Section Insights
Introduction and Guest Welcome
Who is the guest and what have they been up to?
Aaron Delp, a returning guest from Witness Protection and formerly the Cloudcast, discusses his recent experiences, including watching the World Cup in Paris.
- The hosts have not spoken in a while, indicating a personal connection.
- Aaron has been following the World Cup, particularly enjoying the atmosphere in Paris during the France game.
- The conversation sets a casual tone for the episode.
AI Company Dynamics
What are the current dynamics among leading AI companies?
The discussion highlights that major AI companies like Anthropic and OpenAI are not significantly different from each other, and the market is becoming uncertain about which will dominate.
- AI companies have substantial financial resources, allowing them to recover from mistakes.
- There is a noticeable shift in market confidence, with less certainty about which AI model will prevail.
- Talent movement between companies is common, affecting competition.
IPO and Government Involvement
What is the relationship between AI companies and government regulations?
AI companies are navigating the balance between driving innovation and adhering to government security regulations, with some nearing IPO.
- The need for innovation in AI is tempered by national security concerns.
- Companies are seeking to raise capital through IPOs while managing regulatory scrutiny.
- The landscape of AI development is influenced by governmental policies.
Market Fragility and Company Stability
How stable are AI companies in the current market?
The conversation reveals that AI companies are fragile, with key personnel easily swayed, leading to rapid changes in company fortunes.
- Recent mergers and acquisitions can quickly alter the landscape of AI companies.
- The market is characterized by volatility, with companies rising and falling based on team dynamics.
- The fragility of talent in the AI sector can impact innovation and company viability.
Talent and Company Origins
What factors contribute to the success of AI companies?
The discussion emphasizes the importance of foundational talent and experience in AI, suggesting that successful companies often stem from established industry leaders.
- The background of founding teams plays a crucial role in the success of AI startups.
- There is a perception that without experienced leadership, new companies may struggle.
- The analogy of sourdough bread making illustrates the challenges of replicating success without proper knowledge.
Transcript
0:00 4 3 2 1 countdown. I don't even know if it does that anymore. Guess not. All right, we're recording. All right, we'll edit it out. It's fine. 3 2 1 and we're back. And some of you, if you're watching this on video, might be going, "Hey, who is that guy to your left or your right?" You know, joining us from Witness Protection, from a tour of the world, from you know, from formerly the Cloudcast, Aaron Delp, welcome back to the show, man.
0:26 Good to have you on. >> It's been a little while. Oh, it's been gosh, weeks, months. I don't even know what to call this. It hasn't been years. >> No, it hasn't been years, but >> it's been a little bit. >> Yeah, it's good to good to see everyone. Good to be seen again. It's good to see you. Like, we haven't hardly talked at all either recently. We both were like before we record, we're like, "What you been up to, man?" Because we haven't talked in forever. It's been crazy. So, >> yeah. Yeah. Well, before we jump into all sorts of AI news, you've been watching the World Cup or keeping up with that at all.
0:58 >> Yeah. Yeah. Been been watching some games. I mean, here and there when I can, the one that was most interesting, for those that don't know, I work for an AI company headquartered in Paris. And I was in I was in Paris last Friday night for the the France game. >> Okay. >> And pretty crazy. Pretty fun. Like just kind of walking around and all the cafes are just absolutely freaking packed with people wearing jerseys. Is a certainly a fun time.
1:21 >> Yeah. to be in Paris for sure. >> Yeah, I haven't been I I feel like I've had a game on every at least every other night or so. Like it'll just be on. I haven't I don't really have a rooting interest. My wife is deeply deeply following all sorts of the the human interest stories. So like she was, you know, almost almost went up to Boston on a whim to go hang out with the the Tartan Army, the Scottish guys. And so she's she's deeply involved with anything having to do with like just human interest stories around it. So but yeah, we'll see. I think we're recording this right around the end of June. So I think by the time we record next time, I think I don't know how long the whole thing lasts, but we'll probably have a World Cup champion. It seems like right now the only the only two real big story lines I've sort of seen are like the African countries are doing really well and then all of the you know all of the the you know the company the countries you expected to to get through to the next round have all gotten through. So there haven't been any really big upsets at this point.
2:12 >> Yeah. No no big surprises so far. >> yeah. So all right. Well, cool. Well, let's let's dig into, some AI news of the month. The, you know, I I I looked hard high and low. I looked under rocks, under crevices. I could not find any good oldfashioned cloud news. So, the cloud didn't have much for us. Nothing to see here. Move along. I you know when I was thinking about themes for this it does it did feel like a month in which there was I don't want to say there was like a lot of negative news but there was a lot of a lot of news that that sort of felt like oh yeah that's that's the reality that sets in around enterprise when you start talking about AI stuff right so you know I feel like last month the month before we were talking about things like token maxing you know now we're starting to see people kind of come back to life and be like how about we do like token minimizing and token slowing down and stuff you know some stuff like that.
3:02 So, you know, I I think that happened a little bit. I think some of that had to do with just, you know, some some news around the IPOs and the IPOs, you know, maybe being a little a little, you know, softer or a little more muted than people necessarily expected. But what jumped out at you as you were looking through the the big, you know, big news of of the month that that kind of caught your attention?
3:22 >> Well, so I want to talk about the token maxing real quick and the tokening token minning or whatever it was the opposite. I found that one to be one of the funniest quick trends in this industry in a long time. And and here's why. You and I have have been around this game a while and you we know like, hey, if you want to measure people and people know about that measurement, they're always going to game the system. And the the second it was like, hey, we're going to measure you and your token usage. I was like, boy, that's a terrible freaking idea. And it took all of a couple weeks for that to be a terrible idea. And and so like for me it was just like do we learn nothing? Like even with all the things that that have been come to AI and everything that's going on, human nature still prevails through of like hey if you measure a metric humans are going to game that metric, >> right?
4:17 >> Yeah. No, it's >> it's not really an AI story. That's a human story. >> Yeah. No, it it it it feels like it feels like every every few years we forget that like the the really the the really big quote unquote and I'm I'm I'm kind of air quoting, you know, trends that come out of Silicon Valley, you know, that that on the surface you go, God, that makes no sense whatsoever. That that wouldn't fly anywhere else in the world that like it doesn't take long for that to not fly anywhere else in the world, you know. So the >> you know when you know whether it's the latest sort of weird diet fad coming out of Silicon Valley or you know the you know basically puffing your chest out to be like how many tokens did I use like >> well so I got a question for you and and we'll get into the news here in a second but I I kind of want to read the read the >> I don't know which way the winds are blowing you know because I feel like before I went on my hiatus from the show here Google was on top of the world and could do no wrong. Yeah.
5:11 >> And and I don't feel like that's the case anymore. And and like I don't know like where where are the winds blowing right now, Brian? Well, >> tell everyone what's what's happened here recently. >> Yeah. I think the thing that I think that the thing that that Brandon and I have sort of come to come to an agreement upon as as we've been doing cloud news or AI news of the month is is like everybody gets their moment and then very quickly their moment sort of goes away, you know, and >> Yes. I I mean I think that's I think that's sort of the news cycle is that you know every couple of months somebody you know piles up all of their announcements or all of their talking points or all of their media blitzes and you know in some cases it turns out really well like you know like you said like Google a few months ago sounded like they had their act together like they had the right engineering they were aligned to a model that was really working they'd figured out a business model I mean they were they were printing cash on top of the cash that they were printing you know all these But, you know, it's it's not unlike what we saw, you know, a few few years ago with Microsoft where it was like, man, they got their act together. And then, you know, we've seen what feels like rotating months sometimes between Anthropic and Open AI. And so, you know, I mean, I I think to a certain extent, the the takeaway I have from it is is a couple of things. Number one is they're not all that different from each other.
6:30 All right? I mean they they some of them have some structural advantages but you know it feels like number one maybe they're not all that different from each other right so the whole like can you tell which model's best you know I don't know it feels like at any given day some of the smart people from one of the companies is going to go jump ship to the other company and you know people are going to react to that accordingly and then I think the other big thing is like they all have so much money that even when they make mistakes it's sort of like okay cool well you I'll see you in a month, you know, fix fix your mistake or come out with something new. And so, yeah, I don't know. It it does I one of the things I heard recently, I was at an event recently, I think I talked about it in the last show, like there there does very much feel like a distinct difference from a year ago to now where people were like, "Oh, yeah, we're all in on co-pilot or we're all in on Gemini or whatever it would be." And now there there feels like a very distinct we're not really sure which one's going to win. we're trying to work around being, you know, tied to one of them specifically, right? So, I I do get the sense that the market is is kind of understanding the dynamics. they're not exactly sure how to deal with them, but they're they're trying to build some buffers in for themselves so that, you know, if something crazy happens, they're not they're not completely locked into something.
7:49 >> Yeah. And you know what this reminds me of? and the the early days of networking in like the Cisco >> and, you know, like Bay Networks and all of those others where it was like everyone had had to have some kind of, you know, carrier, if you will, or, you know, transport layer. >> And Cisco always wanted to sell you all the bells and whistles because they sold you the bells and whistles. They locked you into their thing.
8:16 >> Yeah. Yeah. And and then all the others had their, you know, proprietary features that they wanted to sell you because it would lock you into their thing. And all everyone wanted was just basic packet transfer. And everyone wanted the ability to figure out which one was cheaper. And yeah, it was a pain to rip them out and put the other one in. But like no one wanted to be locked in, but everyone wanted the features.
8:41 And it just feels like everything old is new again kind of thing. >> Yeah. Yeah. Yeah, I mean there's definitely, you know, there's there's, you know, pe people people want their cake and eat it too, right? They want, you know, it's the old saying like you can you can be fast, cheaper features, you know, like lots of features fast or cheap, pick two. And you know right now people are you know again I think this goes back to this thing we keep saying like we seem to we seem to forget that with every new with every new big trend some of us have lived through it lived it through it before and others are new to the to the thing and they just don't realize like you know you're going to you're going to pick two out of the three at some point. you don't think you need to right now, but you will eventually pick two out of the three.
9:21 And and they have their trade-offs and and the market, you know, the market sort of shakes itself out so that there's going to be the one that is expensive with lots of features and you're going to be locked in and then there's going to be the one that's cheap with, you know, less features and you're going to have to you're going to have to do some work to to work around it and you might decide that it it commoditized itself. And so, yeah, it's going to be, you know, it's going to be interesting in all those same ways. I think, you know, I think the thing we're we're trying to figure out is like where is it going to be different? Where where does it not look like, you know, stuff that we've seen before just with a different paint color on it or something like that.
9:55 >> Yeah. And and maybe that's where we dig into some actual news stories here. I mean, in the notes here, >> I mean, we have a good page of AI news, but but >> I mean, we're not going to go through all of them, but I mean, what are some trends here that that kind of jumped out? Well, I think I think the big I think the biggest news out of this this month really had to to do with the intersection of like well in particular the US government and and these models, right? So, you know, we had, you know, we had Fable and Mythos from from Anthos or from Anthropic come out. So, it was you know, Mythos had been announced as as sort of preview. It was the the scary security ccentric LLM that, you know, was was so scary and so powerful that, you know, Anthropic was only going to allow certain people to have it. And then they, you know, they made another announcement that was like, hey, we went from like the first 50 customers to like 150. So it was like, you know, there was a sense of like, oh, cool. Like they worked out some things and the, you know, these these companies, you know, that have access to it that manage, you know, critical infrastructure are, you know, happy to have their hands on it.
10:58 And then it then they were like, no, no, we're we're we're expanding that out to 150. And and what you weren't really sure of was like, okay, that was pretty quick. you know, did you really get enough feedback to feel like it was it was controllable or manageable or was this just like no, no, no, I got an IPO potentially coming up. I want to get it out there. And then of course, you know, we saw here, you know, a week or so ago that the US government came along and said that they're suspending anthropics, you know, that the to foreign nationals access to Fable and Nethos, right? So, you know, I think that came along. You know, we saw the the US government telling in Open AI to maybe, you know, could you guys slow down how fast your releases are coming out? You know, we saw the US government having a executive order talking about, you know, trying to to, you know, emphasize innovation, but at the same time trying to like put certain controls around stuff. So, I think we're, you know, we're we're starting to very quickly get into a an interesting interaction between these frontier models who, you know, not only are they sort of at the leading edge of the technology, but oh, by the way, they're both very close to being like, we're either filing for IPO, we're very close to filing for IPO, we'd like to get this out in the marketplace so we can get some more money. and you know the US government and you know the various entities around the US government being like we would like to have you drive innovation but at the same time we have to you know we have to do certain things for for security and protection of you know national security and defense and all sorts of stuff. So that that felt like the biggest story of the month in lots of different, you know, played out in a bunch of different characters.
12:32 >> And so here here's what I I also I have an interesting theory. >> Y >> I hadn't had a chance to talk about this on the show, but in the softwaredefined talk Slack, I know I I I mentioned it once or twice there. there's been there's been a theory >> that a lot of the Chinese models tend to do mass distillation >> y >> of the big models >> and for anybody who's not familiar with that what what does that mean you you're now in the model world >> so distillation is basically taking a larger better model and in this instance because it's API because there's different kinds of distillation because this is API like basically shooting queries at the API getting answers back figuring out how it's going to answer, how it thinks, and you don't have like if you had full access, you could get like the chain of thought and the reasoning commands and all the other stuff behind the scenes, but in this instance, you're basically shooting an API, you're getting an answer, and you're distilling down and getting from it how it reacts to certain things. And anthropic has been very vocal, but it's I think it's been decently wellnown.
13:39 Now, by the way, >> I don't want to 100% say, oh, all Chinese models are distilled from American models. No, >> it's not. But is there is it an additional tool in the tool chest? >> Probably. >> Sure. Oh, well, I mean, you know, it's it look I mean, you're not making any sort of like crazy statement. You know, Chinese technology companies have been very well known for, you know, kind of reverse engineering stuff that came out of the US. So, this isn't any different.
14:03 Yeah. >> So, the idea here is like, well, okay, you know, the reason why the Chinese models have always been about six months or so behind is because that's how long it takes to distill everything down and then republish it. >> Yep. Well, if if this is truly limited, in a couple months, we're going to see a pretty big drop off in Chinese models because they haven't had access to the latest ones to distill them. So, that's been kind of a running thing I've been like in the back of my head thinking about, right?
14:27 >> But, I mean, here's the thing, like like the big point is, yes, obviously both Anthropic and OpenAI race for IPOs and then, oh, by the way, OpenAI may be putting theirs off already even though they confidentially filed. And I don't know, it's it's been this weird ever since, you know, SpaceX kind of entered the race by like absorbing up XAI like and then they ended up, you know, being the first to IPO under SpaceX. It's really interesting to see the economics of all of this and the economics endgame. You know, for us, it's always been, hey, you know, you start a startup and then you either get bought or you IPO. And it's some people refer to it as the exit or some people refer to it as the start and you know that's the that's the starting line and you've got a different finish. But either way, what has been the endgame for a lot a lot of these large labs? I guess we're going to kind of find out because we're kind of in uncharted water when it comes to the government kind of stepping in some of this. Yeah, I I think it's it's it's going to be very interesting because you know whether whether you were sort of like pro AAI and you were like look you know I know there's a lot of spending on this but it's you know it's building towards something big and I know that the numbers maybe don't seem like they immediately make sense but you know you would have had the same philosophy when when the internet was kind of growing out or you know you're on the you know you're on the sort of Ed Zitron side of the the spectrum where you're like this is all going to bust it's a giant bubble and and so on and so forth It it does feel a little bit like some of the some of the forces that that will that will impact that are starting to come together, right? Like we're seeing we're seeing some IPOs. We saw Sarabus. We saw, you know, SpaceX. Now, the numbers they put out there seemed fairly high.
16:15 you know, they they saw some initial initial, you know, bounce, but now we're starting to see them them sort of fall back to to gra, you know, so gravity is sort of pulling them back down to to reality. You know, we've seen some concerns about, you know, do we have availability of, you know, data centers and energy and, you know, where are these guys going to find money from and and, you know, whether it's the money that went into investing in SpaceX or whether it went into, you know, Google R, you know, just had a new round of funding and so forth. It does feel like maybe the sources of money are starting to become a little tighter. maybe they're going to cost a little bit more.
16:47 It it does feel like we're, you know, and I don't know if we'll realize this through some sort of crash or whether we'll just sort of see this sort of gradual moving into a new normal, but it does feel like you know, we're at the beginning of whatever AI is going to start to look like at at sort of the end that the the economics of its first phase are starting to, you know, starting to shape up a little bit. And you know, that's that's probably a good thing because it is really hard to to understand a market when the dynamics of it aren't really based in any sort of reality or formula or model that that we're sort of used to. Like as humans, we're sort of used to doing things that feel like something else. So, yeah, it's it's going to be interesting. I think the SpaceX thing is is is interesting to me only be again like I don't really care about the rocket stuff and the outer space stuff and so forth, but it is interesting to see how quickly they sort of moved out of the game, right?
17:42 Like it it was like we bought a bunch of capacity, but we're really nobody's paying us for the capacity, so now we're going to just quickly become sort of a neocloud, but but we don't really play in any other way. And you're like, okay, well that's fine. Like that's the market adapting. But, you know, like a lot of the other things feel like they're for good or for bad going in a in a direction that feels a little bit sort of more natural and normal than than some of the craziness we've seen so far.
18:06 >> Yeah. But I'm I'm going to add this to because I saw it on X actually earlier today. It's so interesting too how quickly things can change even in like what you were just talking about with that space buys cursor for 60 billion which first of all cursor 60 billion. Let that sink in. >> Yeah. Yeah. Yeah. >> To everyone for a second. But then too, they've already announced, okay, they've merged a lot of the cursor code, you know, into I don't whatever, Grock or XAI or whatever the latest models. I can't keep up with them anymore.
18:36 >> And you're already seeing I want to say it was Deepseek level, you know, coding and all these other things. And so like, hey, maybe they're back in the game. But but XAI was, you know, or or Grock was basically all but dead because all the founders left and you basically had, you know, Colossus One and Colossus 2 sitting there doing nothing because you literally didn't have an AI company anymore. And so now you have an AI company again all of a sudden.
19:02 >> Well, it's we'll see. >> Yeah. I mean I I think maybe what I what I read out of that so the yeah the price of cursor you know part of that I sort of chalk up to like okay we live in stupid times so people are paying stupid money and and at least for a few days there like and SpaceX was living on on very stupidly valued money so they you know they spent more than they should have but like they didn't really you know they they had they lost their team.
19:27 I mean, I guess the the thing that sort of jumps out at me is is how fragile it feels like maybe even all of these frontier labs are that the small number of people who know how to do this stuff at any point in time could just get sick and tired of their boss or tired of whatever it is and either like leave or go do something. I mean, like that that feels much more fragile than we've ever really deal with. I mean, I know, you know, like you and I lived through the time when there was like there was only so many people that understood like OpenStack and then they would work it somewhere and then they would sort of leave and go somewhere else, but like, you know, you never felt like, you know, I heard the AT&T's OpenStack team all moved and went over to like Morgan Stanley. You never felt like, oh well, AT&T is not going to be able to process any phone calls anymore or like, oh, Morgan Stanley is not going to be able to process any bank transactions. But like this it does feel like you know if tomorrow like 40 people from open AI were like oh we're going back to Google or like 40 Google people were like hey we're going to start some new company that all of a sudden you'd be like I don't I don't know that they're that good in you know like I don't know that that does feel fairly fragile to me and I think it it ultimately goes back to the thing that we sort of knew which was like you know when this stuff got started and it was the Transformer project which was again what six or seven years ago there weren't very many of those people and it's not like all of a sudden you know 10x or 100x of them just sort of popped out of the woodwork. So, >> you know that that feels >> I mean it goes to that like the like we know our audience loves you know sports analogies here. We we'll put we'll throw another one in here of like, okay, it's almost like this was pro sports and, you know, there's there's just we're paying all these there's stupid stupid money to to the superstars, but there's only like six or seven superstars >> and >> yeah, >> and yeah, like like Google losing two big folks, one to Open AI and and one to Anthropic this month and everyone's like, "Oh no, Google's dead." And it's like I'm pretty sure they got a a pretty good staff over there still. But but that's just how I don't know if it's like >> just the drama like everyone's kind of following it because it's headlines and like this has become something that just is quote unquote newsworthy now? Like does this really really mean like will it turn into actual material differences?
21:56 >> Well, companies I don't I think it's one of those things where again, you know, you're you're you're starting from a very small base of people, right? And you know, if you if you go back to like, I don't know, there was a website somewhere that was like, hey, here's here's where all the original people used to be that used to be at like Google Brain or you know, like you know, it was like here's where they started.
22:16 Then half split over went to OpenAI and then and then there was this sort of distillation of a bunch of people, right? like the folks that that run your company, you know, coming came out of that world and there's, you know, 50 other companies that a few of them went and worked at plates. So, it kind of got distilled and then we've seen some, you know, some moved here, some moved there, but like we haven't like that that number hasn't exploded in any way, right? It's not like, oh, all of a sudden they all took a, you know, a master class and, you know, there's there's, you know, 10,000 more of those people. So, I think we're seeing a little bit of, you know, and we saw this at at Microsoft, like Microsoft tried to, you know, build the super team. and they're going to build the super intelligence team. And you know, on on paper, people were like, they hired a lot of really smart people. They threw a lot of really smart money at smart people. And then it was like, but it's not working out. Oh, why is it not working out? Oh, because they have personality conflicts. And you're like, oh yeah, this is exactly what happens where like, you know, smart people don't want to work for managers they don't get along with, or somebody wants to be in charge of the team, but not everybody can be in charge of the team. And so you got to, you know, like I think we're seeing those sort of natural things happen. And then I think we're, you know, I think there's a certain amount of these people also starting to realize like, hey, we're probably getting really close to me getting that super payday and I better go get that before maybe this doesn't work out, right? Like maybe I'm only getting the like super good payday, not the super super super good payday. And you know, so you get you get irrational behavior and you know, and because the press and analysts don't know how to, you know, determine is, you know, scientist number seven better than scientist number 11, it's like, oh my god, somebody left. So >> yeah. Yeah. I mean, you look at it this way just in the past, >> I don't know, 12ish months.
23:58 >> Yeah. >> I mean, most of the most of the meta team left. and so and then everyone's like, "Oh my gosh, Meta is done." You know, they're not a Frontier Lab anymore. And then yeah, Microsoft and all their problems and you look at Google and you know like Google's doing good, then they're doing bad, then they're doing good, then they're doing bad. Like it almost seems like it is not even cyclic. It's almost like waves in the industry and yeah, it is very almost clicky of like where is there the the quote unquote good AI teams at any given time and it almost seems like it has a little bit of gravity right now.
24:35 >> Well, yeah. I mean, and if you think about it even a little bit, like it was like, okay, the whole Grock team left. Okay, where did they go? Did they just disappear? The Facebook team all like quit and left. Where did they go? Did they all disappear? Microsoft has a gigantic amount of money and they're not cranking out anything on their own. And so it's like the other thing it sort of says is like you know and going back to your sort of Cisco analogy like you know we used to have a kind of an old saying where it was like if if somebody started a networking company we were like well if their founding you know if their founding people didn't come from Cisco Juniper or Stanford they're probably not that good because they were never around learning the really hard parts of doing this stuff you know so yeah you can hire a team of super people but I I my guess is like there is a certain amount of of DNA that you have to have in there like the sort of how do you you know how do you do it like everybody everybody wanted to make sourdough bread at co but like a lot of people were like oh my sourdough thing didn't work and it was like well god didn't you just watch the YouTube video like you should have been able to make sourdough bread I know I think it's I think there's a little bit of that too >> yeah completely completely >> but yeah man I I you know I I feel like we're we're kind of moving into a little bit of and again this might totally change next month with a bunch of new announcements but like this felt like a month that It was very much like it wasn't super sexy. It was a lot of like adult issues going on like companies are trying to you know file for IPO except now instead of acting like teenagers somebody's like do you actually have any money in the bank? How much money do you actually have or oh you would like to do this thing? Well, the US government would also like for you to not leak the nuclear codes out to, you know, like it felt like a lot of stuff that was like, yeah, I guess we're going to have to do that because, you know, that keeps societies from getting out of control or, you know, security problems from spreading wildly. And, you know, it's not super sexy, but it is the reality of, you know, running things that are stable for millions and millions of people.
26:34 >> So, Brian, is this the beginning of the trough of disengagement? Might be. It might be. I mean I mean honestly like that was it was a little bit of why for like the last you know month and a half I've been trying to be very focused on on enterprisy issues and you know as Brandon likes to say like oh god now you're going to talk about governance and and you know like you know like you're going to give me a a login that goes to a you know a dashboard or something but it's like yeah I mean you know cost controls governance accountability like at some point those things come to every market and it feels like we're right at the beginning of that. Now, that doesn't mean we won't go through a couple more, you know, rounds of sort of irrationality, but it does feel like Yeah, Aaron might be calling it the the trough of disillusionment for some of this stuff. So, we'll we'll see. We'll see. I think it I think we'll I think we'll know one way or the other when, you know, when when say like an anthropic goes IPO and we get through a couple of quarters of like, okay, how much are they really burning through?
27:34 How much are people willing to sort of bet on that one? But, yeah, I think we might be might be getting that direction a little bit. So, anything else jump out at you as far as, you know, like like you said, lots of stuff in the in the news. People can go dig into it and source so forth. >> I mean, here's a couple observations real quick. The cerebrous earnings actually considering a first quarter after IPO, I'm decently impressed by that. I'm just going to call it out and say, yeah, I mean, hey, good on them.
27:59 They did good. I'm just kind of looking over these headlines real quick and just making a couple quick comments. Hyundai now has full ownership of Boston Dynamics, the robots, the the viral robot videos, >> the dogs with guns. Yeah. >> And I'm just like, do we I don't know. Like, part of me is very concerned about that. I don't know. Like, I know Hyundai does more than cars, but I immediately think of like Hyundai and cars and Boston Dynamics and the videos they've done. I don't know if that's a I don't know, an intersection I like or I'm comfortable with. But hey, it is what it is. Like, they own it. No, >> I will say this though.
28:32 >> Well, I mean, let me go Let me let me let me put it this way. Like, would you rather have like Loheed Martin own them? >> That's true. That's true. No, and I but I was going to say too, like I've seen SoftBank cash out of a few positions here recently, too, which also may be another sign. You know, SoftBank was had so much money for so long and you're starting to see SoftBank starting to cash out, you know, here and there, which by the way, they had a really crazy presentation. I'm not even going to necessarily go go talk about it, but if you if you click on one link in our show notes, go look at the SoftBanks investor presentation.
29:05 >> Yeah, >> that is some crazy stuff. >> Yeah, >> it's Yeah, it's out there. >> Well, man, I didn't want to I didn't want to overwhelm you with too many things. now that you're back, I don't know if you know, again, I don't know if you're coming back again for a few months. You might be, you might not be. >> Yeah, this might be a popup and go away, but no, I I certainly hope not. you know, it's it's it was a a needed period to kind of focus on the day job and adjust to everything that was going on, but coming up for air.
29:35 >> Yeah. Well, and if you want to see a sign of life, Aaron's face will be on the the YouTube video, so go check out the YouTube link. >> That's right. >> And it's it's not AI, it's actually him. So, all right, man. Well, with that, folks, that is the the June AI news of the month. And we'll be back a month from now with who knows who. Could be me, could be Aaron, could be Brandon, could be three new guests. It could be three new hosts. Who knows? But anyways, folks, with that, thanks for listening. Thanks for telling a friend.
29:59 Thanks for giving us your time every week, a couple times a week. We really appreciate it. And with that, we'll wrap it up. We'll talk to you next month or the next week.
Summary
- The hosts welcome Aaron Delp back to the show after a hiatus, discussing the World Cup and its cultural impact.
- They note a lack of significant cloud news and a shift in AI discussions from "token maxing" to "token minimizing."
- The conversation highlights the fragility of AI companies, with concerns about talent retention and the impact of key personnel departures.
- They discuss the U.S. government's increasing involvement in regulating AI, particularly concerning national security and innovation.
- The hosts speculate on the implications of Chinese models potentially lagging due to limited access to advanced U.S. models.
- They mention the recent IPOs and the financial realities facing AI companies, hinting at a possible "trough of disillusionment" in the industry.
- The acquisition of Boston Dynamics by Hyundai raises concerns about the intersection of robotics and automotive industries.
- The discussion concludes with reflections on the volatility of the AI market and the unpredictable nature of tech trends.
Questions Answered
Who is the guest and what have they been up to?
Aaron Delp, a returning guest from Witness Protection and formerly the Cloudcast, discusses his recent experiences, including watching the World Cup in Paris.
What are the current dynamics among leading AI companies?
The discussion highlights that major AI companies like Anthropic and OpenAI are not significantly different from each other, and the market is becoming uncertain about which will dominate.
What is the relationship between AI companies and government regulations?
AI companies are navigating the balance between driving innovation and adhering to government security regulations, with some nearing IPO.
How stable are AI companies in the current market?
The conversation reveals that AI companies are fragile, with key personnel easily swayed, leading to rapid changes in company fortunes.
What factors contribute to the success of AI companies?
The discussion emphasizes the importance of foundational talent and experience in AI, suggesting that successful companies often stem from established industry leaders.