AI labs want to pump the brakes, but Amazon and SpaceX are still blasting off

0:00This episode is brought to you by Indeed. You're ready to level up your business, but the pressure to find the right person to make that happen is intense. Indeed Sponsored Jobs can help carry the load. It boosts your job post to help it reach quality talent fast. Cut through the chaos with Indeed Sponsored Jobs. Get a$75 sponsored job credit at indeed.com slash podcast. Terms and conditions apply. Hiring now? Then this is a job for Indeed Sponsored Jobs. After years of pushing full speed ahead on AI, OpenAI CEO Sam Altman is now saying maybe we should pump the brakes. Also this week, SpaceX and Amazon are racing to beam satellite service straight to your mobile phone.

0:45And we're keeping our eyes on a couple of deals, including one startup that just raised$9 million because of its AI detection tools. Stick around.

0:59Hello and welcome back to Equity, TechCrunch's flagship podcast about the business of startups. Today is Friday, July 31st. I'm Girsten Goracek, Transportation Editor here at TechCrunch, and I'm joined as always by our weekend editor, Anthony Ha, and Senior Reporter Special Projects, Sean O. Kane. And I have to say, Anthony, it looks like a little bit of a different backdrop. I'm going to say it's giving me beach vibes. Are you at the beach?

1:29Mary Ann Azevedo: I am at a friend's, he's just renting this house for the week near the Jersey Shore. And I've had a very nice couple of days on the beach. I may even look slightly more tanned than usual. Don't answer that. I'm ready to dive back into the world of tech news. How has your morning been? Yeah, so this morning I spent a very early morning, because I'm on the West Coast, writing about Zoox getting its final exemption, which is a big deal because now they can finally launch a commercial robotaxi service with those toaster-like robotaxis that you might have seen in Las Vegas or San Francisco. So that's how I spent my morning definitely not tanning on the beach.

2:15Mary Ann Azevedo: I'm okay if you want to feel a little bit jealous, but we have a lot to get to. The first one, as you mentioned at the intro, is some comments from Sam Altman, which I think tie into this broader discussion of kind of where we are in this AI moment. And basically what Altman said on this podcast was essentially seems to be embracing a decelerationist message right now. Sean, what did you make of that? I mean, I'm going to pull out my favorite segment and talk about Sean doing a victory lap, because it wasn't that long ago that I was talking on this podcast about why we aren't just slowing down with this stuff a little bit and there are myriad reasons why that's the case.

2:58But maybe we've finally hit an inflection point here. I think a big driver of this has to be what we talked about last week with one of OpenAI's models breaking into HuggingFace's data and apparently breaching a few other things around the internet as well. We should couch this. He's not like calling for, you know, a pause like we've seen some people in the tech industry try to do in the past. He was, I guess, very careful with his words and, you know, saying pace it. And so I will see how this holds. I mean, any caution that we see some of these labs throw out there often gets reversed when the incentives push them forward to resume sort of like full speed ahead.

3:39So I remain skeptical. Big surprise. That sounds so cynical. I think he really means it. Now, I will say this. He might have been careful with his words, but OpenAI Ananthropic did sign a petition that does reflect what he did talk about. And I do think, and I do agree with you, I think a lot of this was very much triggered by a hugging face. He had a quote in the story that we covered, really calling this viscerally felt. I think it probably spooked him and a lot of other people and certainly a lot of people in the industry. The hard thing I think here is how do you thread the needle or how does OpenAI thread the needle of continuing to generate revenue, raise money or have a successful IPO and quote unquote like pace development.

4:26So show progress and pace. And I don't know if they can do that. I'll be curious to see how they how they manage both.

4:34Mary Ann Azevedo: One of the things I've been wrestling with is also this question of is sort of acceleration, deceleration the right framework to be thinking about this? Because it, I think, kind of suggests that there's sort of only one path and we're all kind of stuck on this path. But in as much as we get to decide at all is sort of like, do we speed up or do we slow down as opposed to, again, I'm going to really torture this metaphor. But, you know, do we build different kind of guardrails? Do we choose different paths? And so on some level, I'm just very resistant to this framework as opposed to saying, okay, if we're not happy about what models are doing right now, what else can we do?

5:13Mary Ann Azevedo: Is sort of a slowdown, a pause, just a stoppage the only option? And I don't think it is. And I do think that it has become, again, as with so many other discussions about AI, become this site of all these different kinds of debates. And I mean, one thing that I did want to emphasize again, because it's been really interesting to see how the level of alarm around this, which I think a lot of it, you know, these are, again, real concerns and this sense of, oh, my gosh, like, what if we have these autonomous agents and models just running around, hacking each other, trying to stop, you know, prevent hacks.

5:47Mary Ann Azevedo: And it's sort of just all getting out of our control, leading to all these like broader debates about kind of alignment that Rebecca Boulin did a great piece about. But it's worth coming back to, I think, one of the points that we also wrote about at TechCrunch, that this specific hack, yes, it was caused by an open AI model, but it also comes from the fact that it sounds like they just didn't secure the testing site properly. that in theory, this model should just not have been able to get online. And that didn't happen. Now, of course, if you have sort of a powerful, misaligned AI, the risks of that human error, I guess, go up dramatically.

6:25Mary Ann Azevedo: But it does start from just the fact that they kind of didn't secure things the way they should have. I think that's right. I think your point is well taken in the sense of like, we shouldn't only think about this in some linear fashion and whether things are accelerating or decelerating. There's a lot, I think, that could and should be said about just how responsible these companies are being in both parties here. I mean, Lorenzo, one of our colleagues, also wrote a really good piece sort of walking through how serious some of the security researchers who pay attention to this stuff think that the hack really was.

7:00And it really does seem like on both sides of this hack, there were steps that probably should have been taken that would have prevented it. One of the things that I found most interesting in that story was that some of the researchers were pointing out that, you know, what this model did was not some new advanced thing. It was really very human in the way that it thought about trying to break into, trying not to anthropomorphize, but the way that it thought about breaking into Hugging Face and that it was also very loud and messy and sort of it wasn't really trying to hide its tracks. It was more like Nixon's people breaking into Watergate than some real stealthy cyber op because it didn't need to be and it wasn't instructed to be.

7:43And so that should have been more easily preventable. And hopefully this is a sign that these companies will take this forward and be more careful about that stuff. I will say one other thing on the sort of like Excel and Decel side. I don't know if this is the motivation, but I think it's smart. You mentioned the IPO, Kirsten. And I think it's smart of Altman to be able to push this advantage that they have now, which is that they're not going to markets next month or two months from now. He's even floated the idea of going, you know, in 2027 and that they only filed their confidential filing, you know, so that they have the option ready when they're ready.

8:18So if you believe all of that, he has the ability to talk this talk in a way that Anthropic can't because Anthropic's already in conversation with a lot of the bankers and is, you know, headed towards a more near term IPO and is, you know, therefore more restricted in what it can say and how it should be saying it and how the market is going to react to that. Okay, one pushback on that. I think Anthropik's been saying what they kind of want to say. I mean, I would maybe counter that they have, in a lot of respects, disregarded probably the smarter, more diplomatic move and have been pretty voicey.

8:53But I actually want to go back to Anthony's point, which is the only choices are stopping or going full speed ahead. And so my question for both of you is, so what is the right approach? Anthony? If there are these two options and you're saying that maybe AI labs should be, and the industry as a whole shouldn't just think of it in this way, what should they be thinking of?

9:19Mary Ann Azevedo: Well, I think there are a few different ways to approach it. I'm going to say right now that I'm not going to have some grand solution. Solve all of our problems, Anthony. No pressure. This is the new way everyone should think about AI and guardrails and alignment. But I think that certainly one of the things that Sean touched on was this idea of responsibility. And so it's less a question of how fast or slow do we go, but who's responsible when things go wrong? Who has incentives to actually make sure things don't go wrong? I think also there was some interesting posting on social media over the weekend from the CEO of Hugging Face where he went to visit OpenAI and basically said his kind of big suggestion is radical transparency transparency, where OpenAI releases a lot more information about what happened, which I mean, I know they have released a report since then.

10:12Mary Ann Azevedo: And on top of that, basically, they commit a lot of resources to allowing a variety of researchers, including independent researchers, to figure out how to create defenses against this kind of hack. And so again, it's not about sort of fast or slow, but okay, if we're going to commit a lot of resources to building powerful models that could potentially do these kinds of cyber attacks. The same company should commit the same amount of resources to figuring out how to defend against those attacks. Yeah, I mean, I agree with a lot of that. And I am pleased to see the Hug and Face is treating it that way.

10:48I think the question, you know, if we had down this path, or we consider this stuff more, these tech companies consider this stuff more, the question has to be responsibility to whom, right? Is it going to be responsibility to their backers, or their eventual public shareholders? Or is it really going to be responsibility to like the world where all of us could feel the impact of these things going in bad directions. And, you know, I just, unfortunately, I think it's fair to say we are in a moment where regulatory frameworks and oversight are not really something we can, especially at the federal level that we can fall back on for those things.

11:25The other thing I'll make a point before we kind of start to wrap this up is like the wild card here is Elon Musk, as always, right? I mean, Grok and XAI and SpaceX in general is kind of behind these other companies in a lot of different ways, but they bought Cursor. They're sort of building on what Cursor had built. And if you pay attention in between all the racist memes and stuff on X, a lot of the things that Elon Musk is sort of copping to now is just basically being an accelerationist. And he basically said it in the Economist interview he did a few days ago. He has 180'd from his cautious position as OpenAI founder a decade ago.

12:06And should he find the ability to turn those products that he has under his auspices into something that is powerful and really competitive and widely adopted when compared to OpenAI and Anthropic, that could just throw a lot of these efforts out of whack. If there really is an effort to make these things responsible and treat them responsibly and make sure stuff like this doesn't happen again, and in a worse way, you know, he could put a lot of a big thumb on the scale to mess with those incentives. I'm glad that you brought up Elon Musk, because it just isn't the same to have an equity conversation without him.

12:43But to your point about his changing position, he used to have a very different kind of rhetoric around the pace and development of AI. And that has flipped. And at the same time, the responsibility question, then also the next question is, is it voluntary responsibility or is it legislated? And if we go down the road of legislating this responsibility, Elon Musk is going to have a lot to say about that because he has, you know, generally speaking, not liked, again, to go back to this metaphor of guardrails around his other businesses. That is going to be a really interesting fight. It's not just about like, oh, sure, we should take responsibility.

13:28Great. But should it be voluntary? Like, for instance, how a lot of the autonomous vehicle on the federal level is handled? It's like voluntary reporting, things like that. Or is it something where it's really a mandate? And it is interesting to me that Sam Altman is talking about pumping the brakes, and it It clarifies and crystallizes how far ahead the AI industry is from any law or national framework on how to even handle this tech.

13:58Mary Ann Azevedo: Well, that also brings up something that our colleague Amanda wrote this week as well about thinking about sort of the distance between the AI industry, whether those are the people who are kind of the accelerationists, who are just very evangelical for the technology. or those who are sort of the doomers as well, I think all of that can sometimes feel kind of disconnected from the everyday experience of people using AI or not using AI. And all of which is to say, Amanda did this great piece about librarians discovering that increasingly there seems to be a lot of popularity for these workshops, giving people the option to kind of get AI out of their technology.

14:42Mary Ann Azevedo: How do you actually remove AI results from your Google search? How do you turn off AI features on your smartphone? And, you know, like a lot of these kinds of stories, they're talking about like relatively small groups of people. These are workshops with, say, dozens of people, maybe slightly more than that. In a few different states, we're not talking about a mass consumer movement. At least that's not what this story is about. But I think if you think about that in the context also of, you know, polling numbers in general showing that particularly Americans really, really are not excited about AI, about the kind of future that AI may be ushering in, it's important to keep that in mind as we talk about acceleration versus deceleration or a Sam Altman one versus Elon Musk one is that it's not entirely up to them.

15:26Mary Ann Azevedo: They can very much push us towards these different futures, but individual consumers can also say, well, I'm not so interested in that. I mean, our power as one person is limited, but if millions of people are like, you know what, this sucks, which I think you see some evidence of that happening, that actually does shape the direction of these technologies. Well, there's nothing more powerful than a cultural movement that may start in a very niche, and it may start with a real crystallized black and white feeling about whether AI is good or bad. But when it then seeps into the rest of culture and people just adopt ways of life that they are not really sure where it started, that's when it becomes really powerful.

16:09And I kind of wonder if that will happen in our culture and specifically with younger generations and how they use AI and whether these, like what's happening with the librarians, workshop, if these are just going to be little small niche things, or if it will actually end up kind of seeping throughout the culture in a way in which the way certain generations use AI is very different than it is right now. And I don't think we know the answer to that question. The one thing we do know is that people are still investing in AI. And specifically, there is a scoop that Marina Temkin had this week about a company called Prentice.

16:47And not only is this a new AI startup, this is a new AI lab, which is kind of incredible, that was co-founded by Reid Hoffman and Mark Pincus. And apparently they're in talks to raise$100 million on a$1 billion valuation. So people are still putting money in it, even if they're not really sure where it's headed, even if they're pumping the brakes a little bit, and even if they're, you know, trying to find weird niche ways to avoid it. Wow, that's a pretty lofty valuation for a company that we really haven't heard much about yet. And speaking of lofting things into orbit, how do you like that one?

17:26We should talk a little bit more about space. In particular, there was a story this week that I think sums up a lot of what's been happening.

18:00Then this is a job for Indeed Sponsor Jobs. Among some of these companies, and that was Amazon filing to basically create a new satellite network that would connect to mobile phones. They want to build as many as and send up as many as like a little more than 5 ,000 satellites starting in 2028 to create a network. And this is the latest in a bunch of these kinds of announcements that we've seen. Blue Origin is building its own different kinds of satellite networks. Amazon already has one called Leo that it's been building for a little while. And then there's SpaceX with Starlink, obviously. So I'm curious what you guys think of this, because I still think of satellite, mobile satellite networks as things that happen on the fringes that are good in a pinch, but not something you rely on.

18:51But we're seeing a ton of interest and a ton of push into the space from some of these major companies. And I'm still trying to figure out why.

18:59Mary Ann Azevedo: I had the same question. And so I'm going to do the classic thing where I answer your question with a question. But I also wondered about to what extent, in some ways it reminded me of trying to keep track of all the data center announcements and that there's clearly a lot of money flowing into it, but also an element of, hey, there are all these high-flying deals and who knows how many of them are going to become a reality. To what extent does this reflect the reality on the ground now? And I know that certainly with SpaceX, that this isn't just a future fantasy. This is something that to an extent already exists.

19:31Mary Ann Azevedo: But I'm curious for companies that aren't SpaceX, is this something that they're saying, well, we'd like to do this? Or have they already kind of started to put satellites up there and started to create these networks? The thing that maybe makes the most sense to me is we're talking about companies. Amazon obviously has a close relationship with Blue Origin because they're both founded by Jeff Bezos. SpaceX has Starlink. there's really nobody else at the moment, new entrants at least, who can do this kind of thing. And so this is, in my view, at least as much a move of opportunity as it is one of necessity.

20:09Because if they don't do it, someone could come along and try to do this a little bit down the road. And clearly there are existing companies that offer stuff like this. We see SpaceX is buying a bunch of Spectrum from EchoStar. And, you know, they have a deal with T-Mobile to provide their service, you know, sort of in that direction. There's been rumors about SpaceX buying T-Mobile. And so I think it's really another example in some ways, honestly, of a Bezos company following what Elon Musk's company was doing, where Musk was doing this to build a business for himself and take advantage of the opportunity that they have at SpaceX of being kind of the leading launch provider around the world and the sort of synergy that he gets out of that by, you know, doing business with himself and, you know, Amazon sort of following that same path.

20:57And so I think that's probably a big driver of this. Yeah, it seems protective in nature, which Elon Musk is actually pretty good at doing, which is he'll take big, big bets far out before there is really demand for something. And that has worked out for him in the past. And this feels like that as well. I mean, I guess my question for you is, is this, you know, infrastructure before demand? I believe it was the T-Mobile CEO, I think it was like something like two thousandths of a percent of its total network usage is using satellites and most of it's in national parks. So is this a question of infrastructure before demand?

21:40Or are we seeing hints of demand? Like when you're following this, do you actually see hints of demand happening? Specifically for mobile connectivity? I don't think so. And not at the moment. There must be some calculation that these companies are running of, you know, it may be this infinitesimal amount of usage now. But you know, across the globe, in every country, you know those little tiny percentages add up to something that makes sense if you're able to become the provider another thing that i think people should consider when they think about this stuff and i've said this a bunch especially relative to spacex on this show if you can come up with even the slightest justification for something that needs to be launched into space and you have a business whether it's you know in the same company or next to the company that you're running, you're creating a revenue stream for that launch company, right?

22:34Like a large portion of what SpaceX is trying to do now, and arguably even in the future with the sort of data centers and orbit idea, is you come up with a project that requires your launch company to be able to offer that launch service to them. And you're creating a revenue stream for it. And so I think there's probably an element of that playing here. The only caveat to that at the moment is that Amazon's kind of stuck in a spot right now with its LEO satellites where it's turning to other providers to launch them into space because its new Glenn rocket, while it looked better than I think a lot of people expected over the first few launches starting in early 2025, blew up on the launch pad a few months ago in one of the largest explosions that has ever happened in the space industry.

23:26And while they say that they're making very good progress and they're committing to getting that rocket back off the ground by the end of this year, which would be an insane turnaround, it's just something to think about now where maybe this is partially driven by the idea of creating business for Blue Origin, but at the moment it certainly isn't.

23:48Mary Ann Azevedo: One thing I want to mention, partly just to undercut the idea that Kirsten is the only outdoorsy person on this podcast, is that I did have a moment a couple years ago where I could see the need for this, where my partner and I, we were driving into a national park in Maine. We completely lost cell service, which we were warned that that would happen. And it was just, you know, like you're driving, you're like, did we miss the turn? Where are we? You know, like navigation basically just goes down and you're just sort of thrown back on paper maps. And it was like a very harrowing experience. So like I can see the appeal of like, oh no, you get connectivity in those moments.

24:24Mary Ann Azevedo: At the same time, of course, that's like the only time in the last however many years that that's happened to me. So it's hard to see, you know, that that's something I would pay like an ongoing fee for. And so the explanation, Sean, that I think you've offered repeatedly on this podcast, which I think has been very clarifying, is just launch company says we need to do more launches is a pretty good explanation for why they're talking about this. Yeah, that is actually a really good explanation. And I will say that we need to work on our map and compass work next time we're together, Anthony, so that next time that happens.

24:58I will say, though, to the point of the need for satellite mobile, and I'm going to throw in a humble brag in here for a minute, but I was on a backpacking trip earlier this summer and in an area where I never would have been able to have cell service, and I did. And so those pockets of the world in which you don't have that, I think, has diminished greatly. And so then you're like, why do I need this satellite to mobile service? But I want to talk about, if you'll let me, getting into some other deals. One is, I think, kind of close to my heart because we're always complaining on this show and also sometimes just personally between us about AI slop and AI content.

25:43And this is a company that's apparently developed some tools to spot or detect AI content. And the company is called Pangram and they just raised$9 million. I don't know about you two, but I'm excited about them succeeding.

25:59Mary Ann Azevedo: Yeah, this is a company I've heard a lot about in the last few months, maybe the last year, because I think they seem to be the most cited of these various AI detection tools. I think Substack actually just announced that they're also going to be using some pangram tools to identify which newsletters are written with AI, which I'd be very interested to see how that goes and who gets mad about that. I mean, I think it's definitely like a very interesting area with a lot of potential. Obviously, it ties into a lot of the different things we were talking about earlier in the show. And I also couldn't help noticing that, and again, this is not an AI lab, it's not a frontier lab, but it's, you know, doing work that you would expect the opportunity to sort of increase as AI usage increases.

26:44Mary Ann Azevedo: And yet, I mean, the funding that we're talking about here is so small compared to pretty much any other AI startup where we talk about like$9 million. I feel like I'm back in 2015, you know, like that, that seems like a respectable first round back then. But now that's like, you know, nothing. Yeah, not everybody is a hyperscaler, right? Although, you know, you mentioned sort of their relationship with the leading labs. I think there really is an interesting dynamic here about how they're going to keep pace with those new models as a smaller company. By all accounts, their tech so far is actually pretty good, you know, better than attempts we've seen in the past.

27:25and I think it's something along the lines of like one in 10 ,000 of their evaluations they get wrong as far as properly labeling AI which you know is pretty good can they keep up that pace as the models get more advanced and start maybe weeding out some of the telltale signs of AI writing I don't know the other thing I'll point out is that you know I think it's easy to get lost in numbers all the time and one in 10 ,000 is a good hit rate in some ways but also if if you're doing these millions and millions of times, it's going to get stuff wrong. And I think we're already seeing a bunch of pushback from people who are hardcore sub-stackers who don't like the idea of getting this kind of witch hunt dynamic thrown into the mix.

28:08There's also this week LinkedIn added a new option to label things that you think is AI slop, which I'm sure that's going to have an even wider and louder reaction from the most LinkedIn maxed people. To me, it's really a question of these dynamics. How do people start to accept this? And it's more of a social question than it is about whether or not this tech is super accurate.

28:33Mary Ann Azevedo: Yeah, there was this really interesting piece in The Atlantic a few months ago. So even before these most recent announcements, the title was America Has a Pangram Problem, which is a little bit hyperbolic when I think most people have no idea what Pangram is, but does speak to some of those broader social questions you're talking about. Sean. And I think one of the most interesting points in it was this idea that, yes, like these AI detection tools have gotten a lot better over, you know, the last year or so. And in fact, in the funding story about it, it was really interesting because, you know, we actually did test.

29:07Mary Ann Azevedo: I think it gets pretty granular about what are the things that Pangram seems to be able to detect and things where it's still giving you wrong answers. But the argument in the Atlantic piece that essentially it's trickier once these tools are pretty good, because then you do start to accept the answers more. And I mean, obviously on some level, that's what the tools are for, but you still can't accept them a hundred percent. And so you still need to have some humility. You need to, you know, to use the same phrase that you use, I think is sometimes fair is like this kind of witch hunt mentality of just as soon as you run somebody, especially somebody you don't like, they're writing through Pan Graham and it says some, suggests some level of AI contamination then you're immediately like, oh, this person is just producing slop.

29:49Mary Ann Azevedo: And I don't think that's exactly the right way to use these tools. I want to know if they used that Mateo Wong story in their pitch deck. I feel like that's as much as it might have felt hyperbolic, were they looking at investors and saying, hey, see, look, we're a problem. You've got to come in and back us. We're making impact. We've got one more deal to talk about this week. And this one was maybe not something I was expecting, even though I follow this stuff relatively closely, but DoorDash is building its own drones and its own drone delivery business. Kirsten, what was going on with this?

30:23Well, first of all, I 100 % believe that it was in their pitch deck and they should contact us and let us know. And yeah, DoorDash is in the drone business. So to be clear, they have been working with other drone companies, Flytrex and Wing, which is the Alphabet drone delivery company. And they've been doing that for a couple of years. This is a different, very specifically in-house effort, and this is coming through their R &D team. It's called DoorDash Labs. And this is the same team that gave us, and we've talked about on the show, an autonomous sidewalk delivery robot. This is an autonomy and robotics-focused team.

31:01They still have a long way to go. They just received a very important permit, however, that allows them to be a drone operator, the eighth here in the U.S. they have not gone through the certification process for having a fully autonomous drone that you don't have to have visual line of sight which is like a whole other thing. But yeah, I mean this is really remarkable in some ways makes a lot of sense but in other ways it's like why take this on when there are other companies doing this?

31:34Mary Ann Azevedo: Yeah, we've talked about some of these other drone delivery companies and technologies in the past and the impression that I've gotten from those discussions is that often it's not so much about just replacing the delivery options that exist now, but sort of reaching areas or customers that you can't get through, you know, the e-bike delivery service in urban areas. But that seems like it's relatively a smaller market and you have to sort of spend a lot of time and resources in order to actually like make that work in an effective way or some degree of reliability. And so I'm curious to what extent is this, it seems like DoorDash is talking about this mostly as kind of like an experimental initiative versus a major, you know, strategic investment right now.

32:21Mary Ann Azevedo: Is that right? I don't know. I mean, it's easy to put it under this is an experimentation, but I don't think that they would have been working on this for as long as they would have. And by the way, committed to doing in-house if this was just a pilot and a sort of fanciful exercise. I do think that since they've moved past just partnering, which, by the way, they say that they're going to maintain those partnerships with other companies and have committed real dollars to this, I think that they want to see if it'll work. The question is, will it? And I think that right now they're going through something that Uber went through on the ride hailing side, which is how do we grow our business?

33:02How do we be everywhere? We're already the largest, I believe, in the U.S. DoorDash is already the largest on-demand delivery service. How do you grow? Well, do you do it by offering drones and sidewalk bots? I guess you do that. And do you want to own the entire stack? Which to me is really the interesting part here. They clearly want to own all the software and the hardware and really control which mode is used to deliver your burrito. Yeah, I'm just glad we've finally arrived at the eventual future of being able to have nine different robots race to deliver you a burrito in every type of mode.

33:41One point to what you just said that I think is interesting to think about is this is happening at the same time that we are seeing a pretty now public split between Uber and Waymo happening. And that's happening because Uber and Waymo were both going with this like heavy partnership approach. And, you know, part of me wonders if Waymo is thinking about whether it should have been more vertically integrated and sort of full stack, or even if it isn't, you know, I think a DoorDash or companies like it could look at stuff like that happening and think, boy, if like those big companies can't play nice together, this is validating for us that we're going in this direction.

34:23Mary Ann Azevedo: Yeah, I mean, listening to it also just reminds me of, you know, how in the early days of DoorDash, when they're just coming up, like how much they talked about. And of course, you know, of take this with a grain of salt because it's startup marketing, like how much their real focus was on essentially like the logistics side of things and just being as sort of fast and efficient as possible. Now, is that actually why they've become the biggest delivery startup right now? I'm not totally sure, but I can see them sort of trying to maybe reclaim that in some of these new areas. But we'll see how it develops.

34:55Mary Ann Azevedo: In the meantime, of course, we are out of time. Next week, we're not going to be doing this news roundup. Instead, we're going to have an interview I did with the historian Jill Lepore about her book, The Rise and Fall of the Artificial State. She had a lot of pretty interesting, critical thoughts about how big tech companies are affecting our democracy. So stay tuned for that. And of course, until then, you can follow us at EquityPod on X and Threads.

35:27your podcasts and find out what's next at techcrunch.com slash events. Thanks so much for listening and we'll talk to you next time.