03:12.679 in together. And this isn't all in partnership in terms 0:03:12.720 --> 0:03:16.800 of building out you know, the AI compute that open 0:03:16.840 --> 0:03:19.680 ai needs for everything that they're offering to the world. So, yes, 0:03:19.720 --> 0:03:22.720 it's a huge deal, and it also says a lot 0:03:22.840 --> 0:03:25.200 about you know, how much needs to come together for 0:03:25.480 --> 0:03:27.800 you know, this entire ecosystem to operate. So you know, 0:03:27.840 --> 0:03:29.720 we are setting up you know, certainly there's a lot 0:03:29.760 --> 0:03:34.040 of engineering work, but our teams are working together on hardware, software, 0:03:34.120 --> 0:03:37.120 We're ensuring the supply chain, all of those elements are 0:03:37.160 --> 0:03:40.960 set up and ready to deliver on this massive commitment. 0:03:41.520 --> 0:03:44.480 Greg, talk us through a little bit about the players 0:03:44.480 --> 0:03:46.080 that you need to also lean on. This has been 0:03:46.200 --> 0:03:47.680 years in the making, as you say. 0:03:47.520 --> 0:03:48.080 With a m D. 0:03:48.280 --> 0:03:51.280 But what other cloud providers were involved? How you thinking 0:03:51.320 --> 0:03:54.240 about this working with an Oracle or others out there. 0:03:55.120 --> 0:03:57.640 Yeah, we really think of this as an industry wide effort, 0:03:57.960 --> 0:04:00.200 and in general, we think that compute is something that 0:04:00.560 --> 0:04:03.080 does require the entire supply chain to really wake up 0:04:03.120 --> 0:04:05.760 and to really to start building much more than people 0:04:05.760 --> 0:04:08.680 we're planning on. I think this starts from energy to 0:04:08.720 --> 0:04:12.080 try to get far more power to be built. Things 0:04:12.120 --> 0:04:13.880 like nuclear I think are going to be very important 0:04:13.880 --> 0:04:16.120 to come online. The cloud providers are an important part 0:04:16.120 --> 0:04:17.640 of this as well. So we're going to be deploying 0:04:17.680 --> 0:04:20.000 AMD in our own data centers. We'll be deploying them 0:04:20.279 --> 0:04:22.560 together with cloud providers. You know, we have a deal 0:04:22.600 --> 0:04:25.919 with Oracle, lots of other cloud providers out there. You 0:04:25.920 --> 0:04:28.320 can really see that we're very much in the We 0:04:28.440 --> 0:04:30.720 just want compute as much compute as possible. We think 0:04:30.760 --> 0:04:32.640 this is important for the economy, we think this is 0:04:32.640 --> 0:04:35.120 important for the nation, we think this is important for humanity. 0:04:35.440 --> 0:04:37.600 And so really we're working with everyone in this whole 0:04:37.640 --> 0:04:40.640 industry in order to get as much compute power online 0:04:40.720 --> 0:04:41.680 as quickly as we can. 0:04:42.279 --> 0:04:46.440 Lisa, I'm sorry specifics where is this data center going 0:04:46.480 --> 0:04:50.960 to be? Is it one single site? Is it Oracle 0:04:51.800 --> 0:04:53.039 that we'll partner with you on this? 0:04:53.520 --> 0:04:56.680 Well, actually, what this really is is an announcement of 0:04:56.720 --> 0:04:58.440 what you know, AMD and open a are going to 0:04:58.480 --> 0:05:00.800 do together. You know, open a I has a lot 0:05:00.800 --> 0:05:03.280 of partners in terms of you know, where they deploy 0:05:03.680 --> 0:05:06.159 I imagine a lot of it will be in cloud 0:05:06.200 --> 0:05:08.800 service providers. It's really up to you know, open Ai 0:05:08.920 --> 0:05:12.120 and Greg and Sam and the team. But the way 0:05:12.160 --> 0:05:15.760 to think about it is, for this amount of compute, 0:05:15.800 --> 0:05:17.160 it's going to have to be in a lot of 0:05:17.160 --> 0:05:18.320 different places. 0:05:18.360 --> 0:05:20.560 It's a massive amount, multiple locations. 0:05:20.120 --> 0:05:24.279 Multiple locations, I would imagine, you know, multiple providers to 0:05:24.360 --> 0:05:26.160 really get this online as fast as possible. 0:05:26.520 --> 0:05:30.679 Greg, there is a lot of focus on where open 0:05:30.760 --> 0:05:32.919 ai is going to get the money from to fund 0:05:32.920 --> 0:05:38.280 all of this. Sam Altman's big picture commitment is well documented, right, 0:05:38.320 --> 0:05:41.000 and the numbers to his mind are in the trillions. 0:05:41.520 --> 0:05:45.000 But have you specifically thought about debt financing for this 0:05:45.080 --> 0:05:49.200 relationship with a MD? Have you thought about doing a 0:05:49.240 --> 0:05:53.640 specific equity raise? You are very committed across multiple projects. 0:05:53.880 --> 0:05:54.080 Yeah. 0:05:54.120 --> 0:05:55.520 Look, the way that I would the way that I 0:05:55.520 --> 0:05:58.640 would look at this is that AI revenue is growing 0:05:58.680 --> 0:06:02.360 faster than I think almost any product in history, and 0:06:02.600 --> 0:06:04.719 that ultimately, at the end of the day, the reason 0:06:04.720 --> 0:06:07.359 this compute power is so important and is so worthwhile 0:06:07.440 --> 0:06:11.040 for everyone to build is because the revenue ultimately will 0:06:11.040 --> 0:06:13.760 be there. Now as a company that is trying to 0:06:13.800 --> 0:06:16.080 move as fast as we can, we look at everything right, 0:06:16.080 --> 0:06:19.680 we look at equity debt, we look at trying to 0:06:19.680 --> 0:06:23.120 find creative ways of financing all of this. That's been 0:06:23.160 --> 0:06:25.000 actually a huge focus of us for the past couple 0:06:25.040 --> 0:06:27.840 of years as thinking about how can we possibly build 0:06:27.920 --> 0:06:30.280 the amount of compute that is required in order to 0:06:30.400 --> 0:06:33.359 really transform this whole economy into an aipowered economy. And 0:06:33.440 --> 0:06:37.200 so I think you'll see lots of creative ideas, but fundamentally, 0:06:37.360 --> 0:06:38.680 I think at the end of the day, it is 0:06:38.720 --> 0:06:39.359 because we believe. 0:06:39.839 --> 0:06:41.919 Sorry to jump in an interrupt and carriage, just forgive 0:06:41.960 --> 0:06:46.839 me on this one. The condition of AMD issuing the 0:06:46.920 --> 0:06:50.560 stock to open Ai requires you to spend money basically 0:06:50.600 --> 0:06:55.200 because you have to deliver that gig awad of capacity first. Lisa, 0:06:55.240 --> 0:06:58.039 I have to ask you if you have assurances that 0:06:58.120 --> 0:06:59.240 open Ai is good for it. 0:06:59.640 --> 0:07:02.200 Well, let me be clear. I mean, this deal is 0:07:02.240 --> 0:07:05.160 a win for am D, it's a win for open Ai, 0:07:05.360 --> 0:07:07.600 and it's a win for our shareholders. And that's kind 0:07:07.600 --> 0:07:10.160 of the way we put this together. I have full 0:07:10.200 --> 0:07:14.120 confidence in you know, open Ai, Sam, Greg Sarah. I mean, 0:07:14.160 --> 0:07:17.480 this is a massive opportunity for us right now. Right here, 0:07:17.600 --> 0:07:20.840 it's about who has the most compute and how fast 0:07:20.840 --> 0:07:23.200 can we get it online? And we're committing to doing 0:07:23.240 --> 0:07:27.040 this together. And the fact is as open ai buys chips, 0:07:27.080 --> 0:07:29.880 that's great for AMD. Our revenue goes up, our earnings 0:07:29.920 --> 0:07:32.600 go up. You know, we expect that it will also 0:07:32.760 --> 0:07:37.240 be very very accretive to our shareholders from day one. 0:07:37.720 --> 0:07:39.800 And as we do that, you know, we're very happy 0:07:39.840 --> 0:07:41.720 to have open Ai as a deep partner and we 0:07:41.800 --> 0:07:45.120 win together. So it's like a virtuous positive cycle in 0:07:45.160 --> 0:07:47.480 how we build out. You know, this big vision for 0:07:47.560 --> 0:07:50.440 having all this compute out there, and yet we. 0:07:50.480 --> 0:07:53.200 Still question as you were just talking about greg some 0:07:53.240 --> 0:07:55.080 of the other supply chain elements. You're talking about the 0:07:55.160 --> 0:07:58.080 need for nuclear for power. What's really interesting is we 0:07:58.480 --> 0:08:02.320 are you feeling confident enough about the rest of the compute, 0:08:02.320 --> 0:08:04.040 the supply chain is there? Is this going to be 0:08:04.200 --> 0:08:07.560 US manufactured? From your perspective, were you looking and also 0:08:07.600 --> 0:08:09.320 building out internationally with MD. 0:08:10.360 --> 0:08:13.920 Yeah, we've been looking at really all options our preference 0:08:13.920 --> 0:08:15.880 and really the core thing that we try to do 0:08:15.920 --> 0:08:17.600 is build as much as possible in the US. And 0:08:17.600 --> 0:08:19.280 you can see the commitments that we've made over the 0:08:19.280 --> 0:08:21.920 past year, you know, five hundred billion dollars of investment 0:08:22.000 --> 0:08:25.040 in the US, and that's not stopping. We're continuing to build. 0:08:25.240 --> 0:08:28.520 I do think that international that there it is also 0:08:28.520 --> 0:08:30.320 going to be important for the world to have compute. 0:08:30.840 --> 0:08:32.560 I think that computer is going to become this like 0:08:32.679 --> 0:08:36.240 national security strategic resource, and every country is going to 0:08:36.280 --> 0:08:39.440 need computational power, and so that we are really not 0:08:39.679 --> 0:08:43.000 limiting our sort of sites in terms of where to build. 0:08:43.040 --> 0:08:44.600 But we do think it is important that the US 0:08:44.720 --> 0:08:47.439 leads in this technology, leads in computational power, and we're 0:08:47.440 --> 0:08:49.559 expanding the supply chain. But you can see that we've 0:08:49.559 --> 0:08:51.920 really been working with partners across the globe in order 0:08:51.960 --> 0:08:54.360 to actually meet the demand that we expect to becoming 0:08:54.360 --> 0:08:55.240 in upcoming years. 0:08:55.720 --> 0:08:58.960 Lisa, the manufacturing of these chips, will you look to 0:08:59.000 --> 0:09:00.840 Intel at all for it? Do you think of the future? 0:09:02.240 --> 0:09:04.800 Well, as you know, the supply chain is something that 0:09:04.840 --> 0:09:08.920 we work on, you know, very very meticulously. I think 0:09:08.960 --> 0:09:11.960 we have a very strong supply chain. We're certainly deeply 0:09:12.040 --> 0:09:15.680 partnered with you know, TSMC across the supply chain. You know, 0:09:15.800 --> 0:09:19.599 just to that earlier question, we're absolutely prioritizing building in 0:09:19.600 --> 0:09:22.800 the United States because I think that's super important. This 0:09:22.880 --> 0:09:26.160 is the US AI stack. We want to have as 0:09:26.200 --> 0:09:28.320 much of it in the US as possible, and you know, 0:09:28.400 --> 0:09:30.480 we continue to really look at, you know, how do 0:09:30.559 --> 0:09:32.800 we ensure that there will be a strong supply chain, 0:09:32.840 --> 0:09:33.640 you know, going forward. 0:09:33.960 --> 0:09:37.120 Greg Sam posted on x that this deal with a 0:09:37.240 --> 0:09:40.079 m D is incremental to what's already being done with Nvidia. 0:09:40.720 --> 0:09:42.560 But as least know so, I spent quite a lot 0:09:42.600 --> 0:09:44.840 of time looking at them I family and the newer 0:09:44.880 --> 0:09:48.040 generations of products to come. Is there a very clear 0:09:48.160 --> 0:09:52.679 specific benefit to using a m D technology for inference 0:09:52.840 --> 0:09:56.080 relative to the capabilities of Nvidia, or do you just 0:09:56.120 --> 0:09:59.840 see it broadly as some sort of diversifying factor. 0:10:00.240 --> 0:10:01.920 Well, I would look at it this way, that there's 0:10:01.960 --> 0:10:05.080 a huge fixed cost to getting AI models running on 0:10:05.120 --> 0:10:08.400 any platform, and so that when we look at what's 0:10:08.440 --> 0:10:11.240 out there, that actually getting AI training to work is 0:10:11.640 --> 0:10:14.160 a huge, huge amount of lift. That's something we've really 0:10:14.200 --> 0:10:16.679 only done the work for in Vidia, but for inference, 0:10:16.720 --> 0:10:20.679 that that's something that's much more that there's an easier 0:10:20.840 --> 0:10:23.080 barrier to entry there. And one thing we found is 0:10:23.120 --> 0:10:25.680 that I think that the work that Lisa and team 0:10:25.720 --> 0:10:27.560 have been doing on the M four to fifty series. 0:10:27.960 --> 0:10:30.000 It's looking like it's going to be a really incredible chip. 0:10:30.200 --> 0:10:32.679 I think that there's the way that these things work 0:10:32.760 --> 0:10:35.839 is that there's niches for different balances of memory and 0:10:36.440 --> 0:10:40.040 computational power, and so as we have a diversity of workloads, 0:10:40.040 --> 0:10:42.760 we're finding that having a diversity of chips also really 0:10:42.800 --> 0:10:44.079 accelerates what we're able to do. 0:10:44.400 --> 0:10:46.280 Lisa. At the beginning of this conversation, I said, there 0:10:46.280 --> 0:10:49.640 are both operational and financial milestones to be met, and 0:10:49.880 --> 0:10:52.000 Greg explained, you've got to start somewhere. So in the 0:10:52.000 --> 0:10:55.120 first instance, one giga what But would you just sort 0:10:55.120 --> 0:10:57.280 of draw out the pathway to that first giga what? 0:10:58.200 --> 0:11:01.040 You know, it seems like you're prepared to move quickly here. 0:11:01.200 --> 0:11:01.440 Yeah. 0:11:01.440 --> 0:11:03.679 Absolutely, And and maybe ed, if I can just build 0:11:03.720 --> 0:11:06.920 on something that Greg said, I think he's absolutely right. 0:11:07.000 --> 0:11:09.560 You know, we're a believer in there's a diversity of 0:11:09.559 --> 0:11:12.360 workloads and there will be a diversity of workloads across 0:11:12.720 --> 0:11:17.040 you know, customers, models, use cases, and from that standpoint, 0:11:17.480 --> 0:11:19.720 you know, we feel really good about how we're positioned. 0:11:19.760 --> 0:11:22.800 You know, we we love the work here because you know, frankly, 0:11:23.320 --> 0:11:25.880 you know, open Ai is the ultimate power user of 0:11:25.920 --> 0:11:28.920 our chips and and and test us in very good ways. 0:11:29.360 --> 0:11:32.280 So I think that's that's what gives us confidence that 0:11:32.360 --> 0:11:34.280 you know, the technology is there. And then to your 0:11:34.360 --> 0:11:37.080 point about milestones, Yes, I mean this is you know, 0:11:37.120 --> 0:11:41.559 clearly a case where we are tied to each other. Uh, 0:11:41.600 --> 0:11:44.440 the first gigawatt of deployment is super important. We're going 0:11:44.520 --> 0:11:47.000 to start that, you know, second half of next year, 0:11:47.240 --> 0:11:49.520 and we're going to build on from there. And it 0:11:49.600 --> 0:11:53.640 really is not just the technology, but you know, commercial milestones, 0:11:53.679 --> 0:11:57.080 adoption milestones, and and just how we proliferate the capability 0:11:57.120 --> 0:11:59.280 going forward. But I'm looking forward to building this as 0:11:59.280 --> 0:12:01.400 fast as possible. Well, you know, we're already working with 0:12:01.480 --> 0:12:05.040 a number of cloud service providers who are also very 0:12:05.120 --> 0:12:07.720 active on our technology, and I think this is a 0:12:07.760 --> 0:12:10.680 great catalyst to get the industry to build faster. 0:12:11.679 --> 0:12:14.640 Tied to each other is such an interesting turn of phrase. 0:12:14.720 --> 0:12:19.199 And Greg, look, you are seeing more AI users and 0:12:19.320 --> 0:12:22.880 chip makers and designers becoming more financially tied to each other. 0:12:23.360 --> 0:12:26.000 Is this going to continue? Is this the step forward 0:12:26.080 --> 0:12:27.920 for how you see this financing going forward? 0:12:28.840 --> 0:12:32.480 Well, I really see the world transitioning to this AI 0:12:32.480 --> 0:12:35.240 powered economy and The interesting thing is within open AI 0:12:35.520 --> 0:12:39.199 that we've really seen what it's like when your progress 0:12:39.520 --> 0:12:42.480 is limited and accelerated as true sides of the coin 0:12:43.400 --> 0:12:46.199 by computational power, like teams within open EI, that their 0:12:46.280 --> 0:12:49.040 ability to deliver really is tied to the amount of 0:12:49.040 --> 0:12:51.000 compute that they get. And I think we're heading to 0:12:51.040 --> 0:12:53.720 a world where that is how the whole economy will function. 0:12:54.000 --> 0:12:56.520 And we're starting to see it right that people having 0:12:56.559 --> 0:12:59.079 access to better AI tools. If you're a coder, you're 0:12:59.120 --> 0:13:01.320 able to do far more, or if you have access 0:13:01.480 --> 0:13:04.360 to better AI models, And we're heading to a world 0:13:04.360 --> 0:13:06.640 where if you can have ten times as much AI 0:13:06.760 --> 0:13:10.080 power behind you, you will probably be ten times more productive. 0:13:10.360 --> 0:13:11.840 And so I think that we're moving to a world 0:13:11.880 --> 0:13:14.200 where the whole industry is waking up to the fact 0:13:14.200 --> 0:13:17.000 that we have just not planned. We have not planned 0:13:17.040 --> 0:13:20.559 for this moment where this explosion in AI demand is happening. 0:13:20.720 --> 0:13:22.960 So it's happening all the way from the power to 0:13:23.000 --> 0:13:25.560 the silicon, and I think this whole industry has to 0:13:25.600 --> 0:13:28.400 find a way to actually rise to meet the occasion. 0:13:28.960 --> 0:13:31.800 Lisa, you have given us a look into the future 0:13:31.800 --> 0:13:34.800 before about how you see the total addressable market the 0:13:34.880 --> 0:13:38.800 industry now that the ink is dry with open AI 0:13:38.920 --> 0:13:43.280 and Greg, are you rethinking either your bigger picture analysis 0:13:43.760 --> 0:13:46.720 of the market for AI accelerators and GPUs or do 0:13:46.800 --> 0:13:49.880 you see AMD now having an improved position in that 0:13:49.960 --> 0:13:52.760 market relative of course to your friends at Nvidia. 0:13:53.280 --> 0:13:56.360 Well, again, I think, and I've told you before, I 0:13:56.400 --> 0:13:59.880 believe that this is a huge market we have saw 0:14:00.200 --> 0:14:04.440 is just the AI accelerator TAM being you know, over 0:14:04.480 --> 0:14:07.720 five hundred billion dollars in TAM over the next few years. 0:14:07.960 --> 0:14:09.760 I think some might say, you know, maybe I was 0:14:09.800 --> 0:14:13.080 a little conservative in that TAM analysis, but the way 0:14:13.080 --> 0:14:16.160 to think about it is there's so much need from compute. 0:14:16.200 --> 0:14:18.080 I mean, you just heard it from Greg, so you 0:14:18.080 --> 0:14:20.800 know this is a huge pie and you're going to 0:14:20.880 --> 0:14:23.680 see the need for you know, more players coming into it. 0:14:23.760 --> 0:14:27.280 And you know, from my standpoint, this is a big 0:14:27.360 --> 0:14:30.680 validation of our technology and our capability. You know, as 0:14:30.760 --> 0:14:32.720 much as we love the work with open AI, we're 0:14:32.760 --> 0:14:35.120 working with a lot of other customers as well. There's 0:14:35.160 --> 0:14:37.040 a lot of excitement in the industry around m I 0:14:37.080 --> 0:14:38.760 four fifty, so we're ready for it. 0:14:39.520 --> 0:14:42.840 M d C Lisa s f and AI President Greg Brockman, 0:14:43.040 --> 0:14:44.840 it's been a joy having you on the show. Thank 0:14:44.880 --> 0:14:45.760 you both very much.