← Big Technology Podcast
Big Technology Podcast
Is OpenAI Ready To IPO?, The Datacenters in Space Myth, The Kids Boo AI
00:59:49
Loading summary
Transcript157 lines
- [00:00]
A
Is OpenAI ready to IPO? How about anthropic? SpaceX is going public, but data centers in space might be a myth and Eric Schmidt is booed loudly by the new grads. And maybe they have a point. That's coming up on a Big Technology Podcast Friday edition right after this. I'm just back from ServiceNow's knowledge 2026 in Las Vegas and the conversations I had there are ones you're going to want to hear. I sat down with their President and CPO Amit Zaveri on the platform strategy powering enterprise AI Chief People and AI Enablement Officer Jackie Canney and Chief Digital Officer Kelly romac on what AI really means for the workforce, the technical leaders behind ServiceNow's Nvidia partnership on shipping AI at scale and Ulta Beauty on deploying ServiceNow's technology across 1300 stores. If you want to know where enterprise AI is actually headed, not the hype but the real story, you can find these videos on my YouTube channel. Search Alex Cantroots on YouTube. Depending on who you ask, between 80 and 95% of enterprise AI projects fail to get AI to work for you. You don't need more tokens, you need better people. A board pair, powerful proprietary tools with senior engineers who've seen it all. That combination means your project doesn't stall, doesn't drift and doesn't fall. It ships. Whether you're a startup that needs to get to market or an enterprise with complex legacy challenges, Aboard delivers exactly what your business needs Fast. Aboard is your partner for AI transformation. Visit aboard.com and let's build something together. Welcome to Big Technology Podcast Friday Edition where we break down the news in our traditional cool headed and nuanced format. We have so much to cover for you today on a very big Newsweek. We're going to talk about the new numbers that we're seeing from OpenAI and Anthropic, as both seem to be ready to file to go public. SpaceX, of course, filed to go public this week. We'll take a look at the numbers and debate whether data centers in space are going to be a thing. And then of course, it's commencement season and what does that mean? Booing AI loudly? And we'll talk about what it means for the AI industry as public approval is as low as it's ever been. Joining us as always on Fridays is Ranjan Roy of Margins. Ranjan, great to see you.
- [02:08]
B
Eric Schmidt's getting booed and Elon Musk has found the largest addressable market in human history. I Think we got a. We got a lot to talk about this week.
- [02:18]
A
So much to discuss. Yeah, the Schmidt boos, I mean, weren't the only boos about AI. We had three commencements so far where the speakers have mentioned AI and then actually seemed somewhat baffled by people booing them. So we'll get into that at the end of the show. But most importantly, I think we'll start talking about the numbers that we're seeing from OpenAI and Anthropic as both gear up towards filing for IPOs. Maybe even this year. We could potentially even see the paperwork for OpenAI come through as you and I record today, Ranjan. So that might come through, but we already have some insight into how the business is doing and I think it's worth us discussing a how the business is doing and also whether these are companies that should go public at all. All right, this from the information. OpenAI generated nearly 6 billion in revenue in the first quarter, boosted by Codex. OpenAI generated about 5.7 billion in revenue in Q1, nearly 1 billion more than Arch rival Anthropic generated in the same period. But it's not all roses. OpenAI told investors that its adjusted operating income margin was negative 122% in the first quarter. That means for every dollar of revenue the company generated, it lost $1.22, even after excluding some large items such as stock based compensation, according to one of the sources here, Ron John, take us to the Wall street side of things. You've spent time on trading desks. Is that bad? Is that bad?
- [03:47]
B
I honestly think I was watching on cnbc they had like a chiron, like something like open. Is OpenAI the new WeWork? I think, I mean, you know, like the margins that were actually the operating income margin that was released was not that surprising to me. We all know they're losing money. Like, we know that their costs are exceeding their revenues. I guess, like the fact that it's at the operating income level, you know, it's not good. But were you surprised by this kind of number?
- [04:21]
A
So actually I want you to go a level deeper. So the, the surprising thing to some was that it was operating income, right? So that's after you take away the things that typically make companies unprofitable, they were still unprofitable. So talk a little bit about the difference between operating income and what we would otherwise see and whether that is a big deal. I'm actually hoping to learn from you on this one.
- [04:44]
B
Well, no, operating income still would include all the traditional things that go into like what are the costs of goods sold, any kind of R and D, even like SG and A, like all the additional costs, people salaries that still will go into the operating income line. So it's not just like at that level, it's not that shocking that they're losing Overall, you know, $1.22 off of every dollar it's generated. Again, it's not good. It's not good at all. But I think like from they, they've kind of outlined this business for a long time. What is really interesting, we're going to get into the anthropic numbers as well that the Journal had reported on. It's so fascinating to me that we've all known for years now that they're burning money, but it's now going to actually just be directly provided to us how much they're burning and how, what their business model actually looks like. Even though we've all kind of known this. It's funny, it's almost comical to me that like now we get a number like this and people are actually kind of not freaking out, but they're certainly going to be questioning it a little bit more. And I do think again, like as in the reporting, stock based compensation was excluded in this, but that's not that uncommon. Obviously like when you think about the economics of the business, it's definitely going to be an SBC heavy like profile in terms of all of their employees. But overall I was not shocked by these numbers.
- [06:26]
A
Okay. And so I titled this episode at the beginning of the episode title is is OpenAI Ready to Go public? And so the question is, why is it going public with these numbers? Right. The numbers look bad. And I think that the reason why, like typically what you'd want to do if you're a startup, and correct me if I'm wrong, but you want to get your economics in shape, you want to grow and then when you start to turn profitable and you can show a path to lots of profitability, which is what stockholders want. As soon as you can draw that line, that's when you hit the public markets. So why is OpenAI deciding that it's going to go public now?
- [07:05]
B
Alex, come on, come on.
- [07:08]
A
Is I too naive?
- [07:09]
B
Yeah, like they can't, they just can't. I mean, they can't what get. When you're saying like kind of get your financials in order, like it's not a profitable business. We've all known this, like it's not. And is anthropic profitable, we'll get into in A moment. But like they have been telegraphing this forever. I think they can't just turn. I guess the biggest difference is that theory kind of held where there was the Uber which again it took them like six years after IPOing to actually become truly profitable at like a unit economics level. But like the thesis had been you spend on sales and marketing to grow, but once you have those users and they're sticky, you can just kind of like turn off the sales and marketing spigot and then suddenly you're profitable and that's what you do ahead of an ipo. But as we've said, the economics of a gener or an AI company are just fundamentally different than software that like it's more akin to industrial companies or something because you're paying for the compute costs so the resources in can scale linearly or somewhat linearly with your actual revenue. So I think that's the biggest difference that like the they no one has figured out the business model and it's this week actually I think was a pretty big one because it became clear.
- [08:38]
A
Okay, so this is going to be the punchline of my question which is if OpenAI is clearly not ready, then why IPO now? I think your answer was premised on this idea that like, all right, well it's natural to IPO now. They just raised 122 billion. Okay, so why IPO now? And that's where, let me give, let me give my perspective and then you can sort of rebut me here.
- [09:04]
B
Got it.
- [09:04]
A
I think it is entirely a result of Anthropic's success. So you have a couple of curves that are running in parallel right now. So you have OpenAI which made more money than Anthropic. But Anthropic's revenue and user growth it seems like has also led. Revenue in particular has accelerated faster. We have this Wall Street Journal story telling us mind blowing growth is about to propel Anthropic into its first profitable quarter. Right. So they're going to make 10.9 billion in the second quarter. And the Journal says it's an explosive rate of growth that will help it turn an operating profit for the first time. So OpenAI is the leader, but Anthropic is gaining quickly. Right. And I think what OpenAI wants to avoid is, is Anthropic going public with its rapid rate of growth. That by the way doesn't seem like it's sustainable. And that's part of this profitability thing that Anthropic is profitable now, but it would prefer to be unprofitable because it would be prefer to be building to meet that demand. Remember a couple weeks ago we heard from Dario Amande who said, you know, we demand grew 80x and we really wouldn't like some more normal growth. OpenAI has been building in a way that Anthropic has said is reckless. Dario called it yoloing. And they're going to be able to meet that demand. So when is the window to go public? The window to go public is before Anthropic. Because Anthropic, if Anthropic goes before you and that story hasn't played out entirely, it's going to look like they are growing faster. It's going to be much more difficult to tell your story to Wall street that oh yeah, they're leading us now, but because of our bets on infrastructure, we'll be able to lap them. You'd much rather just go before him for them and set the narrative or otherwise you might have to wait three or four years to IPO. And I don't know if OpenAI has that long. That's my theory.
- [10:47]
B
I agree. I actually do. I think it was in our Discord chat for some one of our listeners that actually made the comment that you need to view everything these companies do as who can get out to market first. Like that's it. And I would say the last two months, that's everything I read through that lens. Space and SpaceX has already filed their S1 and oh my God, we can get into those numbers this week.
- [11:15]
A
We will get into those numbers.
- [11:17]
B
Oh my God, those numbers are something else. But so they have already filed RS S1 and have positioned themselves squarely in the AI camp. Like this is not a space company anymore, this is an AI company. So they're already getting out there and Elon Musk is Elon Musking in ways that we can only just sit back and marvel at, I think. So between OpenAI and Anthropic, the craziest part of this whole thing to me is like going public is not. It feels like this end state for these two companies, but it's supposed to just be a financing event that gets you going to the next level. They don't need two IPO in theory, when you can raise $120 billion, as we said, I think it's five or six times bigger than the biggest capital raise in an IPO ever. They're OpenAI's less fundamental.
- [12:14]
A
Yeah, about four times bigger.
- [12:16]
B
Yeah. So like I agree it's still always been baffling to me is the goal just to provide liquidity to all investors onto retail. I mean, maybe that's it. But, but I think, but, but I do agree, they, it does feel like this almost desperate race to get out first. And I mean, again, the cynical read on that is just to reward all investors and employees by pushing the stock of these companies onto retail in this weird, desperate way that makes it feel like they don't actually feel they're worth what they are.
- [12:53]
A
Yeah, exactly. I mean, if you really believed in the exponentials, would you be doing that? But again, think about Rush.
- [12:59]
B
It's not a rush then.
- [13:01]
A
But this is the thing. If you, here's the argument. If you believe in the exponentials, you're going to believe you're going to need more money to build infrastructure, right? Yes, it's just, you know, a one time financing event. But you could get a lot of money from this ipo and you need to be smart about where you get the money. Right. And so if you believed in the exponential, you would want to take the biggest possible funding round. You can you look around, right. Again, we've talked about this on the show. The Gulf states might not be into funding OpenAI and anthropic right now. Remember Dario said, we're going to go back on some of our values and take Gulf State money. What happened to that, that round? I mean, the big Gulf State round never materialized. Maybe some came in in some ways. And so then you have to go public. Then you have to go public. And when you, if you do that, it's important to get the most money you can, the best valuation you can, because after that you don't really have, you know, another IPO that you can do. And so what, what OpenAI is doing, I'm pretty sure is they're looking around and they're like, now or three years from now, we might not have three years from now. We need that money. We need to be on the market. We can't let Anthropic go first, set the narrative and then have to like sheepishly follow them with our stats not looking as good because this rotation into our products because we have the infrastructure hasn't fully happened. And that's to me, what's going on right here.
- [14:19]
B
Okay, I had forgotten your theory, which actually makes this all make sense to me, that the Gulf states are not going to be available to be that final, final tranche of financing or the next few rounds even, especially with the Iran war. Yeah, yeah.
- [14:35]
A
Have you seen what's happened to Qatar I mean they're, they're stuck. Their natural gas is on hold because of the straight of hormones.
- [14:44]
B
No, no, you're, you're right, you're right. So I mean it is crazy to me that is that singularly causing this kind of like almost panic. And I think you could be right because it's almost again like SpaceX is targeting a raise between 40 and 80 billion dollars. That sounds big. That's like a regular Wednesday for private market financing for OpenAI or any of these companies. Like it's gotten so stupid. So, but you're right, that, that actually, that makes things make more sense now.
- [15:20]
A
Anthropic is in a really interesting position again and I want to go back to this profitability here. Profitable also means like you know, you're, you're making more money than you're spending on the build out, which like if you're trying to hit that Goldilocks zone of building out enough to keep growing without mortgaging the company or without going bankruptcy, you actually underbuilt if you're profitable. That's the logic. And so I'm curious, Ranjan, what you think about Anthropic in this situation?
- [15:47]
B
Okay, so the Wall Street Journal reporting again, like the numbers, you start to see 4.8 billion in sales in Q1. That's supposed to more than double to $10.9 billion in the second quarter. Now we are only about halfway through the second quarter, so already we're in extrapolation land. There's been a lot of growth, but we all know Codex Show 5.5 showed up from OpenAI recently. I've seen endless chatter around people actually switching over from Anthropic. So the idea that it's going to be a straight line or not even straight line, I'm guessing it looks more like an exponential curve to 10.9 billion I think is ambitious when we're only halfway through the quarter. Again, in any other situation where you're looking to go public, they would be providing backward looking numbers not predictive of the next quarter and a half. But let's get into that profit. They say they're going to turn an operating profit of 559 million in the quarter ending at the end of June. So I got to give Ed Zitron credit on this one. I feel like I go back and forth sometimes. He's certainly a little too bearish and negative on the industry that I do, I work in myself. But like this is where he's great. So in this SpaceX S1 document, they talk about how Anthropic has agreed and this is one of the big like revenue boosters for them to spend $1.25 billion per month through May 2029. But they explicitly say in May and June 2026 it is at a reduced fee. We don't know what that reduced fee is, but that reduced fee could be zero. And this is honestly like the fact that they specified May and June 2026 are the reduced fee. Suddenly the Wall Street Journal gets this leaked investor prospectus saying that that Anthropic is gonna be profitable for this quarter ending and then even adding the kind of like this does not guarantee profitability in future quarters. This, I don't wanna go full conspiracy theory, but like if you just think anthropic going to SpaceX. Hey guys, you guys are gonna IPO? How about we commit to I think 15 or no, no, like $28 billion in spending for your business and be the first gigantic customer. That's a long term revenue. All we need from you is you cut us a deal on the next two months, give it to us for significantly discounted or free to make our numbers look good and everybody's happy. And that's what it looks like to me.
- [18:44]
A
But okay, this is all predicated on the idea that Anthropic wants to show profitability. Why would they want to show profitability? That means that they miscalculated their build out.
- [18:55]
B
No, no, it's Wall street. It's what you will. No, it, you will still appreciate profitability. You appreciate control over your business. Like I still think the, they missed the build out narrative. It's going to, it's not going to be the dominant one going into, as these companies go public, the, the losing money again. The fact that that operating income margin actually surprised people is just a reminder that people have kind of been closing their eyes. Even though we all kind of know they're losing a lot of money, it's going to move back to that. And Anthropic knows that. And that's why they very smartly and tactfully were able to not only come up with these numbers, but leak them to the Wall Street Journal and get the reporting.
- [19:44]
A
I don't buy it. I mean I, I definitely accept it as a theory, but it seems to
- [19:50]
B
me that why May and June?
- [19:53]
A
That is a good question. Okay, I don't know the answer to that.
- [19:57]
B
Why reduced fee for May and June only?
- [20:00]
A
So your, your contention is basically that this was a ploy to leak some numbers to say OpenAI is yoloing. Our revenue is accelerating. We are going to be profitable. So buy our IPO that we haven't announced yet.
- [20:18]
B
Not, not only to leak numbers to go to the market with those numbers there. If they file their S1 sometime going into June 30, suddenly you are going in as a profitable company. It changes. And as we said, get like, if they go out with that narrative, OpenAI can be shouting from the background, but they didn't get their compute right. They're not investing and everyone will look at them. How much money are you losing? It's, it's. I mean, it's, it's kind of brilliant.
- [20:51]
A
But I do think, yeah, all right, maybe it is this boss move to just be like, listen, like they are. You know, they, they might tell the world that they're going to have this infrastructure advantage on us. They're losing a dollar 22 cents every dollar they spend. We've shown an ability to be profitable. Go with us.
- [21:09]
B
The market will love that over the. The alternative.
- [21:15]
A
Like I told you, that's the way that it should be done. Except if you're making tricks to do it. So let me just ask.
- [21:21]
B
Oh, yeah, you're right. You're right.
- [21:22]
A
That is the dream lives, except maybe not exactly in this format. Doesn't that catch up to the company? Like, let's say they do go public and, you know, they sort of had this sort of sweetheart deal. You're gonna eventually, like, think about if you're in Dario's perspective, you're gonna go public, you're gonna get that money. Do you want to be the CEO of a company that's just gonna have that on your back? Being like, they never got back to that point.
- [21:47]
B
That. That's what I genuinely is still surprising to me. I mean, I think the drying up of Middle Eastern money in Gulf State sovereign wealth is like, possibly the impetus for it. But I agree, like, it's still like the feeling of urgency when they've been raising this much money, when it's not going to be in aggregate that much more money, like in pure additional financing for any of these companies. But it will be a shifting of who owns all of that to public market investors, which is a much larger capital pool. But, like, why that feeling of like, I mean, not desperation, but just that race to IPO still bad. It's still weird. It still feels weird to me.
- [22:39]
A
So I, I think the question needs to be asked and I don't think that this is the case, but we should at least talk it through. Maybe they're seeing a top and, you know, I think that's kind of had like, heretical to say, but, like, let's at least explore it. So where could the top be? We talked about OpenAI making this pivot to Codex, and we asked if it was coming from a position of strength. And you could say, well, maybe not. If the consumer ChatGPT doesn't have much further to grow. This is from the information article. OpenAI faces another challenge growing ChatGPT usage. OpenAI's weekly active users for the quarter averaged about 905 million users. Although the company hit about 920 million weekly active users in February, the lower average figure for the quarter implies that usage during the rest of the quarter was weaker than the February snapshot. The company expected to hit a billion weekly active users by the end of last year, but warned employees to expect rough vibes as consumer AI rival Google made gains with its Gemini Chatbot. Ooh, I don't know. That's. That's potentially rough for OpenAI. If it just can't grow ChatGPT, it's seen some contraction. What do you think?
- [23:53]
B
I mean, I think it is. It is and it isn't. Because isn't this the company that is now going to focus maniacally on the enterprise, which, if that's the case, then I mean, who cares about your 920 million active users or 905 or whatever it is? Like, and again, is the entire game going to be Codex? Like, I don't. Do you think, do you think they're going to still go. Do you think they go into IPO with a dominant story still being consumer plus enterprise, even after they told us it's going to be enterprise?
- [24:28]
A
I think it's going to be enterprise. And I think again, and this is something I've brought up in all of my interviews recently, is, you know, a lot of the growth of generative AI has been on the back of, like, novel uses that's gotten people interested in the technology and sort of established a baseline behavior, but hasn't exactly become the norm. Like, I think we're still searching for, like, what is the key use of generative AI? And remember where we've seen these spikes, image generation, voice generation, and now this agentic stuff. The question is, what is going to be the mainstream use that everybody goes for and is intuitive and useful for everyone? And you know that we don't know whether the Codex and Claude codes are going to be that type of mainstream use or whether they're another flash in the pan.
- [25:18]
B
I Still, I don't know. I guess it's still how you define like just day to day search. I consider to be the killer use case. Like day to day information gathering and retrieval knowledge partner. I mean, I don't know everyone. Actually. I don't know if this is like depressing or romantic or something in between, but like the other day my wife and I were sitting at like separate ends of our couch and we were both dictating into our computers and we were both like whisper flowing into like our, our computers. And I'm like, yeah, I know, but you have to do it especially at a lower voice because you're like. Because you don't want to whisper into the other person's AI. But this was like.
- [26:10]
A
It's just seen in my head.
- [26:12]
B
I know. No, no, I kind of. And I'm like, oh my God, are you, are you whispering too? That. She, she got it from me. She always saw me dictating. And then like it's ingrained in everyone I know, my parents. Like, like, it's just the chat experience is already there, but that should be reflected.
- [26:33]
A
This is what I'm saying. That should be reflected in the, in the billion users of chat GPT. Right. So it's just taking longer to get there than anticipated. I'm just making this case. Let's. Let's let it breathe a little bit.
- [26:44]
B
900, 920 million users. What do you want?
- [26:48]
A
Come back. Yeah, again, Again, the word that has been used is exponential. If it's true exponential growth, that's where it should be. If that's what you're pitching Wall street, that's where it should be. So that maybe if that's leveling off. That's what I'm saying. Thinking about through the, the thinking through the impetus to go public, maybe it's like, all right, well, this is probably the best story we're ever going to tell. Let's go.
- [27:10]
B
Do you think there's been no real numbers around the actual compute utilization or total queries per user, Anything like that that you've seen is there?
- [27:23]
A
Just from Google. Google gave this week that they are now processing 3 quadrillion tokens. It's gone.
- [27:32]
B
That doesn't.
- [27:33]
A
9.7 trillion to 3.2 quadrillion per month in two years, according to Sundar Pichai, who, by the way, I mean, I won't pile on, but it was a Fairly underwhelming Google I.O. kind of as we expected. No big. Not a real big moment for them. But anyway, let's put that aside.
- [27:49]
B
Well, does that, does that include. See already theirs is a little different because they can just route, which they are increasingly doing, traditional Google search to AI mode. And everyone who is just using Google search is now ending up, whether they want to or not, using AI. So that one still doesn't feel as real to me like, or important or relevant like to me. I want to know, are people, out of those 920 million users, are they using it significantly more today on average than they were three months ago or six months ago? That's the only question that matters to me.
- [28:29]
A
Yeah, yeah, okay, just floating it out there. How about, how about canceling cloud code when you see the money stack up? We've heard from Uber, for instance, instance, talking about how they like exhausted their cloud budget in four months. Many other companies are also talking about the ROI here and Microsoft, interestingly, although maybe this isn't indicative of Claude, but Microsoft, according to Verge, starts canceling CLAUDE code license licenses. Microsoft first started opening up access to CLAUDE code in December. It was part of an effort to get product managed project managers, designers and other employees to experiment with coding for the first time. Microsoft is now planning to remove most of its cloud code licenses and push many of its developers to use Copot Copilot CLI instead. Here's what Chamath says about this and then we'll turn to you. He goes Microsoft pulling CLAUDE is the first but not the last. The issue isn't that the tool isn't useful. The issue is that without context and oversight, the tool can spin forever and generate an enormous cost burden that when cascaded across an entire employee population, makes using the tool economically untenable. Again, like thinking like, well, is this the top? Like if you were going to talk about anthropic that way, maybe they're just like, oh, people are going to get religion about the amount of tokens they're spending and our growth rate is going to slow. I don't know feasible.
- [29:49]
B
I. Okay, I think we're starting to maybe put together a semi coherent theory across all this. I'm not getting, I'm not going to full coherence.
- [30:01]
A
These companies are going to both get a trillion dollars, they're going to get trillion dollar market caps, they're going to raise all the money in the world and we're here like why are they in a bad position? But hey, this is our job.
- [30:12]
B
I'm going to give you kind of what I've been seeing happening over the past six months and it was interesting that this article specifically references this time period. Of the last six months is that CLAUDE code kind of gets released and starts picking up. It is a revolutionary product. And then no one has even thought about token budgets ever. Everyone is allowed to use it, who is given access as much as they want. Every software engineer specifically just cranks out whatever they want. Usage explodes, revenue explodes. And then in the. But it's an experimental time. There's been a lot of. I've been seeing around and a lot of discussions I've been having around even trying to attribute ROI or measure even individual productivity. Was it actually in line with the, the costs associated with it? People are starting to severely question. So you're starting to see this. You. I mean again we went from token maxing which we all knew was this kind of point in time that we'll always look back at and be like what were we thinking? But it was, I mean it was just that moment that everyone's like okay, we're, we're really running with this. This is something new and different. Everyone needs to be using it. And then everyone stopped and was like holy shit, how much does it cost? And it was, we know it's heavily subsidized anyways, so like when you put all that together, you start seeing all of this happening. Even chamath like to me and this is like at writer the type of stuff we're working on. We have our own foundation models but interoper model routing based on request being model provider agnostic on the foundation model side is where I see this going. It's not going to be one mod and I mean we've talked about this for a long time. One model to rule them all systems that can route and chamath talking about. I rarely agree with chamath but I mean like if different models we're already seeing it. Like if it's video like VEO is very expensive so seed dance or cling or depending on the type of request and complexity routing to a different model is the way I see things going. And I do think like the, the, the graph of revenue growth for Claude for the last six months we will never see again. It was just something where no one, yeah, no one was checking anything. No one cared. And now suddenly they all stopped and we're like oh wait, what are we doing?
- [32:49]
A
Okay, great. So now let's just end the segment with three questions for you. First, do you think. So I have now brought out the best arguments I could to, you know, sort of say this might be the top. Is this the top?
- [33:08]
B
Define top.
- [33:10]
A
Well, actually just answer it the way that you're thinking.
- [33:14]
B
I. Well, it's more again like I genuinely believe over a multi year period overall AI is going to become maybe not Elon Musk's $28.5 trillion TAM for SpaceX, but something dramatic. But I think we're going to start to see that like going from 4 billion to 10.9 billion estimated. I don't think they're going to get to 10.9 billion. And I know Claude Anthropic has proved everyone wrong, but the last six months were something different and it's already stopped and we've been, I've been seeing that it's being reported on now. So I don't think it that kind of like growth that has never been seen before in the history of. I think like any company anywhere is going to continue in the same way it was. So I think we are at a top of that kind of movement and momentum.
- [34:09]
A
Right. I mean if you continue to see the type of growth that Anthropic has seen, like in three years from now they would be the entire stock market, right? Just like. Yeah, just 10xing every year. The way that they are is crazy. But I, so I, I get that it's like not the top, but it's the end of the acceleration curve which would, of this particular acceleration curve which would indicate to me that it would be the perfect time to go public because that's what you want to show the market. You don't want to show the deceleration, you want to show the acceleration market likes those type of curves. And that's going to lead me into these next two questions which is are these companies ready to go public? So let's start with OpenAI. Is OpenAI ready to go public?
- [34:55]
B
I think I'm still, I'm going to say yes because I think they're ready to go public as they'll ever be. And the story that they're going to have to tell the market and the market is going to have to buy is one of we are over investing in compute and everyone else is going to be screwed and we're going to be in a good place and so they're as ready as they'll ever be. What about you?
- [35:21]
A
I agree.
- [35:22]
B
Okay.
- [35:23]
A
I agree. This is a rare moment of full agreement and you. I'll take it. Okay. Anthropic. Is Anthropic ready to go public?
- [35:32]
B
I mean with that SpaceX accounting stuff, they better be. And if you're, you put it all together, you get selectively to the Wall street journal a little S1 filing only revealing from an exit Tron blog post some potential shenanigans. I think you've put everything in place. We've talked about a lot. They are, I mean, for the kind of like high level narrative of just kind of like research lab. Nice guys. They are a communications maestro. So it seems like they're tying this all together. So they better go now.
- [36:13]
A
Yeah, I think I really should just title this episode or Anthropic and OpenAI. Or OpenAI and Anthropic Ready to go Public. But yeah, I agree with you also. Make it happen. Do it. And also honestly, like, this is such an important cog of the economy that whether they're ready or not, they should just do it right for the public. It's sort of like important for us to get a look into these numbers. So instead of like guessing like should Oracle be worth triple or zero, we can just kind of have a much better understanding. So. So I'm all in favor.
- [36:44]
B
That's why they, that's why they should do it for the public.
- [36:48]
A
If you really cared about safety and the good of people that you're building AI for, go public so we can see whether you're BSing us or not. That is my pitch. I think they'll listen.
- [36:58]
B
I think they'll, I mean, bringing. What did Sam say? He wants to deliver magic at Hyperscale. This is how you do it, Sam. This is how you do it. Open the books.
- [37:12]
A
Show us the books.
- [37:13]
B
Show us the books.
- [37:14]
A
We're in complete agreement here. All right, folks, well, if you want an even deeper dive or a similar. Actually a somewhat different perspective on this from someone who's actually brought a company public that had high expectations, tune in on Wednesday, where ex Twitter CEO Dick Costello and I will go through the OpenAI, Anthropic and SpaceX IPOs. And what is going to happen to these companies and their employees once they hit the public market? All right, we've done two. We've done OpenAI and Anthropic, where we got one left. We got SpaceX. And judging from Ron John's earlier comments, you're not going to want to miss his reaction to the numbers that we saw in the filing to go public from that company, which we're going to cover right after this. This episode is brought to you by True Diagnostic. I've been trying to get more intentional about my health lately. Not just how I feel day to day, but what's actually going on under the hood. That's why I checked out True Diagnostic. They offer at home tests that measure your biological age, not just how old you are, but how your body is aging on a cellular level. Their True Age test looks at things like your pace of aging, organ system health, and even risk factors tied to lifestyle, giving you real data to act on. What I like is that it's not guesswork. You can track changes over time and see how things like sleep, diet or exercise are actually impacting your body. And taking the test at home was so easy. If you're seriously about optimizing your health and longevity, this is a really powerful tool right now. Big Technology Podcast listeners can get 20% off at truediagnostic.com using code Big Tech at checkout that's true diagnostic.com and use Big Tech for 20% off today. Choose Truage True Health or the Combo Kit as a one time purchase or a subscription Most leaders know how work is supposed to happen, but when it comes to how it actually gets done day to day across tools, teams and handoffs, they're mostly guessing. That's exactly the problem Scribe Optimize was built to solve. Trusted by over 80,000 enterprises, including nearly half of the Fortune 500, it gives leaders a live view into how work is really happening across approved business apps without interviews, manual process mapping or extra effort from the team. And because it's continuously analyzing real workflow activity, the insights stay current instead of going stale the moment a process changes. You can see which workflows are happening, where time is going to, and which tools are involved. It automatically surfaces top issues, explains why they're happening, and even recommends ways to fix them with estimated time savings. And importantly, it's built with privacy in mind, so activities only captured in admin approved business apps and user level data is anonymized by default. The kind of visibility that used to take months. Now it's just always on. If you're ready to stop guessing and start seeing, Visit scribe how BigTech that's S C R I B E Every company says AI will make employees more productive, but most employees are still stuck waiting on it, waiting for app access, waiting for password resets, waiting for someone to fix a laptop issue so they can get back to work. That operational drag adds up fast, and IT teams are overwhelmed trying to keep up. Serval was built to automate that work. You describe what you want automated in plain English, and Serval builds it for you. No complicated workflow builders, no consultants, just faster support and fewer tickets slowing everyone down. The platform is designed to eliminate repetitive tickets, so it can focus on the strategic work instead of constant firefighting and it enables it teams to build automations using plain English instead of drag and drop Workflow builders. Learn more or start a free four week pilot at serval.com BigTech that's S-E-R-V-A L.com BigTech serval.com BigTech and we're back here on Big Technology Podcast Friday edition. Let's dig in to this delectable SpaceX S1 the prospectus to go public. This from Bloomberg. SpaceX shows a $4.3 billion loss as Musk targets a record IPO. SpaceX filed publicly for what stands to be the largest ever initial public offering, revealing billions in losses in super voting share plan. A super voting share plan allowing Elon Musk to keep the company under its control. SpaceX had a net loss of 4.28 billion on revenue of 4.69 billion for the first quarter, compared with a net loss of 528 million on revenue of about 4 billion a year earlier. You know, I was, I was asked to go on CNBC to talk about this stuff this week. We didn't actually get to the SpaceX segment. We talked mostly about OpenAI and Anthropic. But I went deep into the numbers here and the more I looked, the more angry I got. And Ron John, I get the sense that you have a similar feeling.
- [42:00]
B
I don't know if I'm angry. It's more like I was shocked. So, okay, so OpenAI at 100, negative 122% operating income margin. I'm weirdly okay with that. SpaceX at a total of $18.7 billion in revenue, of which 11 billion is Starlink, and that's the whole thing. And it's a $2 trillion company. I just, I don't know what, how we got here. I thought this was such a, a bigger thing, a bigger, like, vehicle of capital and cash and everything that I. This, this genuinely shocked me. What was your reaction?
- [42:49]
A
I'll explain why I was mad. Maybe annoyed is probably the better term.
- [42:54]
B
Be mad.
- [42:54]
A
Because mad.
- [42:55]
B
Be mad.
- [42:55]
A
I'll be mad. I'll be mad. Fine. Screw it. This is my mad face. Okay, so this is what annoyed me. SpaceX told us that it had the biggest total addressable market of all time, $28.5 trillion. That was the market. Okay, so I'm thinking, you know, you're gonna go to space and do all these things, build the Mars colony. What is the biggest sort of revenue driver there? 93% of those opportunity of that opportunity is AI with enterprise applications accounting for the vast majority. So SpaceX thinks of this 26.5, wait, $28 trillion. $28.5 trillion TAM 22.7 trillion is AI enterprise applications. What I mean this is like, this is, this wasn't your business a year ago. Now you acquire XAI and all of a sudden AI enterprise business applications are what you're going to do. Obviously like your Grok AI application has trailed Google, Anthropic and OpenAI. You're not seeing anywhere near the level of business. And I'm just saying that not as a hater. I'm looking at the numbers. You're trailing those businesses dramatically losing money and you're just not in position to capture this $22.7 trillion. Now maybe you could say our business is going to be data centers in space, but I, I, I don't believe in data centers in space. If you have something go wrong in a data center, which by the way everything industrial you have stuff go wrong all the time. You send a technician. If you have something go wrong in a space data center, you have to send an astronaut. I'm just not anywhere close to sold that that is going to happen and that SpaceX has this 20 plus trillion dollar total addressable market that it's going to capture. You just rented your AI warehouse Colossus 1 to anthropic. So are you a NEO cloud? Are you a builder? It's hard to really wrap your head around the story here.
- [45:07]
B
Even SpaceX. What was kind of fascinating to me is that the actual like space segment was, which is the one that has made money over time, which has still been declining overall and actually like you know, renting out your or building rocket ships for others to launch that all of the growth is supposed to come from Starship, which is actually not even started yet and is just pure R and D but still forecasting out all this growth. Obviously Starlink. StarLink is very profitable. 11, it's like 11.7 in revenue, 4.4 in profit. Billion billion in profit. So it's a very profitable business. But then those aren't even the key parts of the story. It's just a very vague kind of picture of AI with again I still this one was all worth it just for the line that it's like the, what was it the largest actionable, largest addressable market in human history just for that line. This is worth it though.
- [46:12]
A
I mean do you believe in data centers in space? Do you think that's going to happen.
- [46:16]
B
It could, but normally you would do that and then ipo.
- [46:23]
A
Now I'm going to say, don't be ridiculous. Don't be ridiculous.
- [46:28]
B
I know. Okay, fair.
- [46:29]
A
What planet are we living? You must be in space.
- [46:32]
B
Yeah, you're right. Because instead what you would do is buy Twitter, have its revenue drop by half, and then. And then also launch a very successful innovative satellite Internet company, and then ipo, and then build your first data center in space.
- [46:55]
A
Yeah, I mean, Jeff Bezos thinks it's going to happen. He talked about it on CNBC this week. He said the two to three year timeline is a little ambitious. But Blue Origin, which is his space company, recently submitted plans to the FCC to send 51,600 data center satellites into low Earth orbit as part of an initiative dubbed do you think Project Sunrise. Okay, I'm reading this now. All right, maybe. Maybe Elon and Bezos can do it. Or they just send them. They have some sort of networking. They're all satellites up there. They're off Earth. They're cooled by space, and they can somehow.
- [47:28]
B
Do you think people are going to protest that? Do you think people protest the spacers in space? Data centers in space?
- [47:38]
A
Probably. People protest everything.
- [47:40]
B
I know that sounds like parks.
- [47:42]
A
The Park Slope Food Co Op down the block for me is getting protested, so.
- [47:46]
B
Oh, really?
- [47:46]
A
I don't know why. Yeah. Oh, international politics, of course. Because no one. No one really sends a message on international politics. Like a grocery store.
- [47:57]
B
No co op in Park Slope. That's right.
- [48:01]
A
It must stand for what's right.
- [48:03]
B
I was thinking, like, maybe data centers in space is going to be my. Should that be our entire political platform? Because it does solve a lot of the challenges that are surrounding data centers. And it seems like actually this perfect answer to it. It's still funny just because, like, I actually am excited to see why it angers people. I don't know. Maybe there's like, space purists. I don't. I'm sure.
- [48:35]
A
I don't know.
- [48:35]
B
There could be.
- [48:36]
A
Yeah, here's my. I mean, 51,000 Jeff Bezos. That's the thing. Like, it's all fun and games until an astronaut dies because they hit your data center in space. Like, could you imagine, like, the reaction of the public when, like, we lose a spaceship because we decided to make Studio Ghibli's in space instead?
- [49:00]
B
You're right. Maybe. Maybe I'm going to get into space junk. I'm going to go the other way. Space junk is bad.
- [49:07]
A
Let's.
- [49:07]
B
Space junk.
- [49:08]
A
Let's Boom.
- [49:08]
B
Yeah.
- [49:09]
A
The next stop, space junk that's talked about data center in space. Ronald and I will be out there booing signs up.
- [49:16]
B
Yeah, no more space junk.
- [49:19]
A
All right. So, speaking of the commencement addresses, I think we should talk about this. Where Eric Schmidt, the former CEO of Google, was booed this week at the University of Arizona's graduation ceremony. Not only him, there was a woman named Gloria Caulfield, a real estate executive who gave the commencement at University of Central Florida. And she was booed. And then this guy, I love this guy, Scott Borchetta, the CEO of Big Machine Records, he got booed and he kind of yelled at the graduates at Middle Tennessee State University. He goes, deal with it. Like I said, it's a tool just kind of mouthing off at these graduates. Further evidence of AI's reputation problem, which we've been talking about. It seems like it's really hitting a fever pitch now, right?
- [50:01]
B
Oh, I think. I mean, it's still interesting to me that, like, why do you. Why is Eric Schmidt speaking at this commencement and why is he bringing this up? Like it's. That part is still kind of just odd to me in general. And also, are people specifically, do you think they're booing Eric Schmidt or AI? Or both?
- [50:22]
A
I mean, both, but more AI. I mean, you watch the. He's very booable.
- [50:27]
B
Eric Schmidt is very booable. Yeah.
- [50:28]
A
Every time AI comes up, say what you want about Eric Schmidt. Every time AI comes up, that's where the boos get really loud. And you had two other examples. So it's. They're booing AI. This was something that I thought was interesting from Buco Capital bloke, which is a fun follow on Twitter. One of things is, one of three things is true about Big Tech's AI narrative. One, nobody can tell a compelling story about how AI will be good for society. Two, it hasn't occurred to them to tell that story. Or three, they believe it'll be bad, but it's not their job to fix it. And I don't know what's more damning.
- [50:59]
B
Oh, I. Yeah, I agree.
- [51:01]
A
Doesn't that capture it all?
- [51:03]
B
No, no, yeah, it does. It does. And I think. I mean, we've talked about this a lot. AI has a PR problem. You wrote about it today. I think, like, it's still baffling to me, but this is where Dario going out saying there's going to be a decimation of jobs. No. 1 painting an actual, like, kind of positive picture of what this looks like or how. How does. How does the world get better? No one has actually painted that picture. I mean, the meta layoff stuff this week. Did you read about it?
- [51:38]
A
So dystopian. Talk about it.
- [51:40]
B
Yeah, well, so. So meta. I think it was 8, 000 people.
- [51:44]
A
8,010 of the workplace.
- [51:46]
B
Do they call them medins or anything like that?
- [51:49]
A
Metamates. Metamates.
- [51:50]
B
Oh, God. Do they still do that? Is that real?
- [51:53]
A
Yeah, well, the 8, 000 are no longer meta mates anymore. Some consolation to that.
- [51:58]
B
No longer Metamates. I. But there was leaked audio of the
- [52:04]
A
all you gotta talk about this.
- [52:05]
B
This one was incredible. And like, come on. Mark Zuckerberg's communications team. I know he's not listening to you right now, but still try to explain to him how bad this line of messaging is.
- [52:21]
A
This was internal. This was internal. Wasn't. But it was still. Okay, can we have the exact. I'll read the language. Because we have the exact.
- [52:28]
B
Read the language. Read the language, please.
- [52:29]
A
So they were asking about being, you know, so there's like this narrative. Okay. Meta's monitoring all the keystrokes. Some people are like, they monitored my keystroke, then they laid me off because they know how I do my job. This is how Zuckerberg explained it to the company. The AI models learn from watching really smart people do things. The average intelligence of the people who are at this company is significantly higher than the average set of people that you can get to do the tasks. So if we're trying to teach the model coding, for example, then having people internally build tools or solve tasks that help teach the model how to code, we think it's going to dramatically increase our model's coding ability faster than what others in the industry have the capabilities to do. Who don't have thousands and thousands of extremely strong engineers at their company. Okay, I'll just say that sounds logical to me. I don't know. I mean, sue me. What do you think?
- [53:16]
B
That's my point. It. It does sound logical. It is like, so coldly correct yet. And I also did love that he threw a little shade. He's like. And we even got this guy Alex over here, Alexander Wang, founder of Scale AI, who ran a business that had kind of the dumb people doing the coding, those contractors. But you guys, you guys are smart. So us training on your keystrokes is a lot better. It's wild to me that that becomes the kind of narrative knowing that this is going to become such a big topic if they're really doing this. Like, he started to get there and you start to think like, Our models will be better because of this, which will make you more money and Facebook more successful and you more successful. Okay. That's at least somewhat of a reasonable thing. But you are smart, so you should almost feel honored to be trained that your keystrokes are being trained off of versus those less intelligent contractors just at every level. Come on, come on.
- [54:27]
A
I mean, all right, just to. Just to contextualize it to be fair here, like, those keys, like, this is the way we've talked about in the past why they did this. We talked about why this is the way that AI is being trained right now. It's watching people go through environments and using that as reinforcement learning. So in some ways I get it. But I do agree that, like, there's an overall. This is such a difficult. I mean, can you actually possibly message we're monitoring your keystrokes in a good way? No. Should you be doing that? I don't know. It's sort of a toss up. It's kind of creepy. But the thing is silly. Oh, go ahead, go ahead, go ahead.
- [55:03]
B
Hold on, wait. Okay, what if I come in? You know, we are actually at an existential point. Our models on the last benchmarks were laughed at by the industry. We've had. We must, for the entire future of Meta and your jobs, all of you. We need to improve our foundation models. The best way we have decided we can do that is to all come together. I will monitor my own keystrokes and we will together train these damn models and get back in the race. That's my pitch.
- [55:40]
A
I love that. No, that would be much better. That would be a much better way to put it.
- [55:44]
B
I mean, he should have said, I'm, I'm. I'm training my. I am monitoring my own keystrokes dramatically. Open his computer. And he's getting. He's getting monitored too.
- [55:55]
A
Yeah, yeah, you can see. You can see Net Nanny on my computer as well. No, I think this is one of the things, I mean, obviously, like, I don't know, I was going to say something positive, but I'll just say this, that like, Silicon Valley cannot get out of its own way on this. And I think it's clear. Let's go to Marc Andreessen. And I know this is bait because, like, Marc Andreessen is going to go on Joe Rogan to say baity things so the people will pick it up and they'll be like, get outraged. And Andreessen will be like, look at these babies. So I understand that this is bait and I'm gonna I'll take it anyway. But Andreessen, who was making the case that AI is gonna help our productivity, managed to say, while making that case is stunning, he goes. He was talking about the. The bots, right? He says the bots never get frustrated with you. They never get sick, they never get depressed because their girlfriend breaks up with them, and they never file HR complaints. Even if you think these things, don't say these things like you're not the. The botching of the storytelling on this front. And you know, we talk a lot, a lot about comms. So fine, get the criticism. But, like, this is going to have real implications. You're going to have candidates talking about kill switches, moratoriums. You're going to have data centers not being able to be bu. The country is going to turn on this stuff. The booze matter is my point.
- [57:12]
B
The booze does matter. I think. Like, does. Do you think it's solely a function of that kind of personality? Including an. Andreessen just has not lost in a long time. I mean, they lost on Crypto Net Net, but like, it's just. It's this complete invincibility. But still, you know, Trump comes into office and David Sachs becomes the aizar, and these companies become the center of the economy. Like, do you think it is simply that, or do you think like. Like what would be something that would have to happen to actually stop this? Kind of, like, make people a little bit more cognizant of what they're saying and realizing that baiting people like that is not a good thing?
- [58:01]
A
Well, the technology would have to stop, right? Because ultimately, like, say what you want about tech in the US it is. Its march has continued unabated, and the technological progress has been obvious for everybody to see. It's incredible, the technological progress that we're seeing. And it hasn't lost. And honestly, like, it's.
- [58:20]
B
It.
- [58:20]
A
It is at a point where it's going to be diffused in a crazy way. AI. So I think their belief is whether they like me or not, this is going to be. This is going to just, you know, get everywhere soon. And maybe they're. They're not wrong. I don't know.
- [58:35]
B
I wonder.
- [58:35]
A
Although it's not a politically strong message for sure.
- [58:39]
B
Do you think in China, are any of the, like, leading AI figures equally dickish as, like, many of those in. In the U.S. well, you think. I mean, there's allowed to be.
- [58:53]
A
You don't. You don't need to sell the idea because it doesn't matter. It's going to happen no matter what. In the US the public does have some and.
- [59:01]
B
Yeah, but in your.
- [59:02]
A
They're obviously not latitude here.
- [59:04]
B
No, but they're not. They can't be trying to sell the idea because if you're trying to sell the idea, this is the worst possible way to do it. Like
- [59:16]
A
in their minds. I mean, Andreessen's mind. I think when he talks about AI, he's trying to get people excited about it. But I ultimately think that like he doesn't. Yeah, I don't know. He probably cares more about the business leaders than the public and thinks that I want to get. I want to speak to the business leaders and it doesn't matter what the public, the way the public reacts here because they're not my constituency, I guess
- [59:40]
B
I cannot imagine them being sold like trying to sell this. If this is you're selling. Come on. He was an entrepreneur. You have to learn the art of selling at some point. But I'm looking up the. The founder of or the CEO of Dobao, the basically ChatGPT equivalent run by ByteDance. A guy, Alex Zhu. I'm gonna do some more looking into this. I'm curious now, like, are there big personalities who are just saying ridiculously stupid things like.
- [60:15]
A
No. Well, you know what happens to big personalities in the China business community.
- [60:20]
B
Yeah, yeah, okay, fair.
- [60:22]
A
How many interviews has the deep seat guy given in international media? Case in point? Yeah, he is there. Do your thing. Keep a low profile. Make the country proud.
- [60:31]
B
I don't want to say maybe we could learn something, but maybe just tone it down a little. Guys, we're going into an election year. Don't ruin this for everybody. Just.
- [60:48]
A
Well, I can say what one thing, one thing is guaranteed. This is now a political issue, right? Something that's pulling like this, the politicians will. Some politician will learn to exploit and that I think like that is where you could end up getting derailed. In my opinion.
- [61:05]
B
The largest consumer facing artificial intelligence product in the Chinese speaking world. Doubao is named. Named after a steamed bean bun. That is how you make people friendly and happy.
- [61:20]
A
All right everybody, have a great holiday weekend. Ranjan, great to see you as always. Thanks for joining.
- [61:25]
B
All right, see you next week.
- [61:26]
A
See you next week. Thanks everybody for listening and we'll see you next time on Big Technology Podcast. Every Sunday we cover the week's tech news on this week in Tech. Hi, this is Leo Laporte inviting you to join me this week as Berbergin from the Wall Street Journal and Paris Martineau from Consumer Reports. Join Ian Thompson and we'll talk about, of course, OpenAI and Anthropic. They got together with a bunch of religious leaders and decided what religion AI is. They've also figured out how to keep it from blackmailing you. You just say, well, that would be wrong. This week in tech, you'll find it at Twit TV and wherever you get your podcasts. You may think you know McDonald's drinks, but but you don't know them like this. From fruity refreshers like the Strawberry Watermelon Refresher and the Mango Pineapple Refresher with Popping Boba to crafted sodas like the Sprite Berry Blast with berry flavored Sprite topped with cold foam. Who knew ice cold drinks could be so fire six all new drinks are here. Try them all now. At McDonald's. Refreshers contain caffeine.