The US banned Anthropic's Fable 5 release, but the numbers don't seem to care

Just as last week was ending, the US government forced Anthropic to pull its two newest models, Fable 5 and Mythos 5, citing national security concerns after Amazon researchers allegedly found a way to bypass Fable 5's guardrails. Cybersecurity researchers have since signed an open letter calling the move dangerous, and Anthropic itself noted the same jailbreaks exist in other models. So is this a genuine security concern, or just the latest chapter in a messy relationship between Anthropic and the Trump administration? On this episode of TechCrunch's Equity podcast, hosts Anthony Ha, Sean O'Kane, and Rebecca Bellan unpack what the ban means for developers building on Anthropic's platform and for anyone watching the IPO, why it might accidentally be good for the company, and more of the week’s headlines. Listen to the full episode to hear more about: Why the UK's social media ban for users under 16 might be the lesser of two evils What the SpaceX-Cursor acquisition tells us about xAI's strategy (and its gaps) Jeff Bezos's $12B bet on physical AI with Prometheus, the startup trying to build an "artificial engineer" Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod. Learn more about your ad choices. Visit megaphone.fm/adchoices

• 0:00Speaker 00

  Anthropic is fighting with the US government again. But will that actually help its business instead of hurting it? Also this week, the UK announced a ban on social media for teens. And there are a couple of really exciting deals, including a big update for SpaceX. We're going to get into all of that, so stick around.

• 0:22Speaker 00

  Hello and welcome back to Equity TechCrunch's flagship podcast about the business of startups. Today is Friday, June 19th. I'm Anthony Ha and I'm joined, as always, by TechCrunch senior reporter Sean O'Kane. Kirsten's out this week, so we have Rebecca Bellan, who will be familiar to Equity listeners, joining us today. Rebecca, what have you been up to?

• 0:41Speaker 01

  Oh, you know, I've just been hanging out in New York, mixing five, baby. It's been pretty exciting.

• 0:48Speaker 01

  Been pretty exciting over here. The parade's happening as we record this. And what else have I been doing? I interviewed Andrew Yang last week. I interviewed Tiffany Luck from NIA this week. So it's been a pretty, pretty eventful couple of weeks.

• 0:59Speaker 00

  Awesome. Well, people should check that out. We're going to be doing something a little bit different format wise. Listeners may already have noticed that. We're just trying to get into the kind of meaty theme discussions as quickly as possible. If you like it, if you don't like it, let us know. But yeah, let's start with our first theme.

• 1:14Speaker 02

  You know, I've been talking for the last couple of weeks about how much I want to see the risk factors in the Anthropic IPO. Now I'm really interested in seeing what they're going to look like based on what happened on Friday. And I mean, the short version is that we got a new, new features and a new sort of public facing version of Mythos from Anthropic called Fable 5 just about two weeks ago. And then all of a sudden on Friday, we no longer have it. If you, if you use Claude at all, you know that it is not available. They make it abundantly clear. I know there's a lot going on here, Rebecca. I'm hoping you can sort of briefly tell us what's, what's happening.

• 1:52Speaker 01

  Well, I mean, as I'm sure many of our listeners know, the, the US government basically just forced Anthropic to pull its two newest models offline, Fable 5. And then there was also Mythos 5, which was the one that's, it was available to current Mythos users. Right? So Fable 5 was more available to the public. So they sent a letter on Friday that cited, quote, "national security concerns." No one knows what those concerns are. That report has not been made public. They gave no specifics.

• 2:18Speaker 01

  And told them that they had to Essentially ensure that those models couldn't be used by any foreign nationals. So Anthropic was like, okay, I guess we have to just pull the models entirely because we don't know when someone's a foreign national, like a lot of our own employees are foreigners. So it became a problem, but really the stem of it came from, the White House got tipped off to this because of some Amazon researchers that allegedly found a way to bypass Fable Five's guardrails. Amazon CEO Andy Jassy raised these concerns with the White House, and it just kind of spiraled from there.

• 2:52Speaker 02

  This all moved really fast, especially for a Friday afternoon into a weekend.

• 2:56Speaker 01

  Friday evening for us in New York.

• 2:59Speaker 02

  Yeah, at the same time that the administration was ostensibly trying to negotiate some sort of treaty for the war that it started in Iran, I think one of the things that we've been focusing on on the site all week is, and teasing at in the stories about this, is how much this could impact the other companies that are working on similar models? Because I know that OpenAI likes to think that it has something that is sort of Mythos class in the works as well. Let's step real far back for a moment. Anthropic has not had the best relationship with the Trump administration in a way that stands apart from the other leading AI labs. I think there's an element at least of that playing here.

• 3:44Speaker 02

  So do you think that this is going to have implications for those other companies? Or do you think that the Trump administration would be less inclined to turn off the tap on one of those competitors?

• 3:56Speaker 00

  I mean, part of the context here is that, including a number of stories that we've run on TechCrunch, it seems like both the reporting and the analysis from independent security experts suggests that the risk, the actual security risk from Anthropic is not that unique and so that a lot of this seems to stem more as much from just parts of the Trump administration and Anthropic just don't get along very well and so whatever risks there are those things are going to blow up kind of out of proportion just because they it seems like they can't have a civil phone call with each other and so if you're another company on the one hand maybe that's advantageous to you because you can say well

• 4:40Speaker 00

  like we just don't you know get these guys mad at us and we can do what we want but you know that's also like not a great like regulatory landscape to just be like well boy I hope they don't get mad at us.

• 4:52Speaker 01

  On the one hand it definitely feels retaliatory like after the government labeled Anthropic a supply chain risk there's this big lawsuit going on between them it really feels like the White House is just kind of looking out for any excuse to pummel Anthropic and a lot of you know and I feel that way not only because that was my initial reaction but because of what a lot of cybersecurity researchers have said right they say that this should never have triggered an export control they've all like signed an open letter to ask Trump to revoke the order and they say it's actually dangerous to pull these advanced cybersecurity capabilities from network defenders in the US right so Anthropic itself said some of the same jailbreaks could have been found in several other AI models

• 5:34Speaker 01

  like cynically it's like okay are you just pausing Anthropic so that others can kind of catch up to where Anthropic was but at the same time I've also seen reactions that are like Anthropic kinda had this coming right they're like this is too dangerous for anyone to use but not us we're the good guys right so like you know they're kind of talking out of both sides of their mouth you know a week before Fable came out they were like hey we need to slow down AI development guys it's getting really dangerous but then boom here's here's our most insane ever super powerful model go off right.

• 6:08Speaker 00

  Yeah, in some ways, this feels like a microcosm of a lot of the discussion around AI, where it seems like, you know, people like Sam Altman, Jensen Huang are also like, "Hey, let's try to lower the temperature. Why is everybody mad at us?" And it's like, "Well, you spent the last couple of years essentially saying you've like built this god machine that will take jobs away from everyone." Like, it's not exactly a shock that people don't feel great about this. And I think, yeah, Anthropic's, there's something about the way Anthropic talks about Metos in particular, where they're like, "This is, you know, the most incredibly powerful model ever. It's too dangerous to release to the public." And so, on some level, it's like, "Well, okay, let's say that we take that seriously then." And that means that there's going to be an incredible

• 6:53Speaker 00

  level of scrutiny around it. And I do wonder, I mean, it does seem like Anthropic is not happy about this. I want to be like careful about how, like, not overstating that how this could be beneficial to them. But it is also true, you know, we also ran some stories about how, you know, some ramp analysis highlighted the fact that the last time, the last big blow up between Anthropic and the Trump administration was good for the company in at least some ways, that, you know, their downloads of Claude shot up. I think a lot of people who maybe had thought of ChatGPT as kind of the chatbot, the AI assistant before, suddenly they were looking at Claude as maybe the more responsible one, the more kind of quote-unquote "resistance" one. And in the same way that this, I think,

• 7:35Speaker 00

  you know, Anthropic is very stressed out about this, but also this could, again, make their model seem even more powerful than they already seem.

• 7:44Speaker 01

  Yeah, definitely like, "Oh, we're so dangerous." Everyone loves a bad boy, right? Like, it's like, everyone's like, "It's the most powerful model." Like, even Trump says no. Like, of course, I gotta get my hands on it, right? Like, you have to, you know, we're already seeing, I mean, the ramp numbers from May, I mean, this is just, you know, I assume that, we don't have numbers since all of this debacle went down, but I assume that the trend will increase. Like Anthropic's most recent models, like OPUS 4.8, that is so popular with businesses right now. In May, Anthropic's share of, you know, enterprise customers rose to 41%. That's above OpenAI's 39.5%. It's a small win, but it's a win nonetheless.

• 8:20Speaker 02

  How much of this is coming from just how fast we're moving, right? Like, if everybody was developing this technology a little more slowly, I'm sorry for something so blasphemous, but, you know, there's, I'm sitting here trying to think about...

• 8:38Speaker 02

  The ways that these companies are talking about themselves, how they're kind of shooting themselves on their foot by hyping up the capabilities and the worry and then having to walk that back in some regard. And I'm trying to like trace that back to an origin point. And it really only seems to come from how quickly they're moving and how from the moment that, you know, chat GPT really blew up in late 2022, we've just been sort of like nonstop barreling forward. And that there's, it does not seem to me that there is a lot of consideration going on inside of these labs as to whether they should take a more cautionary approach in when it comes to putting this stuff out there. Like to

• 9:19Speaker 02

  your point that you just made Rebecca, like they were talking about how possibly dangerous the complications, you know, could come from mythos and anything branched off of mythos being released. And then all of a sudden Fable was there a week later. What do you guys think about that? Like, I don't think that there's any, especially with IPOs looming, any reason that believe that they're going to slow down. But like, is that it? Or is there more that we should be thinking of as far as like, why this all seems like such a mess all the time?

• 9:47Speaker 00

  I mean, this is a little bit of what we mentioned earlier, but I think that's an element in terms of how the public talks about it. And obviously, like, the government is, you know, part of the the public in that sense and reflects public opinion in some ways. But it also seems like maybe everybody could be more cautious and still, you know, certain people in the Trump administration and certain executives at Anthropic would still just not get along for for whatever reason. And again, without necessarily going too deep into this, it's sort of also complicated because it seems like there are parts of the Trump administration that really like Anthropic. And so you also get these very confusing headlines where on there's certain, you know, certain departments that are embracing mythos and trying to get them introduced to like all the big banks and everything and

• 10:30Speaker 00

  other departments that are like, no, national security risk. And so, yes, I think, obviously, I mean, I do wish that maybe they were proceeding with a little bit more caution. And maybe it would not have made a big difference in this case. And also that it also underlines the extent to which Anthropic has tried to paint itself as the more responsible company. And while that may be true in some ways, I think in a lot of other ways, it really is just another AI company racing ahead.

• 10:55Speaker 01

  I'm not sure if you guys saw, there was a meme that was going around and it's like a three panel meme, kind of like a comic. And the first one is a researcher. Being like, this is the most dangerous AI yet. It could kill us all. It will destroy all global infrastructure. This can't be allowed to fall into the wrong hands. And then it's Trump saying, okay, it's banned. And then it's back to the researchers saying, you can't do this. So it's like, I kind of love that it's banned in a way. Like, I mean, I think it's, you know, there's all these arguments to be made about how detrimental this is for economies and other startups and other people that are relying on, you know, these models. And you can't just have a government unilaterally get involved and say it's banned. But then also, if you're like, saying how dangerous it is, maybe should the government be allowed to do that? I don't know.

• 11:33Speaker 00

  Well, we should probably move on pretty soon. But just real quick, if anyone has any thoughts, any predictions, like, how long is it actually going to be banned for?

• 11:41Speaker 01

  It's coming back soon.

• 11:43Speaker 02

  Yeah, I mean, I got to imagine in the next few weeks, we'll see something. I mean, my question is, is there, does Anthropic put any more restrictions on Fable 5 than it had already put on, you know, there were plenty of people who said there are already too many restrictions on it.

• 12:01Speaker 01

  Can confirm from my own tests.

• 12:03Speaker 02

  Yeah, yeah. I know. One of the first things you tried, it sort of bounced you off immediately. So

• 12:07Speaker 01

  Immediately. I was like, okay, this is really lame.

• 12:09Speaker 02

  So yeah, I mean, I think it probably comes back at some point. You know, I think, you know, with the IPO looming, I don't think this is necessarily the administration like actively trying to put its thumb on the scale as much as it is just that general disdain, you know, between the two parties here that we've seen. But, you know, that question becomes more relevant as we get closer to Anthropic trying to go public.

• 12:33Speaker 00

  So keeping on the idea of the relationship between the government and tech, we can turn our eyes maybe to the UK for a few minutes where they've essentially unveiled this new social media ban. What do we think about that?

• 12:46Speaker 02

  As the parent of a five year old, I am glad I don't have to think about that much yet. I'll put it that way.

• 12:54Speaker 02

  This is tough, right? Like, we're still reckoning with these questions many years on from these services becoming extremely popular. And I don't think anybody really has a good solution for it yet. You know, I think the two of the like, leading ones that we've seen is something sweeping like this, or, you know, more sort of explicit, like, identification technology, which has a lot of drawbacks, especially in the way that governments try to implement that stuff. So, you know, when I think about it at the highest level, I feel like something like this is maybe the lesser of two evils. I don't know.

• 13:32Speaker 00

  Although I feel like those two things also go hand in hand, that sometimes the way you get like, age verification is through identity verification. So, you're probably gonna see both of those things happening in the UK, or continuing to happen.

• 13:44Speaker 01

  Yeah, well, I'd say, I mean, again, this isn't just the UK, right? Australia did it, like, Canada, France, Denmark, I think they're all thinking about this, or, or they've maybe passed their own rules. I think it, I think it makes sense. I think it's gonna be hard to police, and I'll talk about that in a second, but I think it makes sense because we have learned, like, this is like a tech, big tech's having its big tobacco moment, right? Like, we have learned that this is harmful for people's mental health, which then, you know, has carry-on effects. Their physical health is harmful to society, but like, for kids, it's really, it's harmful, right? And so, why, if you have something, a product where there's so much data on just how harmful it is for them, how addictive it is for them, what do you do? Well, you take it away from them, right? Like, kids

• 14:27Speaker 01

  can't buy cigarettes, right? So, I think that it makes sense as legislation. I'm personally happy to see it. Do I think kids are gonna find ways to get around this? Absolutely. But maybe it will spark a change, so that, you know, we don't see, when I was younger, I definitely saw a few teenagers smoking, but now you don't really see that, right? Like, I think that, I think these, this legislation can make change, and I, I hope that this happens, you know, for this type of legislation.

• 14:55Speaker 00

  Yeah, although I have to say that, you know, at least in New York City, we're, there's like, an element to which it's also cultural, and now you're seeing this upswing in young people smoking again, which is absolutely horrifying.

• 15:06Speaker 01

  Oh, wait, smoking? I'm not saying that.

• 15:08Speaker 00

  Yeah, smoking.

• 15:09Speaker 01

  Are there kids where you live? I never see children.

• 15:12Speaker 00

  Not like babies, but like, I'm a 43-year-old man, so a 25-year-old also looks like a child to me. But I think that the thing about these bans, right, is I do feel genuinely torn in the sense that you're right, that the harms do seem very clear, although I think there's some dispute about the extent, because I think sometimes people will point to, for example, just like sort of general increase in, you know, mental health issues among adolescents. And then, but where there's a lot of debate is that, okay, well, how much of that is because, you know, social media versus the fact that there's all kinds of other reasons why that might be the case.

• 15:53Speaker 00

  And so I think that there's, you know, good reasons to have scrutiny around social media and teens. And at the same time, I understand why some people also kind of push back and say, "Listen, like, you know, we've made life so much more restrictive for teenagers in a lot of ways." And, you know, online is one of the places where they can still gather and connect and build community. And so is this sort of actually like sort of going to exacerbate the problem, or at the very least not solve the problem? And I don't have the answer there, but that's why I feel genuinely torn about it.

• 16:31Speaker 01

  Yeah, the last thing I'll say about that is that, like, you just can't, like, I think the problem has been that you can't control what these tech companies are doing to keep people addicted, especially children. Like, I think a lot of the revelations we heard last year about Instagram and the way that they were targeting young girls, right? Like, how they would, you know, if a young girl posted like a selfie and then took it down, that meant, "Oh, she's feeling insecure, let's hit her with an ad for makeup or something." Like, insane, right? So like, you can't really police the way these companies are approaching teenagers and leveraging the data they have on them to, you know... To keep them addicted. So what do you do? Well, I guess you just have to try to control the kids and parents need to have more control over kids. But to your point, Sean, earlier about age verification, that's going to be the way that this is done.

• 17:16Speaker 01

  And Internet users absolutely hate this, right? Like, they despise having to provide their personal information to have an account on Reddit or something like that, right? So like companies like Persona, did you guys hear about them? They're on, they're working with Character, Roblox, Discord, Reddit. And that's been having massive pushback already because there was a code leak, which confirmed everyone's worst fears about what happens when one company has all of my personal data. Well, it revealed, the code leak revealed that on Persona's backend, they can run all these verification checks to be scanned against like politically exposed person watch lists, global intelligence databases. And you're like, okay, I'm just trying to be on Reddit right now.

• 17:59Speaker 01

  I don't need to, I don't need to be part of these lists.

• 18:03Speaker 02

  I propose, as someone who I think much like both of you, came up at what I think is the perfect time for the spread of computer usage and Internet usage, that everybody should have to start with dial-up and slowly progress their way through an ever-expanding open web.

• 18:23Speaker 01

  Right.

• 18:24Speaker 02

  So that they, A, learn to use these tools in a more stepped way. And B, to have an appreciation for what gets lost when just a few companies wind up owning all of this infrastructure and making these decisions. Because that's roughly how we get into these positions, where you have just a few companies that are making decisions to behave in the way that you just described, Rebecca. And very little that we can do to change that behavior. To me, the comparison to Big Tobacco is interesting because There are a lot of similarities here, maybe at the like highest level, but in the specifics,

• 19:08Speaker 02

  there's so many differences where like it went from regulation to social change because there are a lot of social trade offs with smoking, whether you smelled like smoke and people didn't like that around you, or you know, you were quite literally seeing people that you loved to die in a way that was like more tangible and related to the act of smoking. Whereas here, all that stuff's just a little more nebulous and harder to feel the direct effects of. And, and also it feels so much more harmless in the moment, right? Right. Like that, like, how bad is it for me to just like mindlessly scroll Instagram for three minutes, you know, instead of smoking a cigarette and doing that day after day,

• 19:49Speaker 02

  hour after hour. And so, you know, to me, like, in the ideal world, there would be some legislation, there would be some pushback, there would be some kind of haggling back and forth about how to properly do this. And in that process, like, you know, parents would be able to get a better grip on like how to talk to their kids about it. And then also there would be some social, uh, you know, some social conversation, but I just don't know that that's gonna happen in the same way that we saw with big tobacco. And that like, that's where that uncertainty, uh, you know, has me really worried about where this all can go.

• 20:21Speaker 00

  The other thing I wonder about is what potential there is in terms of, and I don't think these are mutually exclusive. You see some, even in the UK legislation or the UK proposal, this idea that both it's sort of an outright ban for, for kids under 16, but also there's other kinds of like restrictions for, you know, 16 year olds, 17 year olds, um, around things like infinite scroll. And I think a lot, you know, even in the US where I would be shocked if we ever saw an all out social media ban, like you're seeing some proposals around like restricting specific features that maybe can limit the harm. And in some ways that feels like it has a lot of potential, but I also wonder to what extent it's just kind of, you know, a Band-Aid on a much bigger problem.

• 21:05Speaker 01

  Well, does this mean, you know, I mean, if this does take off, I mean, probably not in the US and that will really be the use case, but does it mean that there's? There's been so many other types of social media startups that have like tried to come into existence, but maybe there's a shot for them in the future. Maybe this is a time for builders to start building.

• 21:22Speaker 00

  Yes, and to get back to Sean's point, you know, another like important part of regulation is antitrust, which not to get open up a whole can of worms, but you know, a world where you know-

• 21:31Speaker 01

  Wait, antitrust? I haven't heard of her in a long time.

• 21:35Speaker 00

  A handful of companies control everything that we see on the internet. That would be nice. I think that would be good. But let's move on to some of the deals we want to cover. We'll probably go through all these pretty quickly, but the first one is SpaceX, which we talked, you know, a few weeks ago about this deal where SpaceX could acquire Cursor, and it has now said that it is going to acquire Cursor for $60 billion in stock.

• 21:58Speaker 01

  Speaking of antitrust.

• 22:01Speaker 02

  Yeah.

• 22:02Speaker 00

  Go on.

• 22:02Speaker 02

  It sure is interesting to imagine what kind of antitrust considerations we're going to wind up with SpaceX in a few years, especially with the administration changes. Well, yeah, I mean, that all of that is for another podcast, I think. This was, you know, kind of as expected. I think the more we all sat with the Cursor deal that got announced back in April, it seemed more and more likely that SpaceX was going to sort of subsume it and bring it in-house, especially as XAI got, you know, essentially ripped down to the studs. And Musk said that, you know, it was built the wrong way and they're going to build it from the foundation up now. And it was looked pretty hard to, you know, do that without

• 22:45Speaker 02

  doing something like this and acquiring Cursor. You know, good on SpaceX in the sense that like this is all just being done with $60 billion in stock, and not only $60 billion in stock, but $60 billion in stock post-IPO when it was, you know, when it's at a valuation around $2 trillion or so. I know it's falling today as we're recording it, but certainly a easier pill to swallow for them now than it was even just two months ago when they first started talking about this. You know, but also a great exit for Cursor. Like Cursor was another one of these companies that was doing something really- Well, if not like wholly unique, but doing it better in some ways than a lot of other companies trying to do stuff like this in the AI space.

• 23:28Speaker 02

  And, you know, was in the middle of a funding round when the initial sort of engagement with SpaceX happened. And so, you know, in some ways, this is a pretty great exit for them. To me, like, I'm, I'm always interested in how companies like this or startups like this get integrated into the larger thing, right? Like, I assume that they're all bought into Elon Musk's vision. But that might not always be the case for everybody who's there. And we're already seeing things like, you know, sort of cursor build being renamed rock build and, you know, what is cursor? What made those people want to be there, you know, is already sort of changing a lot. So that's the thing to watch here going forward is, you know, how much this really, you know, is this a rip and replace kind of thing? Or does SpaceX have

• 24:10Speaker 02

  a lot more to do now that cursor is being integrated? And we should say, like, the deal's not fully closed yet. It's going to close in the third quarter of this year.

• 24:18Speaker 01

  Yeah, that's a good point about like, do people want to be working for SpaceX? I mean, I imagine a lot would, but also there are probably going to be some massive liquidation event for, for a lot of people. They might just, you know, take their winnings and move on. But yeah, I, this deal is interesting to me because so much of SpaceX's, you know,

• 24:39Speaker 01

  so much of SpaceX's valuation, like I think 71% or XAI, you know, XAI accounts for 71% of SpaceX valuation, I read, right? And then like a solid chunk, like 90% of SpaceX's TAM is AI related, which is kind of funny when you think about it, because if you look at XAI's, you know, user numbers, they're, you know, paltry compare and revenue growth, they're nothing compared to what OpenAI and Anthropic are doing. So it's like, this is a $26 trillion AI-related TAM they're pitching investors. So how are they doing that? Like, how does cursor actually help with that?

• 25:15Speaker 00

  Yeah, I feel like there's this, this tremendous cognitive dissonance. And maybe that's just always true when we talk about any business created by Elon Musk, but where they've like, it's, you know, the biggest IPO in history by like any sort of like, External marker, it seems like Wall Street is embracing this like a moment for celebration. And yet, when we talk about AI, which is, it seems like the most important market for SpaceX, it still very much feels like they're the underdog and presumably Cursor will help with that, but, but it still does not feel like they're in any sense like the leader or in a clear position to become the leader anytime soon.

• 25:54Speaker 02

  Yeah, the way that I've been thinking about this that has clarified things for me is that like, the gap between what's being promised and what is actually happening in the business is so much greater here than it is at Tesla. At Tesla, you can argue whether or not you think they're going to make a network of robotaxis or, you know, an army of humanoid robots that make any kind of economic sense, all day, for sure. But the gap between what the business is now and those things is substantially smaller than what we see at SpaceX. And so like, that to me is like, once we get past this initial hype of the IPO, and they get included in indexes and, you know, some of the buying pressure stops, I'm super curious to see, you know, like, how does this company present

• 26:39Speaker 02

  itself to investors every quarter, every year? Because it's just a, it's a yawning gap between like, what they're going to have to talk about, which is going to be some launches, but the launch business is sort of pales in comparison to what they're promising, like you said, Rebecca. And, you know, the more meaningful things they're going to have to discuss on every earnings call is going to be like, we maybe rented some more compute, which like, yeah, that works in the near term, you know, but like, is that enough to build? And, you know, caveat all this, that like, a lot of this clearly doesn't matter to most people who want to, you know, own SpaceX stock. But I just, I think it's important to think about how big that gap is.

• 27:17Speaker 01

  Yeah, I mean, the gap is where Elon lives, right? Like, I joke a lot that his, like, Tesla's always been a meme stock, like SpaceX is a meme stock, right? Like, this is, this is not based on fundamental numbers here. This is based on like, wishes and hopes and promises and timelines that, that Elon never meets. You know, I think that the, the cursor stuff, like the what I could say about Cursor and why it might make sense, a little bit, I mean, okay, so here's why it might make sense, right? Cursor does not make their own models, right? That they're kind of a rapper, but like, a really great rapper. They have the distribution, they've got a great product, they've got a lot of data. Right? It could be useful to use internally. SpaceX can use this. There's some high quality, you know, human reasoning data. They see what devs write. They ask which solutions work.

• 28:02Speaker 01

  I'm curious if OpenAI and Anthropic will keep selling their models to Cursor. That's just like a side note, something I'm going to be paying attention to. And Cursor has reportedly been building and investing its own models. However, even if all of that gets absorbed and, you know, SpaceX, XAI, whatever, creates their own great coding model, like who's jumping from Claude to Mecha Hitler? Right? Like, you've got a bad, like, rep, you know? And a lot of enterprises are not going to want to mess. I mean, I guess it's coding, so it's not like the same. But still, I feel like it'd be hard pressed to get people to move at this point. You don't have the first mover advantage.

• 28:39Speaker 00

  Let's move on to our next deal, which is also an AI deal. It is our old friend Jeff Bezos, who has raised $12 billion for his new startup Prometheus.

• 28:50Speaker 02

  That's a, you know, it's a lot of money.

• 28:54Speaker 01

  Pocket change for old Bezos.

• 28:56Speaker 02

  I mean, we've seen, there's a lot of money flowing into AI, but clearly a lot of money flowing into companies positioning themselves as like, quote unquote, "physical AI." This is one of them, $41 billion valuation. And that's a lot of money for a company we really don't know much about. I mean, like Rebecca, you poke at this stuff more than maybe I do. I'm curious, sort of, you know, what do you think about from what you've read about Prometheus? And how we should even think about them? Because to me, it's been, you know, I mean, this is something that he's like co-CEO of. Like Bezos is like clearly more involved in this than he is with like Slate Auto or like his other sort of investments. But I don't really know what to make of this company yet.

• 29:36Speaker 01

  Well, I mean, there's not a lot of information about what they're doing, right? Like, this is just their second fundraise. They came out with around late last year, $6.2 billion. They're building an artificial engineer that's meant to automate design and manufacturing for complex physical systems. So like jet engines to drug compounds. That seems like a huge ask. I don't know what foundational model you've got that's, you're going to be able to like, fine tune for those little, those, you know, those incredibly different, very complex things. I think the thing that we find the most when we're trying to like automate and engineer or something like that is, one, that's compute intensive as hell, because you're not just building one agent that does that. You're building like many agents that go into one. Like, you know, task, right? So it's like a task that gets broken up into like many different agents.

• 30:21Speaker 01

  And then also, I'm wondering, you know, where is all that training data coming from to build jet? Like, how do you? This is the problem and like Tiffany Luck from NEA, I said Mia before, I always do that. That we talked about this this week. She says it's like the like to 100X yourself, right? You have to spend so much time knowing your own processes, knowing how your own mind works. Understanding like building AI systems. And I think that's just like the hardest problem when it comes to building these types of, quote, "artificial engineers," right? Like, I don't know how you're going to get there super fast, if at all. Is this? Yeah.

• 31:00Speaker 00

  Well, you were talking about this a couple of weeks ago in the context of aerospace engineering, but building something like a jet engine is also an area where you really don't want to screw it up. Like, if you make mistakes, hallucinations, things like that, that could go really wrong. That's something, not something you want to fix in the, you know, you don't want to discover mid-flight that you that you've messed it up. So I'm very curious to see, you know, how AI gets deployed in those fields where the engineering is like very, very sensitive to any kind of mistake.

• 31:31Speaker 02

  And requires a level of durability. Yeah. I mean, interesting timing here that this is what we're learning. Prometheus is working on as jet basis space company is dealing with the fact that one of its rockets exploded on the launch pad a few weeks ago, you know, really setting them back at the very least a number of months as they try to support NASA in its quest to get back to the moon.

• 31:52Speaker 01

  Yeah. And it's also interesting because Bezos is like, you know, kind of against the AI job loss narrative, right? And he's talking about replacing engineers with artificial engineers. And even though he's laid off, you know, thousands, tens of thousands of workers. But with that, I think we are just about out of time. Thanks to our audience for listening to us yap for the last half an hour. Love it to spend this time with you. Equity will be back next week. Until then, you can find Equity under the handle at EquityPod on X and Threads. Bye. Equity is hosted by TechCrunch senior reporters and produced by Teresa Locan Solo with editing by Cal. Subscribe on YouTube or wherever you get your podcasts and find out what's next at techcrunch.com/events. Thanks so much for listening and we'll talk to you next time.