AMD-OpenAI Partnership, Crescendo’s AI Rollup Strategy, Founders Fund’s Big AI Bet | Oct 6, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Pasricha. It is Monday, October 6th, and what a busy Monday we have today, folks. We have got the big news about OpenAI and AMD's new deal together that we're going to get to. And of course, OpenAI has its Developer Day happening in San Francisco today, so there is a lot going on. We're also going to bring on Venture Fund M13 to talk about the white space one of its partners thinks that OpenAI won't end up encroaching on. And separately, we're talking to Crescendo about the business of AI rollups. Last but not least, we've got a great story for you today about Founders Fund that we just published on our website.

0:53We're going to break that down with our AI and finance reporter, Miles Krupa. There is a lot going on, so let's get right on into it. OpenAI and AMD are striking a major partnership. Starting in late 2026, OpenAI will deploy 6 gigawatts of AMD chips over multiple years. And also, it could take up to a 10 % stake in AMD if certain milestones are met. AMD shares are soaring on the news. And I want to bring on Austin Lyons, Senior Analyst at Creative Strategies, to break this all down for us. Austin, welcome to the show. It's your first time here. How are you doing? Hey, Kosh. Doing good. What an exciting Monday.

1:33Thanks for having me. I know. I was going to say, gosh, we knew we had Developer Day, but this news took us very much by surprise. Give us your initial reactions. What did you think? You know, initially, both surprised and not surprised. Surprised by the terms and the way it was structured. But let me start with not surprised. So I was at AMD's Advancing AI Day, and they had Sam Altman on stage. And Sam talked about how he's been working closely. He and his team have been working closely with Lisa Su and AMD, working to make sure that OpenAI can run their compute on AMD's instinct offering, and making sure that the software was good enough, making sure that the hardware is good enough.

2:14And Sam specifically said he was very excited about the MI450 series, the Helios rack. And he talked about just being really excited about that. So this was sort of like a culmination of his excitement and Lisa's excitement where they came together and said, yes, this is happening. OpenAI is going to be incentivized to truly commit to deploying those chips at scale. These are the latest chips that AMD has put out. Yeah, yeah. In fact, these are actually coming next year. So AMD's most recent ones are the 350 series, and this 450 series is coming next year. And what's important about this one in this Helios rack is that this will be AMD's first sort of rack scale system with 72 GPUs.

2:55So it's finally sort of catching up at that scaled approach to NVIDIA. So you could say this is where they're really starting to become competitive. So Sam Altman expressed excitement for buying into AMD's technology here with this RAC system that it has. What did you make about the fact here that we have this$100 billion letter of intent from NVIDIA, right? A lot of people thought that, hey, this is a play from NVIDIA to sort of deter OpenAI from maybe using other companies like AMD's chips. Here we have OpenAI saying, hey, we're using them all. And not only that, we might take equity in your company too.

3:35Yeah, that's the surprising part. So, you know, NVIDIA came to the table and said, hey, Sam wants more compute and we have a lot of free cash flow. So we have the cash to help invest and pull forward his compute needs and help him get the compute he needs. And so, of course, now, if you put yourself in Lisa Su's shoes, how do you respond to that? You know, AMD doesn't have 100 billion to match and invest in OpenAI. So this was an interesting way for Lisa to sort of counter and say, hey, we want to make sure that OpenAI is incentivized to still use our chips. We don't want to get shut out once we're finally competitive.

4:13and so instead of saying i can offer you 100 billion this was a very interesting way where lisa's basically saying i can grant you warrants um to exercise to basically up to 10 of our company if you partner with us so this was a very interesting counterpunch to make sure that amd still has a seat at the table uh and yeah so i've got to be interesting for jensen well i was gonna say if you're jensen huang this morning are you worried No, I don't think Jensen's worried. You know, NVIDIA is definitely dominant. They have a great relationship with OpenAI. But to me, this just signals that there's just such a demand for compute.

4:52You know, we need more compute. And Andy's saying we can help. Right. What about the AI chip startups that are trying to capture some market share here? If you're one of those CEOs, what are you thinking this morning? Yeah, great question. So if you're an AI chip startup, you're thinking, okay, great. OpenAI is still willing to seek compute from other players. On the other hand, you still have to ask, where do I fit into this world? NVIDIA and AMD are both making these still more flexible GPUs that can run any shape of workload, whether that's video models, whether it's reasoning models, or just the fast thinking models.

5:34AMD and NVIDIA's GPUs can do that. If you're an AI ASIC company, you have to ask sort of, do we have a way to get in here and tell OpenAI that we can run a particular shape of workload even more efficiently than these guys? The good news for these AI ASIC startups is that should they have, like maybe they can say like, hey, we're the best for video model generation, really long context, huge memory, fast memory bandwidth. What interestingly, what's good for them is what OpenAI is signaling is they're willing to do the work to run their software on heterogeneous compute. So obviously, they're running on top of AMD's ROCAM, they're running on top of NVIDIA's CUDA.

6:18So maybe that's... And bring it down, just one level of complexity for us. When you say heterogeneous compute, what do you mean by that very quickly? Yeah, totally, totally. So that just means many different chip vendors. So not just one chip, essentially, you're willing to use NVIDIA, AMD, maybe even Grok one day, you're willing to use all these vendors. Exactly. And why that is important is there is a trade-off there. If you're willing to use lots of different hardware, you have to make sure your software can run on all that different hardware. So there's work that OpenAI has to do. In an ideal world, it's just write once and run anywhere, but we're not there yet.

6:57And OpenAI is saying they're willing to take that effort to make sure their software runs on NVIDIA, runs on AMD, Could they make sure that it runs on someone else, Grok or someone else? We've talked a lot on the show about OpenAI's own ambitions to one day perhaps build their own chip. Do you think this says anything about their own ambitions there? That's a good question. You know, I think they'll still pursue their own chip. I think OpenAI will have a good sense as their workloads shape up what their specific needs are. They'll continue to use GPUs because GPUs are flexible. and as workloads change, as the underlying models change, GPUs are just the best place for those to land.

7:38On the other hand, as some workloads, like let's say deep research, there's a lot of people that run OpenAI's deep research. It's very computationally expensive. Could OpenAI make a deep research cluster that's tuned at the system level just for deep research? Definitely. And still making their own chip might be a way to do that. And last question for you, how do you think about the power demands that all of these deals and all this compute is going to need to take. Did we glean anything from this announcement or did it raise more questions? Akash, that is it. So, you know, the question is, where's the six gigawatts coming from?

8:13Where's the 10 gigawatts coming from? You know, now OpenAI says, we're going to get these chips. And the partner's like, we're going to have the... We'll have them ready for you. We'll give them to you. Exactly. And so naturally, the next question is, where are they going to go? What real estate are they sitting in? And where's the power coming from? Is it Oracle and Stargate? Is it CoreWeave? If you're a neocloud, you have to be excited about this as well. Anyone who has power out there needs to be calling Lisa Su and saying, we should talk. Great. Well, thank you so much for coming on the show, Austin.

8:45Like I said, we've got so much news that is going to continue to come out today with the Developers Day, but we started the morning off big with this OpenAI AMD announcement. That is Austin Lyons, a senior analyst at Creative Strategies. And this is his first time here on TI TV, but we look forward to having you on once again very soon. Okay, for our next segment, aside from its cloud deals and its chip deals, OpenAI has been making a ton of noise on the applications front. That, of course, seems to be the main date of its new CEO of applications, Fiji SEMO. And it all raises a big question around what parts of the consumer technology sector will be left untouched and where there's still opportunities for smaller startups to succeed.

9:25I want to bring on someone who has been thinking about that exact point Anna Barber is a partner at Venture Fund M13. She has written extensively on this topic. Anna, welcome to the show. It's great to have you. It's great to be here, Akash. Thanks for having me. Of course. Especially with me like today. I know. So much news and so much more to come. I want to talk about the sort of applications layer that OpenAI is building. We just talked about the chip layer and its deal with AMD. But, you know, you've put some great thought into the consumer side of this story. And I want to ask you, just open-ended here, OpenAI is doing agents.

10:03They've talked about doing browsers. They've now revealed their social media feed with Sora 2 last week. You seem to think that there is still opportunity for startups to build. And so what I want to know is, where is that opportunity? Because it seems like OpenAI is doing just about everything right now. They're doing a lot. They're making announcements fast and furious, and we'll be expecting more today. But I still think what they're doing fundamentally is expanding their platform to allow others to build applications with more vertical focus on top of it. And I think there's several categories where they're just, they're not going to play as aggressively.

10:40And in some cases, even where they do, they may not be successful. Okay, where are those categories? So a couple of things come to mind. So one is anything that is closely related to your, on the consumer side, so let's look at consumer for a second, your personal data and personal decisions about your life. So I realize they did make an acquisition recently of a team working on a consumer finance application, but I think it's going to be a long time before people turn their portfolio management or their financial life over to OpenAI. I think that's such a personal decision. It's something where I think you need a custom dedicated vertical approach there and a brand and a level of customer support that I think is going to be hard for them to deliver.

11:31Another idea sort of in that same vein, I think, is matchmaking. So there's been a real renewed interest in AI matchmakers. This technology is so great for helping people sort through all the choices they have in the dating world. And I think that's something OpenAI isn't going to want to touch. Okay. So we've got financial advice and we've got matchmaking, which are two very separate, but I guess in some cases, related forms of advice that people are seeking. My question for you is, on the finance side, is this the regulations here that are the reason that you think that OpenAI might not be able to make as deep a step into that space?

12:16Is it the same as healthcare there? Do you think it will just not catch on? I mean, what is it exactly about finance, for example, that you think there's opportunity for? So I think the way I think about this is from OpenAI's perspective, they have so much surface area to cover in terms of their strategy. It's about where are they going to go first. So today, for example, we might expect them to be making announcements about more generalized tools for others to build applications. I think that that's going to be their focus for a long time rather than diving deep into verticals that require a very specific user experience and very sort of personal focus and relationship.

12:58So finance, health care, dating, home buying, any kind of they're going into commerce, but I wouldn't expect them to be going into the major commercial decisions such as buying a home. um and so basically the more the more discretionary the purchase i guess the the bigger ticket items you know whether it's hey who am i going to spend the rest of my life with what house am i going to live in where am i going to put my retirement savings the bigger the ticket the less the reliance on something like ai i think that's right the bigger and and sort of the less right to win a general platform company like ai like open ai might have um you know there's there's liability associated with these decisions.

13:41You know, they're highly personal. They involve a lot of your personal data. Another example is home management. So companies like Apple and Google got into, you know, home management devices. I don't see open AI going there very soon. And so when you flip this around and say, where are the opportunities for startups? I think AI home management, AI personal financial management, AI matchmaking, AI advisory for the big decisions in your life. And then I also think across the board, startups should not feel cowed by how fast OpenAI is moving and everything that they're offering. There is definitely room across the board in all categories for products that are really well designed and well thought out and deliver a superior user interface.

14:34same question. Right. Let me, I do want to ask you about the developer day that we have coming up, but very quickly, I mean, you make this point here that, you know, OpenAI is sort of a general purpose AI company right now. And that might be the reason why someone who's looking for financial advice on buying a home or managing retirement savings, for example, that might be why OpenAI might not be the best fit for something like that. My question for you is, why is it that you think people will trust AI at all to help them with something like that? because there could be startups focusing on that space, but at the end of the day, it's kind of nice to have a person that's helping you with this stuff, right?

15:11I think it's a great point. And I don't think we're going to transition to people turning over the major decisions in their life to AI at any time soon on a wholesale level. So take Sitch, the AI matchmaker that we invested in. Sitch presents itself as a human matchmaker and there actually are humans in the loop and behind the scenes delivering and supporting the AI and giving you advice. And I think it's going to be that way for a long time. I don't see, for example, people taking medical advice from something that's just AI. There's a doctor in the loop. There's a nurse in the loop. There's a practitioner in the loop.

15:45Same with financial advice. So human in the loop, I think, is going to be the model for a long time. And that also presents an opportunity for startups. What about developer day coming up today for OpenAI? What are you expecting and watching for there? I think this is going to be an exciting one, really focused on the application. layer and tools for developers. So I think the thing that's been teased that I find the most interesting is the visual agent builder. So if you think about the host of horizontal agent builders for the enterprise that help link together MCCP servers with LLM actions in an easy to use visual interface, OpenAI is going to be launching its own agent builder.

16:24So that'll be a great opportunity for developers, but also interesting for the horizontal agent builder companies to think about how they differentiate in that environment. But at the end of the day, it's still a platform. It's a platform that allows developers to build on top of it. So that's the first thing. The second interesting thing, I think, is chatter about an AI browser or some kind of AI browsing space. Not sure what that would look like, but it could be another opportunity for applications that are built on top of OpenAI's LLMs to have another distribution space. And it would, I mean, you know, Perplexity has come out with Comet as well.

17:04And a bunch of other companies have started to march into that space with acquisitions too. So I think now would certainly be the time for OpenAI to come out and say, hey, we've got that too. We're not falling behind in that race. I think it's definitely going to be a day to watch. Anna, thank you so much for come on on the show. We really appreciate it. That is Anna from M13 Partners here on TITV. Okay, for our next segment, Peter Thiel in the past has expressed a ton of caution about investing in AI, but a new story we published today in The Information lays out how his firm, Founders Fund, is doubling down on key investments in the sector.

17:40There are some unexpected pivots in this story too. So I want to bring on our AI and finance reporter, Miles Krupa, to tell us all about what he's learned. Miles, welcome back to the show. It's great to have you. Thanks, Akash. So let's talk Founders Fund. Okay, so the company had its annual general meeting, which you're going to tell us more about in a second. They had it last month, and there were some new revelations about how they are looking to invest in AI. What did we learn? Yeah, we really learned for the first time how Founders Fund is kind of seeing venture investing in AI from the top down.

18:17And what they really said is, we don't plan to make that many investments. But in the ones that they have made already, in companies like OpenAI, Crusoe, Cognition, which makes Devin the coding app, they plan to really double down on these investments and put a lot more capital behind these bets instead of spreading it more widely, as some other VCs have done. So how do you square that with Peter Thiel's comments that, hey, we're in a bubble? I mean, it's like 1999 right now. And here he's saying, no, forget it. I'm going in. I'm going all in, in fact. Yeah, I think he's really saying a variation of his investment thesis that he's stuck to for a few decades now, which is basically he wants to back the companies that are going to become the biggest that are going to basically create monopolies in these new product areas.

19:13You know, he thinks probably that most companies, a lot of the sort of copycat companies that are springing up during the AI boom won't have much value or aren't worth the prices that VCs are paying right now, but that basically there are going to be a few big winners out of this technology wave. And he wants to sort of get behind the ones that can scale with a lot of capital. And tell me about venture capital broadly. I mean, I've heard of strategies where people say, hey, we're indexing the market. You know, we are spraying and playing. I mean, venture capital is a game of having many bets and hoping that one of the hundred work out.

19:53This seems like a nuanced approach, I guess, saying, hey, we are so confident in the big ones. And really, these are also of companies that are pretty big already. The likelihood of them failing tremendously seems pretty low over here. Yeah, I think there's an argument that OpenAI's investors in particular are making, which is basically that OpenAI got this head start. They had this viral moment with ChatGPT, and now they've accumulated all this capital, all this distribution, all these inherent advantages from being a first mover in consumer AI that makes them sort of the default winner at this point.

20:30Whether that's true or not, you know, it'll take time to play out. But I think if you're a VC and your mandate is to invest in private tech companies, you know, OpenAI seems to be right now the clear sort of shoulders, you know, above the rest of the competition winner. Yeah, the blue chip AI company. Did you get any insight from the people you talked to as to why his perspective on this may have changed recently? Because Founders Fund, we've reported they didn't participate in some of the earlier rounds in OpenAI. They were an investor in OpenAI, I think come 2013 was maybe the – I should say 2023 was I think the last big round that they participated in.

21:12And then they actually sat out some of the later rounds. Why have they changed their perspective on this? Why is he coming around to sort of doubling down on OpenAI, on AI generally? Yeah, well, it's interesting. One thing I learned reporting this story was that, yes, Founders Fund may have set out an early OpenAI round or two, but they actually invested in three different rounds before this most recent one, where they put in a billion dollars at a$300 billion valuation. So, you know, one thing that sort of came out at the meeting was that Founders Fund has quietly been accumulating a bigger stake in OpenAI.

21:49And I think they would probably say that just reflects their confidence in the way that OpenAI has been able to grow and sort of grow the business while also being able to consume a lot of capital and use that to create the mode that they're creating around their infrastructure, their model training, the way their consumer product works. So I think that's sort of how founders felt and got more confidence in OpenAI over time. And what about Crusoe and General Matter? And maybe tell us what General Matter is, actually, because that's not a company that we've talked about too much on the show. Yeah, it's interesting because these two companies show how Founders Fund is looking to play a bit farther down the stack, as they say.

22:32You know, Crusoe is a data center builder, and General Matter is in the nuclear, the uranium enrichment business, right? So Founders Fund is really thinking in a way that actually a lot of hedge funds think, which is given the demand for AI, how is that going to play out in sort of the physical world? And where is the demand going to be in the future for these things that are necessary to fuel AI? So Crusoe, they've been doubling down on they're going to invest again at a$10 billion valuation. And General Matter is one that they incubated inside the firm. And so they own a really big chunk of the business, more than a quarter.

23:11So they're really bullish on both. Interestingly, you did mention in the story that they haven't made any investment yet in chip companies. Why not? You know, right now, NVIDIA is just so dominant that it's hard to envision an AI chip startup displacing it in the near term without some sort of really big breakthrough in technology. It's not to say it can't happen, but so far they haven't seen anything that can really compete with NVIDIA. Right. Well, Miles, it's a fascinating story. Thank you for giving us an inside look into Founders Fund. It's rare to get that deeper dive into a fund like that.

23:48So we appreciate you coming on. That is Miles Krupa, our AI and finance reporter here at The Information. Okay, for our next segment, one of the new big crazes in venture capital are roll-ups. It is a consolidation strategy that private equity companies made their name on. And now AI-focused companies are employing a similar approach where they are buying up existing businesses, injecting AI into their operations and trying to make them more efficient. I want to bring on Matt Price, the founder and CEO of Crescendo, an AI customer service company that knows this strategy very well. Matt, welcome to the show.

24:24It's great to have you. It's great to be here, Akash. Okay, so let's talk about the business of roll-ups. This is something that we have hinted at in the show. We haven't gone too much into detail about it. Very quickly there, tell us about what Crescendo does, and then we'll talk about the industry at large. Yeah, so Crescendo's created the first AI native contact center, and the first wave of AI has been adopted somewhat within customer care by bolting on AI chatbots onto existing operations, a bit like you would bolt an engine onto a horse carriage to create the first car. We've thought about it differently and to improve customer experience, reinvented customer care from the bottom up in the AI era.

25:04And it provides amazing accuracy and results. And tell me, I know you have a close relationship with General Catalyst. Was the company started inside General Catalyst? Was GC an early investor? How did that relationship work? It was, yes, it was part of the creation fund within General Catalyst where obviously the business there had an amazing vision for how customer care could be transformed and assembled the team, including myself and Andy Lee, our co-founder, who had an amazing success in the contact center world. And we created Crescendo together. Right. And so tell me, the idea here behind the business is what, that Crescendo will, you just bought a customer service company partner hero.

25:49It was a business, the way I understand it, it was a business that operated call centers for certain companies and industries? That's right. So we have very, we created very deep AI technology, you know, for customer care and ranging from AI assistants and all of the things that are required in order to make an AI native contact center. And then obviously there's an important part of this, which are the people, we call them superhumans because they're trained to work with the AI. And so in order to get to scale very quickly, we then went and acquired a company that was very well-versed with that.

26:25And along with that, they had a number of clients. And the goal there was to work with those clients to move them from where they were today into an AI native world and working very well. And so the idea here is that you will acquire not just Partner Hero, but a second, a third, and a fourth call center company like that. merge them all together, I guess, replace some of the employees with AI to do something like a Decagon does, I guess, and just having an AI talk to the customers? That's the vision here? So the important thing is what we're seeing is actually the humans don't need to replace, but they need to be moved into new roles and educated because the AI can handle certain things.