OpenAI CFO Sarah Friar: IPO, AI Rivalries, New Device, and Spending $100B+ on Compute

0:00OpenAI's CFO, Sarah Friar. We've got to get right to it. You have just completed what I regard as the most successful fundraising round in history. We're going to raise actually north of$120 billion. We think AI is the biggest era that we've seen to date. We're just starting to understand what it's going to mean for global productivity and with that, you know, hopefully more affluence, better lives for everyone. Luck is whatever the preparation meets opportunity, but you've got to grab it. Long-time listener, first-time caller. Quite exciting to get to hang out with all the bros here. Hello.

0:39Jason Calacanis:We weren't sure how to start this off, but I thought the best thing was to allow our erstwhile crypto czar to maybe save comments. I saw an article today, I think it might have been in the Wall Street Journal, that the perception is that there's an advantage to IPO-ing earlier if you're an AI company. So now we know SpaceX is going. And then the question is, when are open AI and Anthropic going to go? And I'm curious, how do you think about that? Do you think there is a little bit of a race on? Or, you know, you haven't made a decision about that yet. Like in the end, an IPO, I say this to the team all the time, it's a milestone.

1:17It is not a destination. Do not run your company as if that's some sort of destination. It's just another way to fundraise. We just did, you heard me on the sizzle reel, raise$122 billion in March. And that was to give ourselves maximum flexibility. I feel like my job as a CFO is create optionality for this, not just this company, but just this era that we're living in.

1:45Jason Calacanis:Sorry, Sarah, was that point in fundraising, is that the biggest private republic up until the SpaceX IPO? It is. It is by orders of magnitude. I think the largest IPO to date was the Satya Ramko, which was about$30 billion. So it is actually incredible that you're going to have potentially three IPOs at a scale that will be bigger even than 2001. That time frame, there was a lot that went on in the market, too. But the market has grown. And by the way, the other thing going on in the market is, like, if you look at buybacks, M &A, and so on, it's actually a lot of capital that keeps being returned back to shareholders and cash.

2:21so there is a lot of money sitting on the sidelines but in the spirit of like the question david i think in the end you want to you'll be measured right it's the in the end the market is a weighing machine not a popularity machine no one remembers who went first google or yahoo lyft or uber and i say that not because whether i want to be first or second but i just think it you know the the press loves a bit of drama but in the end we're going to have to build big sustainable, durable companies. And fundraising will be a key component of doing exactly that. Sarah, breaking news. Oh my God, so many people coming at me.

2:57Hi, Jason. I know, it is hard balancing four interviewers at the same time. It's okay. This is my world, by the way, so I'm good with this. Jason. Anthropic just confidentially filed their S1. So does that mean you're third place in terms of the filing? It does not mean anything yet because you have to run now the gauntlet of the SEC and who knows how long that takes for anyone. Yeah. Is there though a benefit to them going forward? And I think unpacking the rivalry with Anthropic is on everybody's minds. So just, I guess you can't talk too much about IPOs. So I'll just pivot to Anthropic was far behind and now they've really, I think everybody would agree in the industry now, blown past OpenAI in terms of developers and corporations, and it seems revenue.

3:49So how did that happen at OpenAI when you had such a tremendous lead? How did Anthropic blow past you guys? So let's talk a little bit about a strategy. Our strategy is different, right? So we are building the AI layer, the infrastructure, and it's really important that there's a single foundation, but then with many interfaces out into the world. So chat GPT is one to the consumer. Over 900 million people use chat GPT weekly and it's become the noun and the verb. It's how most people experience AI for the first time. Kind of fun fact, our economic research team just showed me the fastest growing continents now are Africa, probably not totally surprising since it started a small base.

4:33fastest growing languages are Azerbaijani and Kazakstani. It's Kazakh, which is kind of incredible to talk about where it's going. So multiple interfaces, ChatGPT. Of course, there's Codex. Just hit 5 million over the weekend. We're really proud of that. Coming from almost zero in January. A million users. Go Codex. Help me prepare for this little special up here, too. There's, of course, Frontier, our enterprise offering. and every other way that we can get out there to reach businesses of all sizes. That is a very different strategy. We think that because it's served up on one model, there's a compounding element of advantage that comes from that.

5:17More users, more data, more ability to personalize. ChatGPT asks as a front door. As models get bigger, there's more efficiency. That should lower the overall cost to give you a token in the world. That should compound to higher gross margins, ultimately more ways to pay for compute. And then access to compute is one of the really big competitive advantages at the moment. So we have to all run our own races, but we all have to recognize we're part of an ecosystem that also needs to bring people along collectively. Did you spread a little bit to then too many projects? People were talking about this new gadget, Sora, and then maybe not enough focus on enterprise.

5:56Is that a fair assessment of if there was a mistake in the last year, that was it? No, I think that the world loves to go to binary-isms. Like, are you a consumer company, Sarah? Are you an enterprise company? The reality is we're very much both. We're not one or the other. Right now, our revenue is getting pretty balanced, about 50-50. We are incredibly focused on the enterprise. I spend so much of my time with, I mean, just even in the last week, I could tell you I've been to see Thermo Fisher in Boston. I was with a bunch of banks in New York. I was on the phone with travelers on Friday. I spent this morning on the phone with a tech company.

6:34It doesn't matter the vertical. People are really moving on AI right now. Our new head of revenue, Denise Dresser, in seat since December. She is a force of nature. And so I think the enterprise, broadly speaking, is really firing on all cylinders. But we don't want to leave the consumer behind. Remember, our mission at OpenAI is AGI for the benefit of humanity. Not for the benefit of humanity who can pay or for the benefit of humanity who live in an enterprise. but very broad-based. It's why we offer so much free, because we want people to get a taste. Once they get a taste of intelligence, the ability to come up a commitment curve is incredible.

7:13Our free users do about seven turns, seven questions a day. Our first paid tier do double that, about 15. Our real paid tier, the plus, 20 bucks, hopefully you're all on it or higher, about 3x and Pro about 11x over a free user. So remember when you got your flip phone and you're like, yeah, I don't know, it does make some calls. Now that same phone, think of all the things it does for you. That's the path we're on with intelligence right now. Sorry.

7:42Jason Calacanis:You said something very influential. I think it was about 18 months ago for a lot of us in the industry where you framed a very simple economic trade-off, which was gigawatts to cash. And I think you said one gigawatt is roughly equivalent to about$10 billion a year of revenue to open AI. So comment number one was this one gigawatt equals$10 billion a year of revenue for you. But it's not just you, because you can probably extrapolate that to Anthropic and other folks, Gemini. But then you were really at the forefront of getting access to power and data centers and powered land. It seemed a little crazy, but now it looks like, hold on, there's a huge deficit of supply.

8:22Jason Calacanis:Can you just unpack all of that and explain both the spectrum of where we are and then those specific economics and if that's changed? So first of all, yes, compute is a very scarce resource at the moment. I mean, what we see in our business, we're going up that kind of vertical wall of demand right now, and there's just not enough tokens available. So I'm very grateful that I got to work alongside Greg and Sam. I think very prescient on this. And last year, we were definitely taking some, you know, arrows in the back about why are they out there buying all this compute? And I think, thank God we did, because in 26, we still won't have enough compute.

9:00Where are we on the compute continuum? There's kind of choke points everywhere. And I think they will continue to move back and forth. I mean, you all talk about this and know this as well as anyone here. Whether it's energy, first and foremost, Land power, how we get regulatory environments such that we can build quickly. When you get into the racks and chips themselves, clearly do we have enough in that supply chain? Memory spike is on at the moment. Access to great talent. Do we have enough people coming through our education system? I really worry about this right now. I'm a trustee at Stanford, and I see that we need to keep the focus on education and science.

9:44And then trust. I mean, I actually put that as part of the supply chain. Sam right now is in Saline, Michigan. He's going to be cutting the ribbon in about two hours. So you are getting a sneak preview. But they told me it was okay to say it in the room. That will be, you know, sticking shovels in the ground on a one gigawatt data center, which is part of our Oracle complex. Really important there on the trust side that we don't leave communities behind. I spent seven years of my life working at Nextdoor, doing the hard work of what it means to be local. And you cannot tell people from top down what they need because they will tell you thank you, but no thank you.

10:21I will tell you what I need. And so in a data center like that, we're actually spending a lot of time in the community saying, number one, we're not going to raise your electricity bills. We're going to pay for our infrastructure and our power. It will not be the rate payer that has to pay. Number two, we're going to bring jobs, 2 ,500 union jobs, good jobs, like electricians, HVAC, and so on. We are going to pay our taxes, a billion dollars in taxes just for that data center into Michigan. And on top of that, we're going to invest$45 million going into education for Codex credits to do what you all talked about this weekend.

10:58It's like anyone who's not coming in facile to their new job. I have teenagers using Codex. It would be like I would never hire a finance person who didn't know how to use Excel. And I pretty much probably wouldn't hire a finance person today that doesn't know how to use a tool like Codex. So when I think about investment, we're having to invest ahead of demand. That means we need to both be able to find all of the compute and all the pieces and then pay for it. So that goes back to your capital question on IPO. And then on the other side on the economics, look, the economics do continue to get better.

11:31They're getting better on multiple fronts. I think we are doing a better job of actually showing true value to our customers. And I think you get beyond kind of a cost plus type pricing into something that feels more akin to the value being created. Now, scarcity of tokens helps because it's causing a bit of a compression in cost.

11:49Jason Calacanis:Talk about that and just like without specific names, where you know the landscape exists today in terms of all the power that's available and all the demand that exists across everybody. Yep. What's going to happen over the next year, just at the current course and speed of what is available, of the data centers that's available, of the tokens that's available, of the infrastructure that's available, for everybody. Because, you know, I told this story last week, but, you know, I'll use Anthropic, and one of the frustrating things is at some point it just says, you know, 10.30, it's like, all right, Trimov, see you at 2.30.

12:22Jason Calacanis:Yeah. And that's not a viable experience. Right. And in fairness to ChatGPT, actually, I've never had that with. Yeah, we're quite generous with our tokens. And again, on purpose, we're trying to drive access so people understand. Because if you're on that free tier, not actually getting the latest model, but we're trying to put it in your hands so you get a sense for it, by the way. Because if you're a kid doing homework, like I think about when I grew up and the Encyclopedia Britannica showed up at the front door in Northern Ireland in a tiny little community in the middle of the Troubles. It was like the clouds parted.

12:58And so we want to make sure that people get that feeling, by the way. But the landscape right now, in 26, if you want to buy more compute, good luck to you. Like, tell me, because I don't know where else to find it. I mean, as you know, Elon has some. Well, I was going to say, Elon, ironically, ended up being the one person that had too much compute in a way. But good job on, like, figuring out how to sell that off. In 27, it's pretty limited as well, frankly. Now, there's a couple of things shifting around. when we talk about compute, there's training that mostly still all happens here in the United States for USG reasons, for making sure that a national asset in effect is happening on US soil.

13:41For inference, we want that to be global, and I think particularly in an agentic world, you want much more kind of real-time, even for things like Sora and video, which by the way, yeah, we have, you know, we had to make a really tough choice because we didn't have enough compute. And it uses a lot. Right now, yeah, video does. But video is not over. Like, in particular, when you start to think about where AI is taking us into more multimodality. So remember, we've all been taught by the last generation of technology to talk with our thumbs. It's a disease. You walk around. Everyone's looking down.

14:16They don't look up anymore. Teenagers sit on my sofa at night and talk to each other with their thumbs. I'm like, who are you talking to? And my son will be like, him. I'm like, okay, talk. Multimodality is here. Hopefully, I think you all talked about it this weekend. You're talking to your tool. I talk to Codex every day. And so that is changing rapidly. But that is going to need much more kind of real-time compute because it's an odd experience if I was talking to Chamath. And you're building with Johnny Ive this puck in these earpieces. So maybe tell us a little bit about that process. You've admitted it now.

14:48If I tell you it's a earpiece, Johnny will come and steal my teenage son. I might give it to him, give him to him. You do believe that there should be some... We're changing into a consumer substrate that I cannot tell you what it is, but by the end of this year, we will unveil it. But you've seen it and you've tried it. I have seen it. I've tried it. I am a hand talker. Right now I'm sitting on my hands. Is it paradigm shift? Did you have an iPhone moment? Yeah, when you used it, was it like having an iPhone for the first time? It's very... What Johnny and team are really good at is bringing humanity to devices.

15:24And I don't really know how to explain that well, but when you see it, you feel it. It feels natural in some way? It feels very natural, but it feels very lovable. Really? And I can't really explain what that emotion is. It's intimate in some way in terms of not taking your phone out. And it's seamless is what I've heard from people who played with it. Technology can be very mechanistic, but we all know great design just makes everything fade away. It's what, at the time, the simple is hard job set. This story, just going back to the earlier question, so putting on the CFO hat, help us understand the capital allocation model that you use.

16:07Because a lot of businesses over the last decade, two decades, that have kind of been these outsized returners have found some unique way to deploy capital at a higher ROC than anyone else. And then you end up plowing all your capital into that higher ROC bucket. What is that for you guys? And how do you think about that portfolio approach to having more of these kind of big returner shots? And is there an engine where that gets better over time? There has to be because in the end, the durable, high-value companies created in this era, I don't think they're not going to be magical. They're going to look like the great companies of prior eras.

16:44They're going to create customer value. It starts with the customer and really helps the customer do something different, better, more revenue, more efficiency, right? Thermo Fisher wants to be able to get patient screening done faster so they get FDA approval faster. That's really important. Like if you have a form of cancer where you have weeks to live, the difference between a breakthrough in four weeks and two weeks can literally be life or death. They also have, I'm going to misquote this, but something like 30 ,000, 38 ,000 people in the field selling those amazing, like if you walk into any lab in the country, you'll just see Thermo Fisher plastered all over every device.

17:26Those people want to be more efficient going to work. Like the fastest takeoff of codecs within OpenAI right now is actually in our go-to-market team. Our devs are there, but like if you look at the pace of growth kind of month over month, it's all in GTM. So they want more productivity out of their GTM team. And of course, they're doing things in areas like finance, which I get really excited about. So customer value first. From that, now you need to get to a great gross margin. So how do you get to a great gross margin? You're looking at like the cost of revenue. The main input is compute. The good news on compute is that there is a massive deflationary curve on cost, right?

18:03From ChatGPT 5 to 5.4, I think the deprecation cost was something like 97%. It's kind of an amazing curve. Actually, I'm slightly. From 4 to 5.4, it was 97%. But that happened in like two years. That's kind of wowing, right? That's incredible. Even our newest model, if you look at 5.5 that we just released, we're trying to now translate that back to the customer. So we actually raise prices on 5.5 2x. but if you look at what the cost of the customer is, they're probably still getting a break of about 20 to 30 % cost reduction per token because it's just much more efficient per token. So there's a lot to do in that envelope.

18:44And part of making a capital allocation decision is having to, if you make it on today's cost profile, you actually might misprice the outcomes. You have to lean in a little on the cost profile. And then as we think about the bills, yeah, you are having to make, Like really, my focus today on compute is what's the compute I can buy for 28 onwards? Like that Michigan data center in Saline, I don't think we will be getting compute out of it until probably end of 27, early 28. So that's where you're starting to make your bets. And in fact, where I feel most short of compute right now is starting to look at 30, 31, 32.

19:24So you're having to create a business model. Now, the good news is each year goes by, we get more confidence in the build. We're seeing it massively outperform. And so that's giving us more and more confidence. And the market is coming towards us much more. So how are you making the compute need forecast multiple years out, accounting for all of the architectural and model advancements that are happening, where call it value or utility per unit of power is going up? and help us understand how you kind of estimate that given that there's a lot of technology development going on that has a high kind of variance to it.

20:06Yeah, yeah. So we do have to make multiple assumptions, both on the compute itself. So we assume right now that compute actually on a per gigawatt is getting more expensive because power is getting more expensive, memory is getting more expensive, and so on. However, the intelligence that we get on the other side out because of the deprecation on the chip side is more than making up for that. So in terms of a per unit sold to a customer, it should actually get a lot less expensive for the customer. No model improvement in that. So that's just chip. Exactly. That's just the chip itself. We don't try to overestimate on the model side because sometimes, like 5.5 is an incredibly good model on the efficiency side.

20:46But if you look at something like 5.4, the prior model, it was a really large pre-trained model. It was very expensive. It was actually hard to serve. And sometimes we want to do that really big pre-trained moment. And then we take multiple model turns to be able to kind of drive down on the cost side. I mean, in the near term, like in 26 and 27, I clearly build a model that's bottoms up. So I know what my products are. I have a sense of what the pricing will be. You know, consumer, P times Q. How many wows do I think I have? I can see what the shape of the line is. how many of them will subscribe.

21:23Advertising coming in is also still related to how many weekly actives, how many dailies, how many messages, and so on. So you can do actually a pretty good model job in 26 and 27. That said, the shape of the line keeps taking us by surprise to the upside. When you get into the outer years, you're actually looking more at the compute you've bought and almost just doing an algorithm the other way that's saying this amount of compute should equate somewhat to this amount of revenue. I don't know for certain exactly where it will all come from. Like a year ago, I built a model for investors that showed agentic revenue.

21:58And the story was, we're going to have this thing, we're going to be in the agentic era, we're going to hand it to a developer with natural language, they're going to be able to build. And we think they will pay upwards of maybe$2 ,000 a month for it, which is kind of laughable in hindsight, but nobody believed. They were like, I don't even know what she's talking about. There's no way that will happen. And$2 ,000 a month? Remember when people were losing their minds over ChatGPT Pro being at$200? Like, oh my God, no one will ever pay for that. So why$122 billion? Does it take you to 2031, 2032?

22:34Like, how do you get the calculus on the capital needs as you do that modeling? Maybe even more specific. So the estimates I've seen is that to stand up one gigawatt of AI compute costs about$50 billion. Land, PowerShell, chips, everything, all in around$50 billion. Do you have to front all of that money when you create a new data center? Or how much of it do you do? How much of it can you get debt for? Does 100 billion raise only get you two gigawatts, or does it get you five? What does it get you? It's a great question. So if you look at our compute strategy, and it's crazy how fast the world has changed.

23:15So just two years ago, we were literally one. We had one CSP we worked with, Microsoft, Azure. We sat on one chip, NVIDIA. We had one product, ChatGPT, one price point,$20 a month. So I often use a Rubik's Cube as kind of my metaphor. So we were like one cube in the bottom. Today, if you look at our strategy, it's been to go, first of all, multiple CSPs. Because what CSPs do for us, in effect, is they shift CapEx into OpEx. So you pay as you get the revenue, so as you're actually utilizing the data centers. So in effect, we are writing somewhat on their ability to build CapEx and financing. So today we sit on top of every CSP, Oracle, CoreWeave, Microsoft, GCP, AWS, and a bunch of small neoscalers.

24:03On the chip side, we've also gone for a program of being multi-chip, because we want to make sure you're always on the frontier. I think if you're only on one chip, there's just inherently a moment where you can't be on the frontier, because there's some leapfrogging that happens. So today, NVIDIA remains our absolute priority partner. They have the frontier chip. Our next big trading run in the fall will be done on Vera Rubens. We're really excited about that. And now we're plotting kind of the Feynman series that's coming. But we also now have chips in the pipeline from AMD. Cerebris is already online.

24:36It's been an incredible low latency chip. Great for devs, for example, that want real-time coding. And there's our own chip that we're working on with Broadcom. And then beyond that, there's other ways we've diversified. So now think about that Rubik's Cube. It's become much more multidimensional. and it allows us to effectively utilize investment-grade CSPs in order to be able to go fast and push it back to be more OpEx, not CapEx. Now, we are starting to shift gears into more of a built-to-suit type environment. We announced a data center we're building with SoftBank Energy down in Texas. That's the beginning of something that's beyond a CSP.