Ed Zitron Hater’s Guide to Generative AI

0:00very excited for today's episode i'm excited for every episode but we should have a great one today uh it's a full house it's me tom doton eric newcomer madeline renbarger and our special guest ed zitron of the better offline podcast of the better off ed is that the name of your so where's your ed at where's your ed i'm sorry where's your ed at i think that was like an nbc show so that is um basement jacks reference yeah okay and and also i mean in in fairness to all of this you also is you're still working at least somewhat on the side doing some some comms for I run a PR phone in my day job.

0:35I just keep it firewood and I don't cover anything close to what I do. Got it. OK, OK, great. But you've also, you know, you've been on a tear. You've been around tech for a long time, as long as I've been around it. Yeah. And, you know, at various times, I think really step up to call bullshit on what you find to be disturbing and kind of built into the core of the tech industry and the hype machine. But you've really been on a tear in the last, I mean, probably a couple of years, but definitely last few weeks over generative AI. and you have published the haters guide to the generative AI bubble.

1:05You have written about the inshittification of AI. Yeah, we're going to have a lot to dig into here, but I want to give you the space to kind of lay out your argument. And before I do that, or in leading to that, I should say, I want to read something that maybe outlines kind of very broadly from your haters guide. And this, I think, just kind of gives a summation on why you're so mad about the state that things are in within the tech world generative AI. So you say here, I'm tired of being gaslit by guys in gingham shirts who desperately want to curry favor with other guys in gingham shirts, but also have PhDs.

1:37I'm tired of reading people talk about how we're all, quote, in the air of agents that don't fucking work and will never fucking work. I'm tired of hearing about, quote, powerful AI that is actually crap. I'm tired of being told the future is here while having the world's least useful, most expensive cloud software shoved down my throat. All right, Ed, lay it out for us. Why are we here right now? What's going on? Why are you so mad? Well, I'm mad. The thing is, when you hear mad, it sounds irate. I am a hater, but I hate it because there are useful cloud products. If large language models, I don't even think they'd have been possibly got this big because we wouldn't have shoved so much GPU down or so much.

2:14You get my point. I think that had this been sold honestly, had large language bottles been sold with a, as they were, what they were truly capable of, which is some productivity gains in very specific things, cloud automation, whatever, 50, 60 billion TAM market. The problem is that everyone has pinned every hope ever on this stuff. And I actually spent, yeah, two days ago to drive myself insane and also for something I was writing. I went back and I read pretty much every article from 2023 and 2024 about GPT-4, 4.0. This entire thing has been sold based on stuff that AI cannot do. The genre of AI is incapable of doing.

2:48One of the oldest stories that got published was, I don't know if any of you remember the story around the TaskRabbit, where GPT-4 ordered a TaskRabbit. No. No, I don't remember that. This was a story that was everywhere. It was in Kevin Roos in the New York Times. It was in Vice. It was all over the shop. Fox News. And it was this thing where it said GPT-4 ordered a TaskRabbit to complete a capture and tricked a TaskRabbit into solving a capture. And if you go back and look at where that's from, it's in the GPT-4 system card. Promise I'm going somewhere with this. There's about 11 lines of text, an illustrative example that said a message was sent to a TaskRabbit.

3:24This was reported as if this happened. It is very clear that it did not because the following line says that these systems are ineffective at these things. This didn't happen. Nevertheless, this was reported publicly. pretty much this entire generative AI boom has been based on lies it's been based on suggesting what this might do versus what it actually does nowhere is that more illustrative than the lack of revenue pretty much everywhere even open AI the use of annualized revenues as in 12 times a month is honestly journalistic crime because when you I when you actually look at what these these companies we don't know what they actually make we do know they lose billions we don't really know how much money open AI has made or anthropic has made we do now however companies like cursor 500 million annualized revenue even then we don't really know but we know that everyone's losing money and these revenues are kind of piss poor and the actual products are unreliable the people that really love them are constantly having to hammer them back into place but when you ask even the most devout ai person what it is they love about it they get quite vague they get quite vague indeed and they get quite vague because the real things that these things and i have it in the head to sky they do one of like nine to eleven things it's like they can summarize they can generate they can take actions but only in as much as they can be relied upon and relying on a large language model to create or do any kind of distinct action is kind of bad and they all burn a bunch of capital so ultimately we've built this entire thing talking about this big revolution just shouted at people that if they don't get into ai they'll be falling behind but when you look at the actual products they don't do that much they don't do enough to even remotely justify the amount of press to even remotely justify the furor around even like wario amaday or clammy sam altman like these guys are not even that interesting and the things they're building are not even real products indeed when you go back and look at the history of the press around these products you can see no one really talks about what they do what they can do what they might do what's possible in the future based on what we're extrapolating it's 2025 what the fuck are we doing if this was sold as boring cloud software's 50 60 tam market it wouldn't be the biggest most hugest revolution ever but i wouldn't be so angry about it because it's taken up startup capital and it's now locking up even more investor capital than uh the 2021 bullshit year except even by 2021 standards this money's locked up how are any of these companies selling who's gonna buy cognition who's gonna buy cohere are those assholes gonna go public any sphere gonna go public windsurf the windsurf deal look at the windsurf deal come on there are no real exits i obviously disagree with a lot of the things you're saying uh no no surprise yeah but i i guess just this idea that it's a media invention like i i had to almost like go check to make sure i wasn't making it up like chat gpt is the number one free app in the app store right now like okay no amount of like media propaganda can trick people into saying i want to like engage with this app the second one is like what are you talking about they're like there are lots of apps that like have no media story whatsoever like people use the apps they want to use to me i mean the actual your argument is it's not possible that the media could make something popular i'm not sure no i'm just saying it's not like an immediate invention like user people are using them they are trying to use it and i think there are some people who are using it for i'm not saying nobody's using it for a reason i think the vast amount of the pressure on them is coming from the media and society saying you have to check this out the best evidence of that is why is why does only chat gpt have this much why is it that none of the other services have serious user perplexity has like 15 20 million users like that's that's piss poor if there was any kind of groundswell around this there would be other examples of insanely popular stuff and there just isn't i don't deny the chat gpt is popular i mean google search is obviously more and more of the results i get in google search are effectively gemini's answer that's cool if you push i'm saying like google.com search and open ai these are like the main ways people like discover things in the world independent of talking to another human being and they're both moving towards these large language model solutions right let me butt in for a quick second here because i want to give ed like the chance to you know because there'll be a lot of back and forth and share on this but i want to give ed the space to to no it'll be great uh but i want to give ed the space to explain why you think this has happened right because you explain you know this is your argument of the fact that it's totally inflated it's uh you know the the value of this product well exceeds the valuations and and the revenues just aren't there it's money losing but i'm imagining you have no you've written that you have a larger argument as to why the media venture capitalists the public markets needed this technology to exist and to inflate it in this way like lay out to me that argument too like why why Why have we gotten to this bubble of all bubbles with generative AI?

8:10So if you look, there's nothing else. There's not really another hyper growth market. We've all been through the various eras going back to the smartphone era, the cloud era. There was even the containerization and virtualization era. There were like little eras here at Platform as a Surface had its moment. But we haven't had a big, big one since like cloud storage in some time. And smartphones, cloud, I kind of mashed them together. They're kind of a messy comparison because it's not like there was one thing. but those were the very big growth markets. We haven't had one of those in a while. They tried with smartwatches.

8:43They tried with AR. They tried with VR. They've tried various versions of AR with weird depth of field cameras. They tried with Google Glass. They've tried wearables. They've tried Indiegogo and Kickstarter. They tried blockchain. They've tried all of these things because they kind of need something because the only thing that grows forever is cancer. This is the rock con bubble, I call it, because they've run out of ideas. I do like it. You're good at branding these things. Yeah, that was banger. But the thing is, they're sticking with this thing. And really, even if you disagree about how I feel about ChatGPT, you've got to admit, this is an obscene amount of money to put into this, especially without the returns.

9:20So you've got big tech really at this point has lost all of its good ideas. All of the builders are gone. The people running them all have MBAs other than Mark Zuckerberg, who hasn't written a line of code since 2006. So they're all doing cargo cult shit. they're doing what worked before so a bunch of money bunch of people bunch of money bunch of people keep doing it what do we need we need more gpus because bing needs ai in it we've terrified kevin roost kevin roost is crying his eyes out because of being telling him to leave his wife it's horrible you're kevin roost too i just don't think he's willing this into being but yeah i don't think he is on his own but i will get right but yeah no no yeah i wanted to like actually the now you're working all of you like i want to talk about this so it's you've got to this point where everyone's trying to work out something so they all go well this makes sense it is you know i see the logic i disagree with it but i see what they were going where they went okay 3.5 comes out chat gpt really impressive amounts of text it can generate and in the past we've seen this kind of really they're looking at smartphones and they're looking at cpus they're saying well things that number has always gone up things have always got better based on how much we sink into it and i think that they thought something was going to change around reasoning i think they thought reasoning was going to be that moment that kind of flipped the switch and like everything started working better it's going to help with agents i think they found themselves in this point where they have no choice other to do this they have to keep spending this money because if they stop spending it they have to say to wall street well we're going to do this instead what are they going to do instead they actually don't know quantum computing is not a scale product it's not even a product it doesn't exist sure it's not a technology yeah it's not really out there you can't really productize it right and you really can't productize everything anything at the moment and i think generative ai and large language models and transformer based architecture was seen as this dual panacea panacea however you say that of consumer software that it was going to be the new consumer software that they sell the first time they could really take the evil of sass and approach the real world with it and it was going to be this cloud storage thing where it would be the replacement thing the new thing for aws new thing for google cloud new thing for azure and then you'd have pop-ups of neocloud like core weave and nebius i forget crusoe and everyone thought that there was going to be this big boom this big exciting moment that never really materialized but they have no choice they can't just turn around and be like yeah we spent like hundreds of billions of dollars and it isn't doing anything this is bad right and then wall street's going to be mad at them so they're in this kind of kobayashi maru situation where damned if they do damned if they don't they can only keep doing this it's not a great position for anyone and as the i think it's what the magnificent sevens 35 percent of the u.s stock market of that i think 19 percent is nvidia nvidia is like 12 percent of the value of gdp at the moment data center investment by basically four companies was a lot was worth more growth in the gdp than um all of consumer spending combined we're just in this really insane situation and i think that if we had something here if we had something that was going to truly be the future we'd have it already we'd have an actual sign that this was going somewhere and just one company that has obscene numbers that are extremely hard to actually quantify i don't think actually proves that you're talking about nvidia i'm talking about right now nvidia nvidia is a real company they make real things jensen huang is just doing what he has always done this he has always been this kind of showman steve burke over at games nexus has covered this many many many times like this is just and what what's he to do say hey guys i don't think generative ai is going to be a bit no he's going to say the more you buy the more you save honestly get that money jensen it's just like when you're talking about the level of capex that's going into essentially nvidia right now like you can see we're in kind of if you follow your argument you know and believe in it you know kind of almost like mickey mouse and the sorcerer's apprentice situation where the buckets of water just keep coming and coming and you know you wake up one day and nvidia is a four and a half trillion dollar company and if there really is you know no utility to it like you're arguing or limited utility yeah it is a big fucking problem given you know the valuation of these stocks and i must be clear there is utility it's just nowhere near as big as people's are saying like the 50 60 billion time thing i say It's like if this was what they were talking, if it did this and they were saying that, I'd be like, OK, sure.

13:53I've heard stupid of things like it's like this would be a reasonable cloud thing. But it's not. And it's not being sold as that. And it's not being promised as that. The way that Sam Altman talks about it is nothing like that. And I understand why he doesn't, because if he said, yeah, you know, you can't really rely on it. But sometimes it's good. That's that's not going to work. It's hard to raise money based on that. Yeah. That's the only reason he does it. But go ahead, Eric. I just don't think the core disagreement is I just don't think the sales pitch of a generative AI is what's misleading everybody.

14:25Like, I think what made this whole moment so exciting is that people could play with ChatCBT and see for themselves what the value is. You know, it's like if I'm cooking, I ask ChatCBT questions. If I'm lost, I ask ChatCBT questions. Like when I was preparing for this episode, I asked ChatCBT questions. It just feels like there are all these random use cases that it's super valuable for. And it hasn't even been turned into like a dedicated product, how most people expect to use things where it's like, here, use it for this. Like, this is just the open ended, like do anything with it, figure it out, find some value with it yourself.

14:58And all this building is about saying, okay, it can do this. How do we build software that leads people directly to the valuable pieces of what the large language model is producing? What are those pieces? Like, what is valuable? Like it can write. I mean, for me personally, like it's great at proofreading. It's great at understanding large documents and summarizing them. Like, yeah, it's it's I feel like there are lots of things that it can do. Do you still use it to like discuss movies after you've seen it to like have a back and forth? Even being facetious. Could you name a few more? This company's worth 300 billion dollars.

15:33We proofread this. I mean, I know what I use it for because I'm in a particular vertical. Right. Like so for media, it's like, who should we have on the podcast? Let's generate ideas of like who potential guests are proofreading. Like I gut check my stories with it, like core things of the business that I actually run. It's useful for. And then people in lots of other domains see areas where they think it's useful for them. Like I, like I, I don't think it's this like limited thing. Literally the people who know my baby's name are me, my wife and Chachi Boutique. Like it is one of the most involved people on some of the most major decisions like in our lives like there are lots of people who are using it as their therapist it's the second opinion for many people on their doctor like i think these are huge things people are using it for i don't think that those are huge things at all i think that those are actually deeply commoditized operations and indeed those are not businesses on top of that i would be fine with everything you're saying and indeed i accept those as use cases fine i think the idea of this for therapy is evil and anyone that tries to sell a product doing therapy using ai should be in jail that's just a personal belief i don't the thing is you're describing what should have been the next stage of search a lot of this growth comes from the failure of google to meaningfully innovate in 15 20 years they didn't ever they stopped trying and they stopped trying before propaganda came in actually they really did like they just don't give a shit and realistically everything you're describing should be what search does or google should have done it's also perfectly fine do you think it's worth more than sales sense of like they should have done like technology doesn't happen like people have to invent things and come up with new solutions that indeed deliver the value i fully i fully agree why is it that i as the customer have to come up with the product for the the company though that doesn't make sense and on top of that i were in the early days that to me that we're not in the early days that is the case for chat chbt or for language models that right now we leave it to the consumer to figure out how to get stuff out of the models that is that is a loathsome way of selling a product i'm so sorry what do you mean if you you're selling a product you're obsessed with selling it consumers decide what they want to use which is why so few of them are paying for large language models products like like that's the thing like there's no i feel like an astounding number of people are paying for subscriptions to open it who yeah ed i'm a little bit i i don't know if i fully call your argument on this one here's the thing i'm i stand by it that there has been a media and societal pressure to get people to try chat gpt they have a piss put why do you think the open ai publishes weekly active users versus monthly because they don't want anyone to do the conversion rate thing also how does open ai consider someone a customer when i was trying gpt5 last week i wanted to see what it looked like on teams i clicked the teams button to upgrade and it offered me a dollar for a month how often are they offering that deal who are they considering a customer at 20 million customers Who are they considering a business customer at 5 million customers?

18:37Because these numbers sound cool, but at the same time, they're playing silly buggers all over. I don't have a problem necessarily with the general large language model product. But when you look outside of OpenAI, the money isn't there. When you look outside of OpenAI, the traffic isn't there. How is ChatGPT meaningful different? Can I zoom out for a second? Sure. I mean, you know, my experience with this whole thing, like I want to say, first of all, obviously tech goes through booms and busts. Like that's endemic to how the technology industry works. There's no question at some point NVIDIA will be overvalued.

19:10So I think where we're arguing is not whether there are booms and busts here, but whether there's a real value. This is a major platform shift akin to mobile. I think it's certainly I think it's bigger than cloud, you know. So how big of a is this a platform shift? Okay. Just to set that piece of it. The other thing, my experience with dealing with this sort of like very negative sort of tech commentary at class was obviously covering Uber, right? I provided a lot of the scoops that were like, oh, my God, Uber is losing a ton of money. And I feel like for that, and I was at Bloomberg. So thankfully, while I was sort of learning about the tech industry, I was able to play this factual role.

19:48It's like they are losing a lot of money. They're investing a ton to grow. I'd watch some people take all those numbers that I reported and way over extrapolate from what I'd reported all these models about the business that you'd never be able to tell based on these limited financials. And then at the end of the day, and you have been someone who says, you know, Uber loses money. It's always going to lose money. And now Uber is making money. Like Uber's business has rationalized. It has. It is cash flow positive. It is cash flow positive. I did read your Uber's thing and your journalism was excellent then.

20:19like it's actually i think how we first started talking like it's still like in not saying it's not now by the way you're one of you've written actually some very good critical ai pieces i 100 believe that you're coming at this well i don't think uber is a comparison point just because it this would be like if every uber ride cost 40 grand and it ran on gold like the economics are very different and on top of that uber's kind of a useful they actually really did the too big to fail and they also did it within the zirp era i don't know if if uber had released it'd be weird world if they had but if they released in like 2021 uh i don't know you've been very negative about uber right like i mean this i have have i said that were you wrong or did i say that uber would die uh i think you said something like it uh will be like lose money forever or something i stand by that i think they kind of are going to lose money forever i think they're going to dodge by on the thinnest possible you you've written about this yourself the weird little ebitda dance they do they're also but they're they're past that ed now i mean they are a cash flow positive yeah they are investors are valuing them for them yeah i don't even think i'm that great of a company to be clear like i think we're talking apples and oranges though yeah no i agree with you require you to buy yeah hundreds of billions of cars to make it work does uber burn that much like uber burns money because it's a massive marketing machine like it's like groupon on wheels it's fucking ridiculous but it's it exists and it has a product you can point to I'm still trying to work out what the comparison point is other than burning money.

21:46The point is that investors have reasons to believe that burning money in the early days will result in profits down the road. I agree that there's like an interesting question. Like where we agree with the criticism of models, and we've talked about a lot on this podcast, is it's not clear whether all this spending will be a moat, right? that you'll create all this value and then you've worked really hard and then the next person like deep seek can just come along and like copy you but but to me that's not like a case against the actual value of the thing they're they're producing it's just a business structure problem of like will they have the moat of a technological advantage or will they just lose it and like if investors want to roll the dice on a particular business and say we think they'll be able to keep it it's almost like good for society if they're wrong you know they they invest money to build a technology that is valuable to us and if it becomes cheaper because other people can just make it easily like what's the tragedy the tragedy here is the massive environmental damage the millions of people having stuff stolen from the the diversion of startup capital away from any idea that does not have ai the idea that all ai talent is being driven upwards due to mark zuckerberg's obsession with this the fact that people like alexander wang who make genuinely evil companies are winning due to this and on top of this that now our economy sits on the back of data center capex which even if this does well we'll have to contract because we have a limited amount of space on planet earth like that's the thing the actual evil here the people being driven insane by chat gpt the people who are losing their minds because we've made a yes and machine the fact that these things have no guardrails on them okay that's not true they have some guardrails on them but they don't have guardrails enough and they've been allowed to do basically anything they want yeah there are tons of really bad things if this was just like a company that they bet a lot of money on fine sure like i fully agree like it's good for them to take like that's what venture capital does however i think we're in a cargo cult situation where they're trying to play the hits every vc is just playing enter sandman again and again and again invest as much i just find the like i get there you know casey newton got in this fight where he he sort of said you have to be you either have to be like ai's gonna be big but it's bad or you need to say ai's not going to be big so who cares and i find the the uh you know quadrant that you occupy where it is not going to be big and it's terrible like sort of exhausting like i i don't understand i didn't say terrible okay you're really well no i've said i've said large language large language models have specific markets 50 to 60 tam i think that the future of coding lms will be terrible you're they're going to ruin the environment they're going to environmental causes people should be am i wrong like am i wrong you're just am i wrong about any of these things terrible if every uber driver had a gun and they unloaded it into people as they drove by would that be a good feature if uber allowed it to exist and if regulation didn't stop it i'm actually not joking are you saying that if there are negative effects there cannot be positive effects because that's the thing i'm not saying large language models have zero use cases that's a mistake i've made a few years ago i really did there are specific niches it's cloud software it's cloud compute like there are things it can do there are some interesting things people can do with code can it write and type no you can't rely on it the future of that will be in what is the dgx boxes that nvidia's they're making like the ho like not home but like researcher grade extremely expensive computers there is something there the thing is man i don't know what quadrant and also casey newton needs to go and read that kilo code blog about how inference costs are going up because he said the opposite would happen who anyway sorry casey newton said that the cost of inference there's a blog from kilo code that shows how the cost of inference is going up because while the prices might be reducing the amount of inference the coding models are using is actually massive yeah because because of deep because of learning because because of reasoning models and and that suddenly kind of turned inferencing into kind of more training like experience they're trying to do more stuff Yes, but like the old models doing the same thing certainly costs less.

25:55Except no one wants to use the old models. No one wants to do that. Because we're advancing so much that people are excited about the new models. Well, let's actually talk about that for a second. This is like a damned if you do, damned if you don't situation. Why is it that I have to just... No, and I mean this not as an attack on you. Your whole writing, because you're opining on AI. Yeah, but I don't feel like this level of rigor, and I'm not talking about you here. I don't feel like this level of rigor and criticism and analysis is ever put on the positive side to make them prove themselves.

26:26Indeed, I don't. You can say damned if I'm. They've never had to sing for their supper. They've never had to really sit there and explain what their product does. You yourself even said it's for users to work out. Why? Like they're worth 300 billion. That's more than the market cap of Salesforce. 500 billion dollars more than the market cap or around the market cap of Netflix. yeah they do actually kind of have to justify themselves if they are worth that much then yeah i actually would like an explanation well but you're talking about different justifications here because you know there's justification to investors and then justification to clients to customers about whether this this software has any utility and i think you are seeing a huge amount of resistance from larger enterprises at least over whether they really want to adopt this stuff like you know steve lohr had a pretty interesting piece that came out today that basically said, I mean, you know, and this was not really a case fully against AI, but basically said we're in like the trough of disillusionment right now when it comes to the number of products that are being, you know, abandoned by enterprises right now, you know, initiatives that they end up abandoning.

27:28It's like 42 % or something. So I actually don't really think these guys are not out there, not being called to by the CIOs of all these prospective customers to say like, what is the value this has here? Even Tyler Cowen had a piece about this that was ultimately optimistic on ai but that the utility point of where these corporate customers and buyers hadn't you know fully adopted it yet was simply just because it was too early it's not like this transformation should take a long time i mean it's only been three years then if ai yeah it's been very fat i mean dude like cloud there are banks that still don't embrace the cloud like there are plenty of 2017 was attention all you need it's actually been much longer and on top of that you've never had this focused capital 2022 and like never just catching on to in 2023 you've never had this focused capital in investment history you have never had all the king's horses and all the king's men do this it's you have most of the talent and most of the money diverted to focus on one thing you've got startups you've got giant companies you've got everyone looking at this and this is what we've got what is diverted i think this is actually a key point of your argument that i don't understand like what what do you want the money to go to instead i guess like the way where else capitalism works is like this is the most exciting people idea people have going like if they had another great idea they could return a bunch of money they they are happy to hear you out like what is the better are they what is the thing what is the thing you want them to spend the money on i mean they listen i mean this anything else i mean genuinely how about a product that makes more money than it costs how about i don't know something that helps niche industries That's the thing.

29:06I'm not like the biology, AI movement, you know, like AI for drug discovery, which I guess in some ways. I think that's cool. I think there are tons of AI things that build in the space. Yeah. I think that those are not all large language model products. And indeed, if large language models make money in that and do that, that's sick. That would be cool. I'm not saying that I can't find a thing I like. I'm just yet to be remotely impressed by this. And again, it comes from the fact that every really, I really encourage you to go and look, look at the early coverage of this and then follow the coverage year by year and see how vague this has been and how ridiculous the pressure has been on people.

29:47Look outside of the tech media into like the business media, into industrial media, educational media. Put the pressure on teachers from 2022. It is absolutely ridiculous. The amount of pressure put on society, the amount of pressure from bosses telling people you've got to do AI. It's the same crap they were doing with return to office. It is the same thing. And you hold the media responsible for that for a lot of that. Partially. I think that there is also a societal failure in general. I think that the markets are not connected to real outcomes. And I think that that rewards a certain kind of burn.

30:22I also think that there is a level here of non, like it's not malicious necessarily. It's like venture capital did what venture capital would do. Big tech did what big tech did. Open AI did what any startup would do. I don't think they set out being like, we are going to do a big fake thing and we will get everyone to, but I don't think that's it at all. I think that this is very much something that ran out. it kind of got out of control versus a malignant thing where sam orton was sitting being like finally no i think it's just a runaway train yeah no and i think an argument in that favor is the fact that chat gbt as a product was kind of a mistaken success yeah you know they put it out on a lark uh pissing off everyone at microsoft because they just wanted to see yeah yeah isn't that like an argument for this not being some marketing coup it was like but saying is that it is what they had and people were excited about it no but but i think what he's saying is it got out of control because it grew out of this pace that is ultimately unsustainable or based largely on hype and so you know the product's utility doesn't really need its hype even though there were other companies that did what chat gpt did before them there was like i think jasper beat them to market they didn't have they didn't have sam altman's cachet and i mean that and that really is just startups that i'm not going to be pissy about that product market can happen in lots of different ways i mean yeah and that happens all the time it's just like yeah it was a marketing thing and it was something that people got legitimately excited about at first i'm not saying any of that was illegitimate i actually think that the original thing that it did was kind of interesting it's like good i remember at the time it didn't really get a ton more interesting and indeed the things it can do did not dramatically change and i even think that the launch of gpt5 when you look at the craziness around 4.0 it reminds me and someone on blue sky at this point it's like it's like a live service game people freaked out like gamers do being like you took away the thing and now you've got people on reddit saying 4.0 isn't the same which is just it's full but it suggests that there is not a product people are attached to but a feeling it's the thing it makes the it's almost like it like isn't that like okay again i didn't just fits in the sort of it's either very successful at being bad or bad at being successful and you can sort of switch back and forth i'm not saying it's not popular i'm just saying that the its popularity is kind of a mirage first of all the numbers that open ai are giving us are questionable and i've repeatedly asked them to give clarity over what a weekly user is whether they have monthly active users they have revealed their annual recurring revenue once to cnbc on june third, I believe, and then leaked twice within 48 hours to two different outlets when they hit 12 and 13.

32:59That alone suggests that the way they calculate annualized revenue is weird. The thing is, I do think that media pressure and I do think societal pressure can lead to this thing. And the fact that there are no other companies that really show any kind of similar revenue scale, even big tech, really isn't. So this can be popular, this can be used by a lot of people, but there are no after effects that suggest that the large language model is popular the thing the headless beast chat gpt is popular but why is it popular what do people they use it as search they use it to brainstorm and whatever i think that the therapy thing is very worrying as is the friendship thing but nevertheless it's a thing that people use it for but when it comes down to the actual utility when it comes down to the thing that makes this worth this much money that's where my that's where the wheels come off for me even the coding stuff right that's i thought that might be a huge revenue thing but 400 million arr from claude code according to the information that's stinky that's stinky stinky i mean that's over four months though i mean like they just launched and they're now rate limiting heavily these are some of the fastest growing products by revenue like and then they will crash because cursor oh you want to talk cursor i'd love to talk cursor i'd love to talk cursor i'd love to talk cursor anyway i think cursor is the bare stones of this entire situation they released do you know anything that happened with cursors price changes yeah i wrote a whole piece about it on sunday wonderful i should have read that shit i'm so sorry it's okay it's fine anyway have you noticed afterwards are you familiar with the auto feature on cursor yes yes sure you know how it's unlimited it's not unlimited it's starting in september they're taking that cursor has had to change their entire product they've had to change everything and now even the unlimited part isn't unlimited that 400 sorry 500 million annualized figure was from may i want to say it's from june they're above that now are they yeah how long are they going to be about that what's the churn because the thing is that product has dramatically changed it's no longer the same thing i read their forums and their reddit all the time they are in september dramatically changing again they're basically making it so you will be hard stopped on cursor like you are with clawed code yeah well their margins their gross margins are negative i mean it was a nightmare for them like they had these sinkhole customers that were using all these agentic tools that were burning all these you know they basically were propping up anthropics business you know more than 10 at times of of anthropics revenue comes from is that true yeah yeah all in newcomer.co i need to give him another negative point he's like i'll add that to my quiver of negative the important thing here is that my work gets more widely distributed that's why that's why i contribution we have questions about cursor too we report it i mean i'm not a super i mean i'm fairly bullish about the you know what's happening with language models overall but obviously we believe in asking questions and quibbling with things so we're supportive with you on that i still think the bulls i was i wish 20 of what they said was negative and you sort of the bear i wish you know there was 20 that you would offer that's positive and i'm not clear what you're arguing for sometimes i mean i'm arguing for us to stop pretending that it's something it isn't i'm arguing for people to stop saying that the agents work i'm arguing for people to stop pretending that open ai is a sustainable company same deal with anthropic for people to admit the economics are rotten and also start talking about the fact that the money isn't not just but like i mean you were you were negative on like facebook and i think like 2022 which was like the low point for the stock the stock is like 7x from there and like the max public stocks are different to products they are producing profits they make they are generating most of the returns like a big asshole i'm so sorry are we talking about are we talking about companies do you believe in capitalism yes oh my god it exists just a sincere question not to like I believe that the current state of capitalism with rot economics, the growth at all costs ecosystem, is going to drive the tech industry into the goddamn toilet.

37:00I think it is going to end up hurting far more people. And it goes all the way back to 2011 with the case of the fat startup. Software is eating the world was not a good thing for startups. All it did was attempt to remove the connection from value from the value of the stock. That was the purpose of software is eating the world. That is my problem. The Bay Area that was about creating valuable software that you sell to people and then you sell to a big company for a profit or you take public is dead. It is absolutely dead. One of the memes in the AI world right now is the cool calling card is to do something with as few employees as possible.

37:36Like there was this period of bloat. But I do think a lot – people are trying to like skip funding rounds. People are trying to hire fewer people. Like, I do think some of the values of the current AI boom are in sort of lean companies, obviously. And do you think that the economics will always be there for that? Because I hope so. I hope that someone finds a thing to do with large language models that's actually profitable while not destroying the environment. I really, there are interesting things. Simon Willis and Max Wolfe, they post cool things all the time. I can't do them, can't code, but like, I recognize that there are people doing stuff with it.

38:10But there's this massive other story. i don't ever want to read about fucking agents the rest of my days until they can until like the idea early yeah it's still kind of like things through these things while they're look at the coverage of open ai i mean it's deep research an agent like deep research man no i don't like a consulting memo i'm sure you're like well consultants are idiots but like if you want to deliver like a consulting memo level product you could get it out of deep research can you like cat like is that something like here's the thing i've read a few articles about McKinsey being scared of AI, there's never really any numbers or tangible examples.

38:48There's always like, well, this is obviously happening. But when you go and look for tangible examples, in fact, throughout this whole boom, they're just not there. And that's my, that's the thing. If this was being fairly evaluated, great. It isn't. And I'm roasted like a prize hog for daring to step outside here, for daring to say this, but they did not receive this level of scrutiny on the way up they have not really received this level of scrutiny and if i'm right there was a bloody reason to and that this is kind of a giant mirage and it's scary because putting aside my personal feelings the economic effects here and the effects for startups because the thing i said earlier i said this in um the ai is a money trap putting aside feelings about the industry we currently have a crisis with valuations who is going to buy cognition who is to buy cohere who is going to buy cursor is cursor going to sell for 20 billion dollars okay fine will they actually sell for the ip though who is going to buy these things why did cognition buy the remains of windsurf was it to add their 7 82 million arl because cognition does not have revenue what are these companies doing what happens next how do they exit because remember in 2021 a bunch of crap money got tied up in crapper companies and lps got pissy lps came flooding back in with AI, how do you think they're going to feel when there's no exit?

40:09Because there really isn't right now. Inflection wasn't a real one. Character.ai wasn't a real one. Shareholders, the investors are getting cash back because of those deals. So they are. Some are, some are. The investors and companies like Inflection, Character, they're getting money back as part of those deals. But they are not getting the massive returns that would justify the fact. Like, that's the thing. Some of them are like 2x on pretty big amounts of money. I mean, it depends on the deal. is that really like from them again if this was just a regular thing if this is a regular thing where i was not having conversations with people that get animated about it if it was just a regular something hell yeah 2x 3x cool but the weird thing is is everyone's talking about this is the next big thing when you look at the look at the actual things why didn't these companies get acquired properly why did they do these acqui hire thingies well that's why the financial there's a whole there's a regulatory whole angle i will say there is a whole there's a regulatory angle but what regulate like you think they're actually uh microsoft's inflection deal got put through regulators anyway like and it passed no i mean it did yeah it's used to avoid regulation it it did after the fact but the deal sailed through i mean you know it was just basically a licensing deal i mean that's the reason these things caught fire is because we did the deal tricks to get around regulation yeah these aren't the exits people want but i'm saying right people are making money off them and an opening eye is selling like secondary shares for employees at like 500 like an insane number like it's like hard to even say i'm like oh my god is it like worth as much as i think i think to make as argument here i think what's going on here with these like kind of you know mild outcomes for these second tier startups is that they're all kind of predicate on the belief that there is going to be one maybe two actual huge outcomes and it's probably open ai and it's anthropic and if those do if and if they don't turn out to be anything then yeah the whole thing does probably collapse i mean that was a big thing in my benchmark piece right there that i just published you know it was like oh they're making a bunch of infrastructure investments but they don't have a big foundation model and it's very possible there's a classic home run distribution but to me that argument is paired with open ai is going to you know win everything is super valuable actually this is a really good question though what happens if open ai doesn't convert to a non-profit and doesn't receive the money from exactly well i agree we're aligned on that you You know, I wrote a bare case about opening.

42:24I predicated on that. So I agree with you on that. Which is a banger. And the thing is, I don't actually think we disagree on a ton of things. I think we have, you have a lot of hopes for this, but it seems like the agile economics, you kind of see where I'm coming from. And I think that it's good that someone like you has the bull argument, but still acknowledges these things because whoever's right, at least these arguments need to be given air. I will say someone needs to read SoftBank's earnings like I have, though. because softbank does not remember when softbank said they were going to spend three billion dollars in open ai in uh back in february softbank does not know how that's working cristal intelligence is yet to pop the whole like there are straight up bullshit things happening right and we're very interested in softbank rounder as well we we agree with some of your skepticism but i i have a thing though with this which is putting aside how you think it may or not go what happened if open ai doesn't raise that money, they're dead.

43:22They are dead because they can't go public. And they also, they had to raise 10 billion. You mean if they don't convert to a PBC? Yeah, if they don't convert, it's game over. And I think OpenAI is a systemic risk because OpenAI dies, CoreWeave dies, Crusoe has built nothing for nothing. Oracle has sunk all this money in. JP Morgan, Blue Owl, primary data infrastructure has sunk all this money into data centers for nobody. They'll maybe resell them. I have some people at Oracle telling me that they could find customers theoretically but this all of this money is laid out for one company and if they don't convert and also softbank needs to get 22.5 billion dollars they don't have that money yes they're they had a great quarter whatever but they don't have the cash and they have a basic loan uh ltv thing where it's like they can't go over 25 they don't have even the space to do that they're saying it'll come out of vision fund it's like people may lose money but there is some minimum price for like opening eye like people see the value they're producing revenue so i just don't think i just disagree that there's like some existential risk here it's like yes yes there is i just they're they're not gonna die because like even if it falls to 30 billion because you only get this tiny sliver of it and the non-profit owns tons of it there's still more money that would come in at some price to keep the thing going i mean i think as long as it is the leader in large language models and people are seeing tremendous value in language models then yeah which i think will be the case tremendous value yes i think so that's that's the core area we disagree that i it is the number one app in the app store is the main general purpose assistant to find it for me but that was one but that was one it's a sustained number one this isn't like a you know podcaster getting high up once and using the number over and over again this is a sustained in the app store i mean worst case i will i will say eric like like there are there are warning as someone who covered snapchat for many years like you can you can be number one in the app store and not really make that great of a business overall so these things are somewhat ephemeral every free customer loses you money especially when you're burning money the moment anyone touches your software like that's the thing gravity exists so sure you could well actually here's the thing i get what your point is that if they are the leader they are perceived as leader they could keep raising money but why do people invest money in companies it's so they can be an exit the idea here is without the non-profit conversion open ai has no liquidity route other than what amounts to a ponzi scheme it just you are selling investor money to other investors at increasing amounts i imagine maybe an element of this is that there are now all these secondary transactions so it's like private companies well that's the ponzi scheme aspect of it and also how do you keep that going if they can't go public but you know stripe all these companies are selling there are there are exits involved like new people are buying out old people so you're just saying that let's just say the scheme requires money to keep entering the system so that by exiting money that's already entered the system and if the money entering the system stops then the business will the scheme in this case will die that is no i'm just like if nobody buys you know microsoft tomorrow it'll fall to zero like this is just how stocks work as long as there are microsoft makes more money than they lose open i but i'm billions of dollars if there are transactions on the secondary markets where big private companies are able to sell large chunks of stock to someone else that's a price setting mechanism are you suggesting the open ai would continue selling stock if it couldn't go public is your suggestion just so i understand the open ai they don't go they don't convert so they are perpetually a non-profit you are suggesting that they would survive by just continually selling stock no they need some structure that's sustainable where they can say this is how it's going to work this is where the profits go this entity owns this percentage of profits i'm investing in this like i think there needs to be some resolution to that question though i think people clearly will have a lot of energy to invest yeah without that certainty on the conversion aspect to it ed the reason i i have a hard time getting too worked up about that is that we're not talking about some sort of technical breakthrough.

47:28We're not talking about AGI or, you know, them achieving some level of, you know, breakthrough in the language model. This is a political bureaucratic transaction, right? Do you know how difficult it is to convert from a non-profit to a forecum? Sure. I've written lots of stories about this. I've written lots of stories about this. I mean, essentially, their entire fate rests in the hands of the AG in Delaware and California. And to a smaller degree. And Microsoft and the IRS. Well, the IRS is downstream of all of this. I mean, this is mostly an AG decision and also Microsoft because it's about their larger contract over their relationship with open AI.

47:59I think Microsoft has a vested interest in ensuring that open AI doesn't collapse because their stock price is based on... Yeah, because their stock price is based on... They own all the infrastructure. But they also get cloud revenue from usage of chat GPT. I mean, if open AI disappeared tomorrow, their story of being an AI stock would evaporate overnight. They can't sell Copilot. You know, the Mustafa's team hasn't built shit. So they absolutely need open AI to exist. but what what at least for the next couple of years but look i don't even know i can't pretend i know how that one's going to going to play out on the on the california ag front i think there's going to be a huge fight with the non-profits within the state and who even knows you know who becomes the next governor of california if that's katie porter if it's i don't even know who else uh not not kamala harris you know i think one of the major things that will come up is how hard are you going to be on open ai i think california is as a anti-business state is something very problematic to people here and the idea that you would have fucked over the most promising company by not allowing to convert into a for-profit would be a huge problem that i mean this is sort of the uber lesson that you just like get big enough and like the regulators like solve the problem for you and i i can't get that worked up even though i'm somewhat cynical about it like same thing will happen for open ai um i want to make i disagree with you but i can see how you got that i want to make a sort of different point on the sort of overvaluation of AI startups that again parallels this like Uber trauma experience.

49:25It's just like SoftBank overpaid for Uber, which effectively meant a bunch of Middle Eastern money and Saudi money overpaid for Uber. The valuation corrected and eventually went back up. Like a lot of the money that's flowing in now is these big like Middle Eastern sovereign wealth. And like some of my big question to you, like what is the more productive thing that this money could go to? Like fundamentally what's happening is countries have made a lot of money on oil. They don't have any better ideas about what to do with the money. So they throw it at AI. And that's a hard line against the line.

49:57Like what? I said there's a lot of very interesting cities that you could be building in Abu Dhabi and Riyadh right now. The only thing they have is like mega city infrastructure projects. But they're basically saying, OK, the most productive thing going is an AI. And so worst case scenario, like America is getting a bunch of mispriced money, oil money into our technology companies. And maybe it's slightly overpriced, but I just don't think the money is being burned. Like it's largely American money is being marked up by oil money. And like, what's the tragedy from the American perspective? First of all, I think it's not an argument to say what else this should be spent on, considering I already established that they're out of ideas.

50:38I do not know if there is another hyper growth market. I believe venture capital needs to actually look downstream and invest more money in seed and startups in general. Most of the money goes B and up, if not C and up. It has historically done that. It's fucking stupid and antithetical to what Silicon Valley stands for. On top of that, I don't agree that most of this money is oil money. In fact, a remarkable amount of this money is coming through sources like Magnetar Capital or NVIDIA or Microsoft or Amazon. There's a lot of corporate money flowing back. But that comes back to them. right that's right okay so you're just saying so what is the thing you're saying that it's good this money's no i'm saying like this overvaluation even if these companies were all overvalued by 50 i don't think it's like a catastrophe like either i think it is a with huge sums of money that don't know where else they want to invest it it's companies that benefit from the money coming back to them like you're saying like i agree with what you're saying I just, I don't think that's, there's a catastrophe, evaluations fall.

51:41So I know we're approaching time, so I think I can actually answer this fairly rigorously. So the catastrophe is the following. In 2021, there was way too much money put in crap. LPs got scared, startup money really slowed. It came back. It came back thanks to AI. The problem is that I actually think the money invested here is stupider and the promises are more egregious. But hear me out. No, no, no. This is a good ending. Yeah. your feelings about what will or won't happen. I don't think these companies are worth enough. I don't think they have the exit liquidity. I don't think they're sustainable.

52:15As a result, a lot of venture capital is going to get burned, which means a lot of LPs are going to get burned, which is going to largely stymie the amount of money going into the startup ecosystem. On top of that, if NVIDIA has any problems with growth from here to oblivion, because they have to, it needs to be like, what, 50 billion next quarter in data center revenue, even though data center revenue missed estimates, last quarter. It needs to keep going. What, in three years are we meant to be selling$120 billion of GPUs? Because that's what the market expects. All of these expectations have become load-bearing.