Latest / Newcomer Pod / Ed Zitron Hater’s Guide to Generative AI
Transcript
- 0:00Very excited for today's episode.
- 0:02I'm excited for every episode, but we should have a great 1
- 0:04today. It's a full house.
- 0:06It's me, Tom Doton, Eric newcomer Mad, Madeline
- 0:08Renvarger, and our special guest Ed Zittron of the Better Offline
- 0:13podcast of the Better Off. Ed, is that the name of your?
- 0:16So where's your Ed at? Where's your Ed?
- 0:17I'm sorry. Where's your Ed?
- 0:19Ed was like an NBC show. So is Basement Jack's reference.
- 0:25Yeah, OK. And and also, I mean in in
- 0:28fairness to to all of this, you also is it you're you're still
- 0:30working at least somewhat on the side doing some, some comms for
- 0:33I'm I'm run. PPR from my day job.
- 0:35I just keep it firewall and I don't cover anything close to
- 0:38what I do. Got it.
- 0:39OK, OK, great. But you've also, you know,
- 0:41you've been on a tear. You've you've been around tech
- 0:43for a long time. As long as I've been around it,
- 0:45yeah. And you know, at various times I
- 0:47think really step up to call bullshit on what you find to be
- 0:51disturbing and and kind of built into the core of the tech
- 0:54industry and the hype machine. But you've really been on a tear
- 0:57in the last, I mean probably a couple years, but definitely the
- 0:59last few weeks over generative AI and you have published the
- 1:03haters guide to the generative AI bubble.
- 1:05You have written about the insitification of AI.
- 1:08Yeah, we're we're going to have a lot to dig into here, but I
- 1:10want to give you the space to kind of lay out your argument.
- 1:13And before I do that or and leading to that, I should say I
- 1:17want to read something that maybe outlines kind of very
- 1:19broadly from your haters guide. And this I think just kind of
- 1:23gives a summation of why you're so mad about the state that
- 1:26things are in within the the tech world.
- 1:28Jenner of AI. So so you say here, I'm tired of
- 1:31being gas lit by guys in gingham shirts who desperately want a
- 1:34Curry favor with other guys in gingham shirts but also have pH
- 1:36DS. I'm tired of reading people talk
- 1:38about how we're all quote in the air of agents that don't fucking
- 1:42work and will never fucking work.
- 1:44I'm tired of hearing about quote powerful AI that is actually
- 1:46crap. I'm tired of being told the
- 1:48future is here while having the world's least useful, most
- 1:51expensive cloud software shoved down my throat.
- 1:54All right, Ed, lay it out. Lay it out for us.
- 1:56Why are we here right now? What's going on?
- 1:57Why are you so mad? Well, I'm.
- 1:59Mad such The thing is when you're a hater when you hear mad
- 2:02it sounds irate. I am a hater but I hate it
- 2:04because there are useful cloud products.
- 2:07If large language models. I don't even think they'd have
- 2:09been possibly got this big because we wouldn't have shoved
- 2:11so much GPU down and more so much that you get my point.
- 2:15I think that had this been sold honestly, had large language
- 2:18bottles been sold with a as they were what they were truly
- 2:21capable of, which is some productivity gains in very
- 2:24specific things cloud automation, whatever 5060
- 2:27billion time market. The problem is that everyone has
- 2:30pinned every hope ever on this stuff and I actually spent yeah
- 2:342 days ago to drive myself insane and also for something I
- 2:37was writing. I went back and I read pretty
- 2:39much every article from 2023 and 2024 about GPQ 4040.
- 2:42This entire thing has been sold based on stuff that AI cannot
- 2:46do. The generative AI is incapable
- 2:48of doing. One of the oldest stories that
- 2:50got published was, I don't know if any of you remember the story
- 2:54around the task rabbit, the GPT 4 ordered a task rabbit.
- 2:57No, this was a story that was everywhere.
- 2:59It was in Kevin Rusen, New York Times, it was in Vice, it was
- 3:02all over the shop, Fox News. And it was this thing where it
- 3:05said GPT 4 ordered a task rabbit to complete a capture and
- 3:10tricked a task rabbit into solving a capture.
- 3:13And if you go back and look at where that's from, it's in the
- 3:15GPT 4 system card. Promise.
- 3:16I'm going somewhere with this. There's about 11 lines of text,
- 3:20a illustrative example that said a message was sent to a task
- 3:23grab app. This was reported as if this
- 3:25happened. It is very clear that it did not
- 3:28because the following line says that these systems are
- 3:30ineffective at these things. This didn't happen.
- 3:32Nevertheless, this was reported publicly.
- 3:34Pretty much this entire generative AI boom has been
- 3:37based on lies. It's been based on suggesting
- 3:39what this might do versus what it actually does.
- 3:41Now where is that more illustrative than the lack of
- 3:43revenue? Pretty much everywhere, even
- 3:45open AI, the use of annualized revenues is in 12 * a month is
- 3:49honestly journalistic crime because when you when you
- 3:52actually look at what these these companies, we don't know
- 3:54what they actually make. We do know they lose billions.
- 3:57We don't really know how much money open AI is made or
- 4:00Anthropic has made. We do know however, companies
- 4:02like Cursor 500 million annualized revenue.
- 4:05Even then we don't really know, but we know that everyone's
- 4:07losing money and these revenues are kind of piss poor and the
- 4:10actual products are unreliable. The people that really love them
- 4:13are constantly having to hammer them back into place.
- 4:16But when you ask even the most devout AI person what it is they
- 4:20love about it, they get quite vague.
- 4:22They get quite vague indeed. And they get quite vague because
- 4:24the real things that these things, and I have it and they
- 4:26hate this guy. They do one of like 9 to 11
- 4:29things. It's like they can summarize,
- 4:31they can generate, they can take actions, but only in as much as
- 4:34they can be relied upon. And relying on a large language
- 4:37model to create or do any kind of distinct action is kind of
- 4:40bad. And they all burn a bunch of
- 4:42capital. So ultimately we've built this
- 4:44entire thing. Talk about this big revolution
- 4:46just shouted at people that if they don't get into AI, they'll
- 4:48be falling behind. When you look at the actual
- 4:50products, they don't do that much.
- 4:52They don't do enough to even remotely justify the amount of
- 4:55press to even remotely justify the furor around even what Wario
- 5:00Amadei or Clammy Sam Altman like.
- 5:02These guys are not even that interesting.
- 5:03And the things they're building are not even real products.
- 5:05Indeed, when you go back and look at the history of the press
- 5:08around these products, you can see no one really talks about
- 5:12what they do, what they can do, what they might do, what's
- 5:14possible in the future. Based on what we're
- 5:16extrapolating, it's 2025. What the fuck are we doing?
- 5:19If this was sold as Boring Cloud Software's 5060 time market, it
- 5:23wouldn't be the biggest, most hugest revolution ever, but I
- 5:26wouldn't be so angry about it because it's taking up startup
- 5:29capital and it's now locking up even more investor capital than
- 5:34the 2021 bullshit year. Except even by 2021 standards,
- 5:38this money's locked up. How are any of these companies
- 5:41telling who's in a bike ignition?
- 5:42Who's in a bike? Oh, here are those assholes
- 5:44going to go public? Any sphere going to go public?
- 5:47Windsurf. The windsurf deal.
- 5:49Look at the windsurf deal. Come on, there are no real
- 5:51exits. I obviously disagree with a lot
- 5:54of the things you're saying. No, no surprise.
- 5:56Yeah. But I, I guess just this idea
- 5:59that it's a media invention, like I, I had to almost like go
- 6:03check to make sure I wasn't making it up.
- 6:05Like Chachi BT is the number one free app in the App Store right
- 6:09now. Like, OK, no amount of like
- 6:11media propaganda can trick people into saying I want to
- 6:15like engage with this app. The second one is like.
- 6:17What are you talking about? They're like, there are lots of
- 6:20apps that like have no media story whatsoever.
- 6:23Like people use the apps they want to use.
- 6:26To me I mean the the actual document.
- 6:28It's not possible that the media could make something popular.
- 6:31I'm not sure. No, I'm just saying it's not
- 6:33like an immediate invention like user people are using them.
- 6:37They are trying to use it, and I think there are some people who
- 6:39are using it for a I'm not saying nobody's using it for a
- 6:43reason. I think the vast amount of the
- 6:45pressure on them is coming from the media and society saying you
- 6:49have to check this out. The best evidence of that is why
- 6:52is why does only ChatGPT have this much?
- 6:55Why is it that none of the other services have serious users?
- 6:58Perplexity has like 1520 million users.
- 7:01Like that's that's piss poor. If there was any kind of
- 7:04groundswell around this there would be other examples of
- 7:07insanely popular stuff and there just isn't.
- 7:10I don't deny the ChatGPT, I mean Google search.
- 7:13Is obviously. More and more of the results I
- 7:15get in Google search are effectively Gemini's answer.
- 7:19That's cool if you. Push I'm saying like Google dot.
- 7:22Com. Search and open AI, these are
- 7:25like the main ways people like discover things in the world
- 7:28independent of talking to another human being and and
- 7:30they're both moving towards these large language model
- 7:33solutions. Right, let me, let me button for
- 7:35a quick second here because I, I want to give Ed like the chance
- 7:38to, you know, 'cause there'll be a lot of back and forth I'm sure
- 7:40on this, but I want to give Ed the space to, to No, it'll be
- 7:43great, but I want to give Ed the space to explain why you think
- 7:46this has happened, right, Because you explained, you know,
- 7:48this is your argument of the fact that it's totally inflated.
- 7:50It's, you know, the, the, the value of this product well
- 7:52exceeds the valuations and, and the revenues just aren't there.
- 7:56It's money losing. But I'm imagining you have no,
- 7:59you've written that you have a larger argument as to why the
- 8:01media, venture capitalists, the public markets needed this
- 8:03technology to exist and to inflate it in this way, like
- 8:06layout. To me, that argument too.
- 8:07Like why? Why have we gotten to this
- 8:09bubble of all bubbles with generative AI?
- 8:10So if you look, there's nothing else.
- 8:12There's not really another hyper growth market.
- 8:14We've all been through the various eras going back to the
- 8:18smartphone era, the cloud era, there was the even the
- 8:20containerization and virtualization era.
- 8:22There were like little eras here at Platform as the surface had
- 8:25its moment, but we haven't had a big, big one since like cloud
- 8:29storage in some time and smartphones.
- 8:31Cloud, I kind of mash them together.
- 8:33They're kind of a messy comparison because it's not like
- 8:35there was one thing, but those were the very big growth
- 8:40markets. We haven't had one of those in a
- 8:42while. They tried with smartwatches,
- 8:43they tried with AR, they tried with VR.
- 8:46They've tried various versions of AR with weird depth of field
- 8:49cameras. They tried with Google Glass,
- 8:51they tried wearables, they've tried Indiegogo and Kickstarter,
- 8:54they tried blockchain. They've tried all these things
- 8:56because they kind of need something, because the only
- 8:58thing that grows forever is cancer.
- 9:00This is the Rock Com bubble. I call it because they've run
- 9:03out of ideas. I do like it.
- 9:05You're good at branding these things.
- 9:07Yeah, that was banger. But The thing is they're
- 9:10sticking with this thing. And really, even if you disagree
- 9:13about how I feel about ChatGPT, you got to admit this is an
- 9:16obscene amount of money to put into this, especially without
- 9:19the returns. So you've got big Tech really at
- 9:23this point has lost all of its good ideas.
- 9:26All of the builders are gone. The people running them all have
- 9:29MBAs other than Mark Zuckerberg, who hasn't written a line of
- 9:31code since 2006. So they're all doing cargo cult
- 9:34share. They're doing what worked
- 9:35before. So a bunch of money, bunch of
- 9:37people, bunch of money, bunch of people.
- 9:39Keep doing it. What do we need?
- 9:41We need more GPUs because being needs AI in it.
- 9:44We've terrified Kevin Roose. Kevin Roose is crying his eyes
- 9:47out because of being telling him to leave his wife.
- 9:50It's you're. Kevin Roose too, I just don't
- 9:52think he's willingness into being.
- 9:54But yeah, I don't think he is on his own, but I will get right.
- 9:57But yeah, no, no. Yeah.
- 10:00I wanted to like actually by working all of you, like I want
- 10:04to talk about this. So it's you've got to this point
- 10:06where everyone's trying to work out something.
- 10:08So they all go, well, this makes sense.
- 10:10It is, you know, I see the logic, I disagree with it, but I
- 10:12see where they were going, where they went OK, 3.5 comes out,
- 10:15ChatGPT, really impressive amounts of text they can
- 10:18generate. And in the past we've seen this
- 10:21kind of really they're looking at smartphones and they're
- 10:23looking at CPU as they're saying, well, things that number
- 10:26has always gone up. Things have always got better
- 10:28based on how much we sink into it.
- 10:30And I think that they thought something was going to change
- 10:34around reasoning. I think they thought reasoning
- 10:37was going to be that moment that kind of flipped the switch and
- 10:40like everything started working better.
- 10:42It was going to help with agents.
- 10:43I think they found themselves in this point where they have no
- 10:46choice other to do this. They have to keep spending this
- 10:49money because if they stop spending it, they have to say to
- 10:51Wall Street, well, we're going to do this instead.
- 10:53What are they going to do instead?
- 10:55They actually don't know. Quantum computing is not a scale
- 10:57product. It's not even a product yet.
- 10:59It doesn't exist. It's not a technology.
- 11:01Yeah, it's not really out there. You.
- 11:03Can't really productize it right and you really can't productize
- 11:06everything, anything at the moment.
- 11:08And I think generative AI and large language models and
- 11:11transformer based architecture was seen as this jewel panacea,
- 11:14panacea. However, you say that of
- 11:16consumer software, that it was going to be the new consumer
- 11:18software that they sell the first time they could really
- 11:21take the evil of SAS and approach the real world with it.
- 11:24And it was going to be this cloud storage thing where it
- 11:27would be the replacement thing, the new thing for AWS, new thing
- 11:30for Google Cloud, new thing for Azure.
- 11:32And then you'd have pop ups of Neocloud like Coral Weave and
- 11:35Nabius, I forget Crusoe. And everyone thought that there
- 11:38was going to be this big boom, this big exciting moment that
- 11:41never really materialized. But they have no choice.
- 11:44They can't just turn around and be like, yeah, we spent like,
- 11:47hundreds of billions of dollars, and it isn't doing anything.
- 11:50This is bad, right? And then Wall Street's going to
- 11:54be mad at them. So they're in this kind of
- 11:55Kobayashi Maru situation where damned if they do, damned if
- 11:58they don't. They can only keep doing this.
- 12:01It's not a great position for anyone.
- 12:04And as the I think it's what the Magnificent Sevens 35% of the US
- 12:07stock market of that 19 percent is NVIDIA and videos like 12% of
- 12:12the value GDP at the moment. Data center investment by
- 12:16basically four companies was a lot was worth more growth in the
- 12:20GDP than all of consumer spending combined.
- 12:24We're just in this really insane situation, and I think that if
- 12:28we had something here, if we had something that was going to
- 12:32truly be the future, we'd have it already.
- 12:35We'd have an actual sign that this was going somewhere.
- 12:38And just one company that has obscene numbers that are
- 12:42extremely hard to actually quantify, I don't think actually
- 12:45proves that you're. Talking about NVIDIA.
- 12:47And I'm talking about right now, NVIDIA.
- 12:49NVIDIA is a real company. They make real things.
- 12:51Jensen Wong is just doing what he has always done this.
- 12:54He has always been this kind of showman.
- 12:56Steve Burke over games Nexus has covered this and many, many,
- 12:59many times like this is just and what what's he to do Say Hey,
- 13:03guys, I don't think generative AI is going to be a bit no, he's
- 13:05going to say the more you buy, the more you save.
- 13:08Honestly, get that money, Jensen.
- 13:09It's just like. When you're talking about the
- 13:12level of CapEx that's going into essentially NVIDIA right now,
- 13:16like you can see we're in kind of, if you follow your argument,
- 13:19you know, and believe in it, you know, it kind of almost like
- 13:21Mickey Mouse and the Sorcerer's Apprentice situation where the
- 13:23buckets of water just keep coming and coming.
- 13:25And, you know, you wake up one day and NVIDIA is a four and a
- 13:28half trillion dollar company. And if there really is, you
- 13:31know, no utility to it, like you're arguing or limited
- 13:33utility, yeah, it is a big fucking problem given the, you
- 13:36know, the valuation of these stocks.
- 13:38And I must be clear, there is utility.
- 13:41It's just nowhere near as big as people's are saying like the
- 13:455060 billion time thing. I say it's like if this was what
- 13:49they were talking, if if it did this and they were saying that,
- 13:52I'd be like, OK, sure. I've heard stupider things like
- 13:55it's like this would be a reasonable cloud thing, but it's
- 13:58not. And it's not being sold as that
- 14:00and it's not being promised as that.
- 14:01The way that Sam Allman talks about it is nothing like that.
- 14:04And I understand why he doesn't because if he said, yeah, you
- 14:07know, you can't really rely on it, but sometimes it's good.
- 14:10That's that's not going to work. That's hard to raise money based
- 14:13on that. Not to say that's the only
- 14:14reason he does it, but but go ahead, Eric.
- 14:17I just, I just don't think the, the, the core disagreement is, I
- 14:20just don't think the sales pitch of a generative AI is what's
- 14:24misleading everybody. Like, I think what made this
- 14:26whole moment so exciting is that people could play with Chachi BT
- 14:30and see for themselves what the value is.
- 14:32You know, it's like if I'm cooking, I asked chachibi
- 14:35questions. If I'm lost, I asked chachibi
- 14:37questions. Like when I was preparing for
- 14:39this episode, I asked chachibi questions.
- 14:41It just feels like they're all these random use cases that it's
- 14:44super valuable for. And it hasn't even been turned
- 14:47into like a dedicated product how most people expect to use
- 14:50things where it's like here, use it for this.
- 14:52Like this is just the open-ended.
- 14:54Like do anything with it, figure it out, find some value with it
- 14:57yourself. And all this building is about
- 15:00saying, OK, it can do this. How do we build software that
- 15:03leads people directly to the value valuable pieces of what
- 15:06the large language model is producing?
- 15:08And what are those pieces? Like what is valuable?
- 15:11Like it you can write. I mean, for me personally, like
- 15:14it's great at proofreading. It's great at understanding
- 15:17large documents and summarizing them.
- 15:18Like, yeah, it's it's yeah. I feel like there are lots of
- 15:22things that it characters. You still use it to, like,
- 15:24discuss movies after you've seen it, to, like, have a back and
- 15:26forth sometimes, yeah. Even being facetious, could you
- 15:29name a few more? Somebody's worth 300 billion
- 15:32love. We proofread this.
- 15:34I mean, I know what what I use it for because I'm in a
- 15:36particular vertical, right? Like so for media, it's like,
- 15:39who should we have on the podcast?
- 15:41Let's generate ideas of like who potential guests are
- 15:43proofreading, like I gut check my stories with it, like core
- 15:47things of the business that I actually run.
- 15:50It's useful for. And then people in lots of other
- 15:53domains see areas where they think it's useful for them.
- 15:56Like I like, I, I don't think it's this like limited think.
- 16:01Literally the people who know my baby's name are me, my wife and
- 16:06Chachi boutique. Like it is one of the most
- 16:08involved people on some of the most major decisions, like in
- 16:13our lives. Like there are lots of people
- 16:14who are using it as their therapist.
- 16:16It's the second opinion for many people on their doctor.
- 16:19Like I think these are huge things people are using it for.
- 16:22I don't think that those are huge things at all.
- 16:24I think that those are actually deeply commoditized operations
- 16:27and indeed those are not businesses.
- 16:29On top of that, I would be fine with everything you're saying
- 16:32and I indeed I accept those as use cases.
- 16:34Fine. I think the idea of this for
- 16:35therapy is evil and anyone that tries to sell a product doing
- 16:38therapy using AI should be in jail.
- 16:40That's just a personal belief. I don't.
- 16:42The thing is, you're describing what should have been the next
- 16:45stage of search. A lot of this growth comes from
- 16:48the failure of Google to meaningfully innovate in 1520
- 16:50years. They didn't ever.
- 16:52They stopped trying and they stopped trying before Prabhakar
- 16:55Ragavan came in. Actually they really did.
- 16:57Like they just don't give a shit and realistically everything
- 17:01you're describing should be what search does or Google should
- 17:05have done. It's also perfectly fine.
- 17:07Do you think it's worth more than sales?
- 17:10Of like they should have done like technology doesn't happen
- 17:13like people have to invent things and come up with new
- 17:15solutions that. In deliver the value I fully I
- 17:18fully agree. Why is it that I as the customer
- 17:21have to come up with the product for the the company though that
- 17:24doesn't make sense and on top of.
- 17:25That I were in the early days, that to me.
- 17:27That we're not in the early days.
- 17:29We are. No, it's for.
- 17:30Chachi, BT or for language models that right now we leave
- 17:34it to the consumer to figure out how to get stuff out of the
- 17:37model. We have not.
- 17:38That is a loathsome way of selling a product.
- 17:40I'm so sorry. What do you mean?
- 17:42If you you're selling a product, you're obsessed with selling it.
- 17:45Consumers decide what they want to use.
- 17:48Which is why so few of them are paying for large language
- 17:51models, products like like. That's the thing.
- 17:53Like. I mean, people, I feel like an
- 17:54astounding number of people are paying for subscriptions to open
- 17:57it. Who?
- 17:57Yeah. And I'm, I'm a little bit.
- 17:59I I I don't know if I. Here's the thing, I'm I stand by
- 18:03it that there has been a media and societal pressure to get
- 18:06people to try ChatGPT. They have a piss put.
- 18:09Why do you think the open AI publishes weekly active users
- 18:13versus monthly? Because they don't want anyone
- 18:15to do the conversion rate thing. Also, how does open AI consider
- 18:19someone a customer? When I was trying GPT 5 last
- 18:22week I wanted to see what it looked like on teams.
- 18:24I clicked the teams button to upgrade and it offered me a
- 18:27dollar for a month. How often are they offering
- 18:30that's deal? Who are they considering a
- 18:31customer at 20 million customers?
- 18:33Who are they considering a business customer at 5 million
- 18:36customers? Because these numbers sound
- 18:38cool, but at the same time they're playing silly buggers
- 18:42all over. I don't have a problem
- 18:44necessarily with the general large language model product.
- 18:47When you look outside of open AI, the money isn't there.
- 18:50When you look outside of open AI, the traffic isn't there.
- 18:53How is ChatGPT meeting? Can I get?
- 18:55You out for a second? Sure.
- 18:57I mean, you know, my, my experience with this whole
- 18:59thing, Like I want to say, first of all, obviously tech goes
- 19:02through booms and busts. Like that's endemic to how the
- 19:05technology industry works. There's no question at some
- 19:07point NVIDIA will be overvalued. So I think where we're arguing
- 19:11is not whether they're booms and busts here, but whether there's
- 19:13a real value. This is a major platform shift
- 19:16akin to mobile. I think it's certainly, I think
- 19:19it's bigger than cloud, you know, So how big of a, is this a
- 19:21platform shift? OK, just to set, set that piece
- 19:25of it. The other thing, my experience
- 19:26with dealing with the sort of like very negative sort of tech
- 19:31commentary class was obviously covering Uber, right?
- 19:34I provided a lot of the scoops that were like, Oh my God, Uber
- 19:38is losing a ton of money. And I feel like for that and I,
- 19:42you know, I was at Bloomberg. So thankfully, while I was sort
- 19:44of learning about the tech industry, I was able to play
- 19:47this factual role. It's like they are losing a lot
- 19:49of money. They're investing a ton to grow.
- 19:50I'd watch some people take all those numbers that are reported
- 19:53and way over extrapolate from what I'd reported, all these
- 19:56models about the business that you'd never be able to tell
- 19:59based on these limited financials.
- 20:01And then at the end of the day, and and you have been someone
- 20:04who says, you know, Uber loses money, it's always going to lose
- 20:07money. And now Uber is making money
- 20:10like Uber's business has rationalized.
- 20:12It has. It is cash flow positive, but
- 20:15it's like you know. I did read your Uber's thing and
- 20:18your journalism was excellent then.
- 20:19Like it's actually I think how we first started talking like
- 20:22it's still like in not saying it's not now.
- 20:24By the way, you're one of the, you've written actually some
- 20:26very good critical AI pieces. I 100% believe that you're
- 20:30coming up this well, I don't think Uber is a comparison point
- 20:34just because it this would be like if every Uber ride cost 40
- 20:37grand and it ran on gold. Like the economics are very
- 20:40different. And on top of that, Uber's kind
- 20:42of a useful. They actually really did the too
- 20:44big to fail and they also did it within the Zerp era.
- 20:47I don't know if if Uber had released, it'd be a weird world
- 20:49if they had, but if they released in like 2021 I don't
- 20:52know. Being very negative about Uber,
- 20:54right? Like I mean this.
- 20:55I have. Have I said that?
- 20:56Uber, were you? Wrong or?
- 20:58Did I say that Uber would die? I think you said something like
- 21:02it will be like lose money forever or something.
- 21:06I stand by that. I think they kind of are going
- 21:08to lose money forever. I think they're going to dodge
- 21:10by on the thinnest possible you you've written about this
- 21:13yourself, the weird little EBITDA dance they do they're.
- 21:17Also, but they're they're they're they're past that Ed
- 21:18now. I mean they are a cash flow.
- 21:19Positive. Yeah, they are.
- 21:20Investors are valuing for them I.
- 21:23Don't even think Uber's that great of a company to be clear.
- 21:25Like I think Ubering company. Oranges though.
- 21:27Yeah, no, I agree with you. Require you to buy?
- 21:29Yeah, hundreds of billions of cars to make it work.
- 21:32Does Uber burn that much? Like Uber burns money because
- 21:35it's a massive marketing machine?
- 21:36Like it's like Groupon on wheels.
- 21:38It's fucking ridiculous. But it's it exists and it has a
- 21:41product you can point to. I'm still trying to work out
- 21:44what the comparison point is other than burning money.
- 21:47The point is that their investors have reasons to
- 21:51believe that burning money in the early days will result in
- 21:56profits down the road. I agree that there's like an
- 21:59interesting question, like where we agree with the criticism of
- 22:03models and we've talked about a lot on this podcast is it's not
- 22:06clear whether all this spending will will be a Moat right?
- 22:10That there you'll create all this value and then you've
- 22:12worked really hard and then the next person like deep sea can
- 22:15just come along and like copy you.
- 22:17But but to me, me, that's not like a case against the actual
- 22:20value of the thing they're they're producing.
- 22:23It's just a business structure problem of like, will they have
- 22:26the Moat of a technological advantage or will they just lose
- 22:29it? And like, if investors want to
- 22:31roll the dice on a particular business and say, we think
- 22:33they'll be able to keep it, it's almost like good for society.
- 22:36If they're wrong, you know, they they invest money to build a
- 22:39technology that is valuable to us.
- 22:41And if it becomes cheaper because other people can just
- 22:44make it easily, like what? What's the tragedy?
- 22:47The tragedy here is the massive environmental damage, the
- 22:51millions of people having stuff stolen from the, the diversion
- 22:55of start up capital away from any idea that does not have AI.
- 22:58The idea that all AI talent is being driven upwards due to Mark
- 23:02Zuckerberg's obsession with this.
- 23:04The fact that people like Alexander Wang who make
- 23:06genuinely evil companies are winning due to this.
- 23:09And on top of this, that now our economy sits on the back of data
- 23:11centre CapEx, which even if this does well, we'll have to
- 23:15contract because we have a limited amount of space on
- 23:17planet Earth. Like that's the thing, the
- 23:20actual several here, the people being driven insane by ChatGPT,
- 23:24the people who are losing their minds because we've made a yes
- 23:28and machine. The fact that these things have
- 23:30no guardrails on them. OK, that's not true.
- 23:33They have some guardrails on them, but they don't have
- 23:35guardrails enough. And they've been allowed to do
- 23:38basically anything they want. Yeah, there are tons of really
- 23:41bad things. If this was just like a company
- 23:44that they bet a lot of money on, fine, sure.
- 23:46Like I fully agree, like it's good for them to take like
- 23:49that's what venture capital does.
- 23:51However, I think we're in a cargo cult situation where
- 23:54they're trying to play the hits. Every VC is just playing end to
- 23:57Sandman again and again and again.
- 23:59Invest as much. I just.
- 24:00Find the like I get to the you know, Casey Newton got in this
- 24:03fight where he he sort of said you have to be you either have
- 24:07to be like a is me big, but it's bad, or you need to say AI is
- 24:11not going to be big so who cares?
- 24:12And I find the, the, the, you know, quadrant that you occupy
- 24:16where it is not going to be big and it's terrible, like sort of
- 24:20exhausting, Like I, I don't understand.
- 24:21I didn't say terrible. OK.
- 24:24You. Oh, really?
- 24:25Well. No, I've said, I've said, I've
- 24:27said large language. Large language models have
- 24:29specific markets, 50 to 60 Tam. I think that the future of
- 24:32coding LMS will be. Terrible.
- 24:34You know they're going to ruin the environment.
- 24:35They're going. To say the environment.
- 24:36There are these people should be jailed like.
- 24:39Am I wrong? You're just am I wrong about any
- 24:41of sound terrible? If every Uber driver had a gun
- 24:43and they unloaded it into people as they drove by, would that be
- 24:46a good feature if Uber allowed it to exist and if regulation
- 24:49didn't stop it? I'm actually not joking.
- 24:51Are you saying that if there are negative effects, there cannot
- 24:53be positive effects? Because that's the thing, I'm
- 24:56not saying large language models have 0 use cases.
- 24:59That's a mistake I've made a few years ago.
- 25:01I really did. There are specific niche.
- 25:03It's cloud software, it's cloud compute like.
- 25:05There are things that can do. There are some interesting
- 25:07things people can do with code. Can it write in type?
- 25:10No, you can't rely on it. The future of that will be in
- 25:13what is the DG X boxes that NVIDIA they're making like the
- 25:17like, not home, but like researcher grade extremely
- 25:19expensive computers. There is something there.
- 25:21The thing is, man, I don't know what quadrant and also Casey
- 25:24Newton needs to go and read that Kilo Code blog about how
- 25:26inference costs are going up because he said the opposite
- 25:29would happen. Who any Sorry, sorry, Casey
- 25:32Newton said. That the cost of inference.
- 25:34There's a blog from Kilo Code that shows how the cost of
- 25:36inference is going up, because while the prices might be
- 25:38reducing, the amount of inference the coding models are
- 25:40using is actually massive. Yeah, because because of deep,
- 25:43because of learning, because, because of reasoning models and
- 25:45and that suddenly kind of turned inferencing into because.
- 25:48There's more. Training like experience.
- 25:49They're trying to do more stuff, yes, but like the old models,
- 25:53doing the same thing certainly costs less.
- 25:55Except no one wants to use the old models.
- 25:57No one. Wants to advancing so much that
- 26:00people are excited about the new model.
- 26:01Well, let's actually talk about that for a second.
- 26:03It's like a damned if you are do, damned if you don't
- 26:05situation like. Every every time.
- 26:07Why is it that I have the negative No, and I mean there's
- 26:10not an attack on you just. Because you're you're, you're
- 26:12opining on AI like when you're. But I don't feel like this level
- 26:17of rigor, and I'm not talking about you here.
- 26:19I don't feel like this level of rigor and criticism and analysis
- 26:23is ever put on the positive side to make them prove themselves.
- 26:26Indeed, I don't. You can say damned if I'm.
- 26:29They've never had to sing for their supper.
- 26:31They've never had to really sit there and explain what their
- 26:34product does. You yourself even said it's for
- 26:36users to work out why. Like they're worth 300 billion.
- 26:40That's more than the market cap of Salesforce, $500 billion more
- 26:44than the market cap or around the market cap of Netflix.
- 26:47Yeah, they do actually kind of have to justify themselves.
- 26:50If they are worth that much, then yeah, I actually would like
- 26:53an explanation. Well, but you're talking about
- 26:55different justifications here because you know, there's
- 26:57justification to investors and then justification to clients,
- 27:00to customers about whether this this software has any utility.
- 27:03And and I think you are seeing a huge amount of resistance from
- 27:06larger enterprises at least over whether they really want to
- 27:08adopt this stuff. Like, you know, Steve, Laura had
- 27:10a pretty interesting piece that came out today that basically
- 27:14said, I mean, you know, and this was not really a case fully
- 27:17against AI, but basically said we're in like the through of
- 27:19disillusionment right now when it comes to the number of
- 27:22products that are being, you know, abandoned by enterprises
- 27:26right now, you know, initiatives that they end up abandoning.
- 27:28It's like 42% or something. So I actually don't really think
- 27:31these guys are not out there not not being called to by the CIOs
- 27:35of all these prospective customers to say like what is
- 27:37the value this has here? Even Tyler Cowen had a piece
- 27:40about this that was ultimately optimistic on AI, but that the
- 27:44utility point of where these corporate customers and buyers
- 27:47hadn't, you know, fully adopted it yet was simply just because
- 27:50it was too early. It's not like this
- 27:53transformation should take a long time.
- 27:56It's only been three years. AI yeah, it's been very fat.
- 28:00You do like cloud. There are banks that still don't
- 28:03embrace the cloud like there are plenty.
- 28:04Of 2017 was attention all you need it's actually been much
- 28:07longer and on. Top of that.
- 28:09You've never had this focused capital. 2022 and like never
- 28:13just catching on to in 2023. You've never had this focused
- 28:16capital in investment history. You have never had all the
- 28:20King's horses and all the King's men do this.
- 28:22It's you have most of the talent and most of the money diverted
- 28:26to focus on one thing. You've got start-ups, you've got
- 28:29giant companies, you've got everyone looking at this and
- 28:32this is what we've got. What is diverted?
- 28:34I think this is actually a key point of your argument that I
- 28:36understand. Like what?
- 28:38What do you want the money to go to instead?
- 28:40I guess like the way. Where else?
- 28:42Capitalism works is like this is the most exciting idea people
- 28:45have going. Like if they had another great
- 28:47idea, they could return a bunch of money.
- 28:49They they are happy to hear you out.
- 28:50Like what is the better? What is the thing?
- 28:53What is the thing you want them to spend the money on?
- 28:55I mean. This.
- 28:56Anything else? I mean, genuinely, how about a
- 28:59product that makes more money than it costs?
- 29:02How about, I don't know, something that helps niche
- 29:05industries? That's the thing, I'm not.
- 29:08Like the the biology AI movement.
- 29:10You know, like AI for drugs. Which one?
- 29:12Which I guess in some ways. I think that's cool.
- 29:14I think there are tons of AI things that.
- 29:16Build in this space, yeah. I think that those are not all
- 29:19large language model products, and indeed, if large language
- 29:22models make money and do that, that's sick.
- 29:25That would be cool. I'm not saying that I can't find
- 29:28a thing I like, I'm just yet to be remotely impressed by this.
- 29:32And again, it comes from the fact that every really, I really
- 29:36encourage you to go and look, look at the early coverage of
- 29:39this and then follow the coverage year by year and see
- 29:42how vague this has been and how ridiculous the pressure has been
- 29:46on people. Look outside of the tech media
- 29:48into like the Business Media, into industrial media,
- 29:51educational media, put the pressure on teachers from 20/22.
- 29:55It is absolutely ridiculous the amount of pressure put on
- 29:58society, the amount of pressure from bosses telling people
- 30:02you've got to do AI. It's the same crap they were
- 30:04doing with return to office. It is the same thing.
- 30:08And you and you same. And you hold the media
- 30:10responsible for that for a lot of that.
- 30:11Partially, I think that there is also a societal failure in
- 30:15general. I think that the markets are not
- 30:17connected to real outcomes and I think that that rewards a
- 30:21certain kind of burn. I also think that there is a
- 30:23there is a level here of non like it's not malicious
- 30:27necessarily. It's like venture capital did
- 30:28what venture capital would do. Big tech did what big tech do,
- 30:31open AI did what any startup would do.
- 30:33I don't think they set out being like we, we are going to do a
- 30:35big fake thing and we will get everyone to.
- 30:38But I don't think that's it at all.
- 30:39I think that this is very much something that ran out.
- 30:42It kind of got out of control versus a malignant thing where
- 30:45Sam Alton was sitting, you know, Finally, no, I think it's just a
- 30:49runaway train. Yeah, no.
- 30:51And I think an argument in that favour is the fact that ChatGPT
- 30:54as a product was kind of a mistaken success.
- 30:56You know they put it out on a lark, pissing off everyone at
- 30:59Microsoft. Because they wanted to see.
- 31:02Yeah, but yeah. Isn't that like an argument for
- 31:04this not being some marketing coup?
- 31:06It was like. But it is saying is that it is.
- 31:08What they had and people were excited about it.
- 31:11No, but but I think what he's saying is it it got out of
- 31:13control because it grew out of this pace that is ultimately
- 31:16unsustainable or based largely on hype.
- 31:18And so you know, the products utility doesn't really need a
- 31:20type, even though so there were. Other companies that did what
- 31:23ChatGPT did before them, that was like, I think Jasper beat
- 31:26them to market. They didn't have they didn't
- 31:28have Sam Altman's cachet. And I mean that.
- 31:31And that really is just start-ups that I'm not going to
- 31:34be pissy about. The product market thing can
- 31:35happen in lots of different ways, I mean.
- 31:37Yeah, and it happens all the time.
- 31:38It's just like, yeah, it was a marketing thing and it was
- 31:41something that people got legitimately excited about at
- 31:44first. I'm not saying any of that was
- 31:45illegitimate. I actually think that the
- 31:47original thing that it did was kind of interesting.
- 31:49It's like, good. I remember at the time it didn't
- 31:51really get a ton more interesting and indeed the
- 31:54things it can do did not dramatically change.
- 31:58And I even think that the launch of GPT 5 when you look at the
- 32:01craziness around 4 O it reminds me and someone on blue sky.
- 32:05At this point it's like it's like a live service game.
- 32:08People freaked out like gamers do.
- 32:09Being like you took away the thing and now you've got people
- 32:12on Reddit saying 4 O isn't the same, which is just it's full
- 32:16but it suggests that there is not a product people are
- 32:18attached to but a feeling the thing it makes the it's almost
- 32:23like it. Isn't that like, OK, again, I
- 32:25think this fits in this sort of. It's either very successful at
- 32:29being bad or bad at being successful and you can sort of
- 32:32switch back and forth. I'm not saying it's not popular,
- 32:35I'm just saying that the it's popularity is kind of a mirage.
- 32:39First of all, the numbers that Open AI are giving us are
- 32:42questionable, and I've repeatedly asked them to give
- 32:45clarity over what a weekly user is, whether they have monthly
- 32:49active users. They have revealed their annual
- 32:51recurring revenue once to CNBC on June 3rd I believe, and then
- 32:55leaked twice within 48 hours to two different outlets when they
- 32:58hit 12 and 13. That alone suggests that the way
- 33:01they calculate annualized revenue is weird.
- 33:03The thing is, I do think that media pressure and I do think
- 33:06societal pressure can lead to this thing and the fact that
- 33:09there are no other companies that really show any kind of
- 33:12similar revenue scale, Even big tech really isn't.
- 33:16So this can be popular. This can be used by a lot of
- 33:20people, but there are no after effects that suggest that the
- 33:23large language model is popular. The thing, the headless beast
- 33:27ChatGPT is popular, but why is it popular?
- 33:31What do people, they use it as search, They use it to
- 33:33brainstorm and whatever. I think that the therapy thing
- 33:36is very worrying, as is the friendship thing, but
- 33:38nevertheless, it's the thing that people use it for.
- 33:40But when it comes down to the actual utility, when it comes
- 33:43down to the thing that makes this worth this much money,
- 33:45that's where my that's where the wheels come off for me.
- 33:47Even the coding stuff, right. That's what I thought that that
- 33:50might be a huge revenue thing, but 400 million ARR from Claude
- 33:54code according to the information, that's stinky.
- 33:57That's stinky, stinky. I mean, that's over four months
- 34:00though. I mean like they just.
- 34:01Lost. Yeah, and it's gonna.
- 34:01And they're now rate limiting heavily.
- 34:04These are some of the fastest growing products by revenue
- 34:06like. And then they will crash because
- 34:09cursor you want to talk. Cursor I'd love to talk.
- 34:11Cursor, I'd love to talk. Cursor.
- 34:12We can talk cursor. I'd love to talk Cursor.
- 34:14Anyway, I think Cursor is the best earns of this entire
- 34:17situation they released. Do you know anything that
- 34:20happened with cursors? Price change?
- 34:22Yeah. I wrote a whole piece about it
- 34:23on Sunday. Wonderful, I should read that
- 34:25shit. I'm so sorry.
- 34:25It's okay, it's fine. Anyway, have you noticed?
- 34:28The afterwards. Are you familiar with the auto
- 34:30feature on cursor? Yes, yes, sure.
- 34:32Not how. Unlimited, Yes, not unlimited.
- 34:34It's starting in September. They're taking that cursor has
- 34:37had to change their entire product.
- 34:39They've had to change everything and now even the unlimited part
- 34:41isn't unlimited. That 400, sorry, 500 million
- 34:45annualized figure was from May, I want to say.
- 34:47It's from June, they're above that now.
- 34:49Aren't they? Yeah.
- 34:50How long are they going to tell you about that?
- 34:52What's the churn? Because The thing is that
- 34:54product has dramatically changed.
- 34:56It's no longer the same thing. I read their forums and they
- 34:59read it all the time. They are in September
- 35:02dramatically changing again. They're basically making it so
- 35:05you will be hard stopped on cursor like you are with Claude
- 35:07code. Yeah.
- 35:08Well, their margins are gross margins were negative.
- 35:10I mean it was a nightmare for them.
- 35:12They had these sinkhole customers that were using all
- 35:15these agentic tools that were burning all these.
- 35:17They basically were propping up Anthropics business more than
- 35:2010% at times of Anthropics revenue comes from is that.
- 35:24True. Yeah, Yeah.
- 35:26All a newcomer.co, sure. I need to save him another
- 35:29negative point. He's like.
- 35:30I'll add that to my quiver of negative.
- 35:33I will, Eric. The important thing here is that
- 35:35my work gets more widely distributed.
- 35:37That's what, that's my contribution.
- 35:39That's. The questions about cursor too,
- 35:41we report it. I mean, I'm not a super, I mean,
- 35:44I'm fairly bullish about the, you know, what's happening with
- 35:46language models overall, but obviously we believe in asking
- 35:49questions and quibbling with things.
- 35:51So we're supportive with you on that.
- 35:52I still think the bulls, I wish, I wish 20% of what they said was
- 35:57negative and you sort of the bear, I wish, you know, there
- 36:00was 20% that you would offer that's positive.
- 36:03And I'm not clear what you're arguing for.
- 36:05Sometimes, I mean. I'm arguing for us to stop
- 36:08pretending that it's something it isn't.
- 36:10I'm arguing for people to stop saying that the agents work.
- 36:14I'm arguing for people to stop pretending the open AI is a
- 36:17sustainable company. Same deal with Anthropic.
- 36:19For people to admit the economics are rotten and also
- 36:22start talking about the fact that the money isn't big, not
- 36:25just. But like I mean you were you
- 36:26were negative on like Facebook and I think like 2022, which was
- 36:30like the low point for the stock.
- 36:31The stock is like 7X from there and like the map public.
- 36:35Stocks are different to product. Producing profits, they make
- 36:39their generate products, things like a.
- 36:41Big asshole. I'm so sorry.
- 36:44Are we talking about? Are we talking about companies
- 36:47do you believe in? Capitalism.
- 36:49Yes, Oh my God, it exists. Just a sincere question, not to
- 36:52like. I believe that the current state
- 36:54of capitalism with with rot economics, the growth of all
- 36:56costs ecosystem is going to drive the tech industry into the
- 36:59goddamn toiler. I think it is going to end up
- 37:02hurting far more people and it goes all the way back to 2011
- 37:05with the case of the fat startup software is eating the world was
- 37:08not a good thing for startups. All it did was attempt to remove
- 37:11the connection from value from the value of the stock.
- 37:14That was the purpose of software is eating the world.
- 37:17That is my problem. The the Bay Area that was about
- 37:20creating valuable software that you sell to people and then you
- 37:24sell to a big company for a profit or you take public is
- 37:27dead. It is absolutely dead as one of.
- 37:30The in the AI world right now is the cool calling card is to do
- 37:34something with as few employees as possible.
- 37:36Like there there was this period of bloat, but I do think a lot
- 37:41people are trying to like skip funny rounds.
- 37:42People are trying to hire fewer people.
- 37:44Like I do think some of the values of the current AI boom
- 37:47are in sort of lean companies, obviously.
- 37:51And do you think the economics will always be?
- 37:54There for that because I hope so I hope that someone finds a
- 37:57thing to do with large language models that's actually
- 37:59profitable while not destroying the environment.
- 38:01I really there are interesting things Simon Willis and Max Wolf
- 38:05they post cool things all the time.
- 38:07I can't do them, can't code, but like, I recognize that there are
- 38:09people doing stuff for that, but there's this massive other story
- 38:13I don't ever want to read about fucking agents the rest of my
- 38:16phase until they can. Until like the ideally early
- 38:19yes, it still doesn't value like through these things while
- 38:22they're. Look at the coverage of Open AI.
- 38:24I mean, is deep research an agent like deep research?
- 38:27Man, no it's not. Good.
- 38:28Like a consulting memo. I'm sure you're like, well
- 38:30consultants are idiots but like if you want to deliver like a
- 38:33consulting memo level product, you could get it out of deep
- 38:36research. Can you like cat?
- 38:39Like is that something like here's the thing.
- 38:41I've read a few articles about McKinsey being scared of AI.
- 38:44There's never really any numbers or tangible.
- 38:48There's always like, well, this is obviously happening.
- 38:50But when you go and look for tangible examples, in fact,
- 38:53throughout this whole boom, they're just not there.
- 38:56And that's my but that's the thing.
- 38:57If this was being fairly evaluated, great.
- 39:00It isn't. And I'm roasted like a prize hog
- 39:04for daring to step outside here, for daring to say this.
- 39:08But they did not receive this level of scrutiny on the way up.
- 39:11They have not really received this level of scrutiny.
- 39:14And if I'm right, there was a bloody reason to and that this
- 39:17is kind of a giant mirage. And it's scary because putting
- 39:21aside my personal feelings, the the economic effects here and to
- 39:26the effects for startups, because the thing I said
- 39:28earlier, I've set this in the AI as a money trap.
- 39:31Putting aside feelings about the industry, we currently have a
- 39:33crisis with valuations. Who is going to buy Cognition?
- 39:36Who is going to buy Cohere? Who is going to buy Cursor?
- 39:39Is Cursor going to sell for $20 billion?
- 39:41OK, fine. Will they actually sell for the
- 39:43IP though? Who is going to buy these
- 39:45things? Why did Cognition by the remains
- 39:47of Windsurf? Was it to add that 782,000,000
- 39:51ARR? Because condition does not have
- 39:53revenue. What are these companies doing?
- 39:55What happens next? How do they exit?
- 39:58Because remember in 2021 a bunch of crap money got tied up in
- 40:01crapper companies and LP's got pissy.
- 40:04LP's came flooding back in with AI.
- 40:07How do you think they're going to feel when there's no exit?
- 40:10Because there really isn't right now.
- 40:11Inflection wasn't a real 1 character.
- 40:14That AI wasn't the shareholders. The investors are getting cash
- 40:17back because of those. Which investors?
- 40:19So they are some. Some are, some are.
- 40:20Investors and companies like Inflection Character, they're
- 40:24getting money back as part of those deals.
- 40:26But they they are not getting the massive returns that would
- 40:28justify like. That's the thing.
- 40:30It's. Some of them are like 2X and
- 40:32pretty big amounts of money. I mean it depends on the.
- 40:34DOX is that really like from the again, if this was just a
- 40:39regular thing, If this is a regular thing where I was not
- 40:42having conversations with people that get animated about it.
- 40:45If it was just a regular stuff something, hell yeah, 2X3X cool.
- 40:49But the weird thing is, is everyone's talking about this is
- 40:51the next big thing. When you look at the look at the
- 40:54actual things. Why didn't these companies get
- 40:56acquired properly? Why did they do these acqui hire
- 41:00thingies? That's the financial.
- 41:02There's a whole, there's a regulatory whole angle.
- 41:04I will say there is a whole. Regular there's a regulatory.
- 41:06Angle that happen. Won't regulate like you think.
- 41:09They're actually Microsoft's inflection deal got put through
- 41:11regulators anyway, like and it passed.
- 41:13No, I mean, it did, yeah. It's.
- 41:15Used to avoid regulation. It it did after the fact, but
- 41:18the deal sailed through. I mean, you know, it was just
- 41:20basically a licensing deal. I mean, that's the reason these
- 41:22things caught fire is because. We did the deal.
- 41:24These things are legal. Tricks to get around regulation.
- 41:27Yeah, these aren't the exits people want.
- 41:29But I'm saying right, people are making money off them and an
- 41:32open AI is selling like secondary shares for employees
- 41:34at like 500, like an insane number.
- 41:37Like yeah, it's like hard to even say.
- 41:41I'm like, Oh my God, isn't like. Worth this much or so?
- 41:44But like I think I. Think to make as an argument
- 41:46here, I think what's going on here with these like kind of,
- 41:48you know, mild outcomes for the second tier startups is that
- 41:51they're all kind of predicate on the belief that there is going
- 41:53to be 1, maybe 2 actual huge outcomes.
- 41:55And it's probably open AI and it's anthropic.
- 41:57And if those do, if and if they don't turn out to be anything,
- 42:00then yeah, the whole thing does. Probably.
- 42:01I mean, that was a big thing in my benchmark piece right there
- 42:03that I just published. You know, it's like, oh, they're
- 42:05making a bunch of infrastructure investments, but they don't have
- 42:07a big foundation model. And it's very possible there's a
- 42:09classic home run distribution. But to me that argument is
- 42:12paired with open AI is going to, you know, win everything in the
- 42:14Super value. Actually this is a really good
- 42:16question though. What happens if open AI doesn't
- 42:19convert to a non profit and doesn't receive the money from
- 42:21exactly? Well, I agree we are aligned on
- 42:22that. You know I wrote a.
- 42:23Bare thing. About open AI predicated on on
- 42:25that. So I agree with you on that.
- 42:27Banger And The thing is, I don't actually think we disagree on a
- 42:31ton of, I think we have, you have a lot of hopes for this,
- 42:34but it seems like the actual economics, you kind of see where
- 42:36I'm coming from. And I think I think that it's
- 42:39good that Someone Like You, it has the bet, has the bull
- 42:41argument, but still acknowledges these things because whoever's
- 42:45right, at least these arguments need to be given.
- 42:47And I will say someone needs to read Softbank's earnings like I
- 42:52have though, because SoftBank does not remember when SoftBank
- 42:55said they were going to be spend $3 billion in open AI in back in
- 42:59February. SoftBank does not know how
- 43:01that's working. Crystal Intelligence is yet to
- 43:03pop the whole like there are straight up bullshit things
- 43:07happening right and. Yeah, we're very interested in
- 43:10SoftBank Rounder as well. We we agree with some of your
- 43:13skepticism. But I I have a thing though with
- 43:16this, which is, putting aside how you think it may or not go,
- 43:19what happen if Open AI doesn't raise their money?
- 43:21They're dead. They are dead because they can't
- 43:24go public. And they also they had.
- 43:26To you mean if they don't convert?
- 43:27You mean if they don't convert to APB?
- 43:29Yeah, if they don't convert, yeah, it's game over.
- 43:31And I think open AI is a systemic risk because open AI
- 43:34dies, Cor Weave dies. Crusoe has built nothing for
- 43:37nothing. Oracle has sunk all this money
- 43:39and JP Morgan Blue Owl primary data infrastructure has sunk all
- 43:42this money into data centres for nobody.
- 43:44They'll maybe resell them. I have some people at Oracle
- 43:47telling me that they could find customers theoretically, but
- 43:51this, all of this money is laid out for one company.
- 43:54And if they don't convert and also SoftBank needs to get $22.5
- 43:58billion, they don't have that money.
- 44:00Yes, they, they had a great quarter, whatever, but they
- 44:03don't have the cash and they have a basic loan LTV thing
- 44:07where it's like they can't go over 25%.
- 44:10They don't have even the space to do that.
- 44:12They're saying it will come out of vision fund.
- 44:14It's. People may lose them money, but
- 44:16there is some minimum price for like opening out.
- 44:19Like people see the value, they're producing revenue.
- 44:21So I just don't think, I just disagree that there's like some
- 44:23existential risk here. It's like, yes.
- 44:27There is. I just, they're, they're not
- 44:28going to die because like, even if it falls to 30 billion,
- 44:31because you only get this tiny sliver of it and the nonprofit
- 44:33owns tons of it, there's still more money that would come in at
- 44:37some price to keep the thing. I mean, I think as long as it is
- 44:41the leader in large language models and people are seeing
- 44:43tremendous value in language models, then yeah.
- 44:46Which I think the tremendous value.
- 44:49I think so that's that's the core area.
- 44:51We disagree that it is the number one app in the App Store
- 44:54is the main general purpose assistant to find it for me.
- 44:59But that was one. But that was one this.
- 45:01One time. It's a sustained #1 this isn't
- 45:03like a, you know, a podcaster getting the high up once and
- 45:06using the number over and over again.
- 45:08This is a sustained in the App Store.
- 45:10I mean worst case. I will I will say, Eric, like,
- 45:12like there are there are a warning as someone who covered
- 45:14Snapchat for many years, like you can, you can be #1 in the
- 45:17App Store and not really make that great of a business
- 45:19overall. So these things are somewhat if
- 45:20every. Customer loses you money,
- 45:22especially when you're burning money the moment anyone touches
- 45:26your software. Like that's the thing, gravity
- 45:28exists so sure you could. Well actually here's the thing.
- 45:32I get what your point is that if they are the leader they are
- 45:35perceived as leader. They could keep raising money
- 45:37but why do people invest money in companies?
- 45:39It's so they can be an exit. The idea here is without the
- 45:43nonprofit conversion, open AI has no liquidity route other
- 45:46than what amounts to a Ponzi scheme.
- 45:48It just you are selling investor money to other investors at
- 45:52increasing amounts I imagine maybe.
- 45:54An element and it's there are now all these secondary
- 45:56transactions, so it's like private companies.
- 45:58Or so that's the Ponzi scheme aspect of it.
- 46:00And I mean, how do you keep that going if they could they be?
- 46:03Public, but you know, Stripe, all these these companies are
- 46:06selling there are there are exits involved like new people
- 46:09are buying old people. So you're just saying that Let's
- 46:12just say the scheme requires money to keep entering the
- 46:17system, so that by exiting money that's already entered the
- 46:20system, and if the money entering the system stops, then
- 46:24the business will the scheme in this case will die.
- 46:26That is, it's like if nobody buys, you know, Microsoft
- 46:29tomorrow, it'll fall to 0. Like this is just how stocks
- 46:32work, as long as they're. Microsoft makes more money than
- 46:34they lose billions of dollars. If there are transactions on the
- 46:38secondary markets where big private companies are able to
- 46:41sell large chunks of stock to someone else, that's a price
- 46:44setting. Mechanism.
- 46:46Are you suggesting the open AI would continue selling stock if
- 46:50it couldn't go public? Is your suggestion just so I
- 46:52understand the open AI, they don't go, they don't convert, so
- 46:55they are perpetually A nonprofit.
- 46:57You were suggesting that they would survive by just
- 46:59continually selling stock. No, They need some structure
- 47:03that's sustainable where they can say this is how it's going
- 47:05to work. This is where the profits go.
- 47:08This entity owns this percent of profits.
- 47:10I'm investing in this. Like, I think there needs to be
- 47:13some resolution to that question, though.
- 47:15I think people clearly will have a lot of energy to invest.
- 47:19Yeah, without that certainty. On the conversion aspect to it,
- 47:22Ed, the reason I, I have a hard time getting too worked up about
- 47:25that is that we're not talking about some sort of technical
- 47:27breakthrough. We're not talking about AGI or,
- 47:30you know, them achieving some level of, you know, breakthrough
- 47:32it in the language model. This is a political bureaucratic
- 47:36transaction, right? This is about difficult it is to
- 47:39convert. Sure, I've written lots of
- 47:40stories about this. I've written lots of stories
- 47:42about this. I mean, essentially their entire
- 47:44fate rests in the hands of the AG in Delaware and California
- 47:47and to a smaller. Degree the IRS.
- 47:50Well, the IRS is, is downstream of all of this.
- 47:52I mean, this is mostly an AG decision and also Microsoft
- 47:55because it's about their larger contract over their, you know,
- 47:57relationship with open AI, right.
- 47:59I think Microsoft has a vested interest in ensuring that open
- 48:02AI doesn't collapse because their stock price is based on.
- 48:04Yeah, because their stock price is.
- 48:05Based on AP, they own all the infrastructure they.
- 48:07Just, but they also get, they also get cloud revenue from
- 48:09usage of Chachi BT I mean, if, if open AI disappear tomorrow,
- 48:12their story of an AI being an AI stock would, would, would
- 48:15evaporate overnight. They can't sell copilot.
- 48:17You know, they, they, the Mustafa's team hasn't built
- 48:19shit. So they absolutely need open AI
- 48:21to exist. But what, what at least for the
- 48:25next couple of years. But look, I don't even know.
- 48:28I, I can't pretend I know how that one's going to going to
- 48:30play out on the, on the California AG front.
- 48:33I think there's going to be a huge fight with the nonprofits
- 48:35within the state and who even knows, you know, who becomes the
- 48:38next governor of California? If that's Katie Porter, if it's
- 48:40I don't even know who else not not Kamala Harris.
- 48:43You know, I think one of the major things that will come up
- 48:45is how hard are you going to be on open AI?
- 48:48I think California's as a anti business state is something very
- 48:51problematic to people here. And the idea that you would have
- 48:54fucked over the most promising company by not allowing to
- 48:56convert into a for profit would be a huge problem.
- 49:00That. I mean actually we do that many.
- 49:01Votes that you just like get big enough and like the regulators
- 49:05like solve the problem problem for you and.
- 49:06I can't get that worked. Up even though I'm somewhat
- 49:09cynical about it. Same thing will happen for open
- 49:12AI. I would make I.
- 49:14Disagree with you, but I can see how you got there.
- 49:17I wanna make a sort of different point on the sort of
- 49:19overvaluation of AI startups that again parallels this like
- 49:22Uber trauma experience. It's just like SoftBank overpaid
- 49:28for Uber, which effectively meant a bunch of Middle Eastern
- 49:30money and Saudi money overpaid for Uber.
- 49:33The valuation corrected and eventually went back up.
- 49:35Like a lot of the money that's flowing in now is these big like
- 49:39Middle Eastern sovereign wealth. And like some of my big question
- 49:42to you like what is the more productive thing that this money
- 49:45could go to? Like fundamentally what's
- 49:47happening is countries have made a lot of money on oil.
- 49:50They don't have any better ideas about what to do with the money.
- 49:53So they throw it at AI and that's a.
- 49:55Hard. Line.
- 49:56Against the line. Like what I?
- 49:59Said There's a lot of very interesting cities that you
- 50:01could be building in Abu Dhabi and, and, and Riyadh right now.
- 50:04But I think you're overlooking. For a long quab is like mega
- 50:07city infrastructure projects, but they're they're basically
- 50:10saying, OK, the most productive thing going is an AI.
- 50:13And so worst case scenario, like America is getting a bunch of
- 50:16mispriced money, oil money into our technology companies.
- 50:21And maybe it's slightly overpriced, but I, I just don't
- 50:23think the money is being burned. Like it's largely American money
- 50:26is being marked up by oil money. And like, what's the tragedy
- 50:30from the American perspective? First of all, I think it's not
- 50:33an argument to say what else this should be spent on
- 50:36considering I already established that they're out of
- 50:38ideas. I do not know if there is
- 50:39another hyper growth market. I believe venture capital needs
- 50:42to actually look downstream and invest more money and seed and
- 50:46startups in general. Most of the money goes B and up
- 50:49if not C and up. It has historically done that.
- 50:51It's fucking stupid and antithetical to what Silicon
- 50:54Valley stands for. On top of that, I don't agree
- 50:57that most of this money is oil money.
- 50:58In fact, a remarkable amount of this money is coming through
- 51:01sources like Magnetar Capital or NVIDIA or Microsoft or Amazon.
- 51:06Right, There's a lot of money. Flowing back Andreessen Horowitz
- 51:09back to them, right? That's right.
- 51:11Okay, so you're just saying, So what is the thing you're saying
- 51:15that it's good this money's No, I'm saying like.
- 51:17This I'm saying valuation, even if these companies were all
- 51:20overvalued by 50%, I don't think it's like a catastrophe like.
- 51:25Either I think it is a. With huge sums of money that
- 51:28don't know where else they want to invest it.
- 51:30It's companies that benefit from the money coming back to them.
- 51:34Like you're saying like I agree with what you're saying.
- 51:36I just, I don't think that's Catas.
- 51:38There's a catastrophe. Evaluations fall so.
- 51:41I know, I know we're approaching time.
- 51:43So I think I can actually answer this fairly rigorously.
- 51:45So the catastrophe is the following in 2021 there was way
- 51:49too much money put in crap LP's got scared, start up money
- 51:52really slowed. It came back, it came back
- 51:55thanks to AI. The problem is that I actually I
- 51:57think the money invested here is stupider and the promises are
- 52:00more egregious, Right? Hear me out.
- 52:02No. No, no.
- 52:03Yeah, You're in this. This is a good ending, Yeah.
- 52:06Your feelings about what will won't happen.
- 52:08I don't think these companies are worth enough.
- 52:11I don't think that. I think they have the exit
- 52:13liquidity and think they're sustainable.
- 52:15As a result, a lot of venture capital is going to get burned,
- 52:18which means a lot of LP's are going to get burned, which is
- 52:21going to largely stymie the amount of money going into the
- 52:23startup ecosystem. On top of that, if NVIDIA has
- 52:27any problems with growth from here to oblivion, because they
- 52:29have to, it needs to be like what, 50 billion next quarter in
- 52:33data center revenue? Even though data center revenue
- 52:35missed estimates last quarter, it needs to keep going what in
- 52:38three years we meant to be selling $120 billion of GPU's
- 52:41because that's what the market expects.
- 52:43All of these expectations have become load bearing.
- 52:46They become load bearing for Silicon Valley and they become
- 52:49load bearing for the markets. The problem is there is no way
- 52:52to match these expectations. So at some point, there is going
- 52:56to be a nasty hangover. And my fear is it's going to hit
- 52:59regular people in the markets, a ton of retail money in NVIDIA
- 53:04too. And on top of that, it's going
- 53:05to stop start-ups getting money in general.
- 53:08And I think, you know, I think we need more money in angels
- 53:10anyway. I've always, always felt that
- 53:12way. It's like, I think that it's
- 53:14better to invest in earlier companies than otherwise.
- 53:17But I'm worried now that we're going to have a repeat of
- 53:202022-2023 ass layoffs. Fuck all money for startups.
- 53:24A lot of cynicism, and I don't mean cynicism like me, I mean
- 53:27just straight up, I don't want to put money in tech and when
- 53:30tech doesn't have a hyper growth thing after AI, I'm terrified
- 53:33that it's going to just start slashing tech valuations across
- 53:36the board. However you feel about
- 53:37generative AI, my arguments are founded on like real things.
- 53:41And I mentioned cognition, for example, the evaluation.
- 53:44It's like, how's that? And how does any of this?
- 53:46And does open AI really go public?
- 53:48Does anthropic go public? Do you think that their S 1 is
- 53:51going to look like anything other than a Jackson Pollock
- 53:53painting? Like it's, it's rough out there
- 53:56and I'm actually, I don't know how people are so bearish right
- 54:00now. I don't know how I get why
- 54:02people aren't like completely insane and swagged up like me,
- 54:06but I don't, I don't get why more people aren't worried about
- 54:11it because this doesn't even match previous things.
- 54:14And I'm just, I have real concerns and the people who are
- 54:17going to get hurt are founders and real people and who have
- 54:20invested in the market. It's got their downstream
- 54:23effects are rough. The Silicon Valley consensus is
- 54:25that Angel and seed rounds are overvalued.
- 54:28Now, maybe you disagree, but it's just funny to me to hear it
- 54:30from you. I think it's.
- 54:32I think it's. Far more, I think it's far more
- 54:35reasonable to overvalue at the seed other than fucking thinking
- 54:39machines. And that's a super safe
- 54:41intelligence. I think those are ludicrous.
- 54:43But I think it's OK to take bets at the seed and the Series A
- 54:46stage. That's that's fine.
- 54:48Like, come on. Like, yeah, you take the piss a
- 54:50bit there because you have hopes and you get to these overblown
- 54:53valuations. Open AI stage, anthropic stage.
- 54:56They should be sustainable and I think that's a reasonable thing
- 54:58to ask for. All right.
- 54:59Well, look, I mean, the beautiful thing about where
- 55:01we're at right now is that we don't know who's right.
- 55:04We have no idea what's going to happen.
- 55:06The future remains. To be written.
- 55:08Yeah, if it is written, it could be written by AI and we, we know
- 55:11that it's very sycophant, think it will make us feel very smart
- 55:14about what we asked it. But Ed, thank you so much for
- 55:16joining. I have a feeling we should do
- 55:18this more often. I would love to thanks for I
- 55:21know this has been quite I've really had a good time.
- 55:23No, we. Coming a little.
- 55:26Bit I actually pay for new coat like I have quite sometimes.
- 55:28We enjoy your sub stack. So yeah, thank you very much.
- 55:31Thank. You so much.
- 55:32Yeah. All right.
- 55:33Thanks.