The BS Bubble

0:00This is an iHeart Podcast.

0:30Thomson Reuters and specialized bikes have since they upgraded to the next generation of the cloud. Oracle Cloud Infrastructure. OCI is the blazing fast platform for your infrastructure, database, application development, and AI needs, where you can run any workload in a high availability, consistently high performance environment, and spend less than you would with other clouds. How is it faster? OCI's block storage gives you more operations per second. Cheaper? Better? OCI costs up to 50 % less for computing, 70 % less for storage, and 80 % less for networking. Better? In test after test, OCI customers report lower latency and higher bandwidth versus other clouds.

1:12This is the cloud built for AI and all your biggest workloads. Right now, with zero commitment, try OCI for free. Head to oracle.com slash strategic. That's oracle.com slash strategic. Wells Fargo announced a new$20 million program in 2025, teaming up with nonprofits to support small business owners. That's how Wells Fargo is helping strengthen small businesses and communities. Wells Fargo, the bank of doing. Learn more at wellsfargo.com slash say do. Run a business and not thinking about podcasting? Think again. More Americans listen to podcasts than ad-supported streaming music from Spotify and Pandora.

1:51And as the number one podcaster, iHeart's twice as large as the next two combined. Learn how podcasting can help your business. Call 844-844-iHeart.

2:05Hi, I'm Ed Zitron, and you're listening to Better Offline.

2:17Better Offline. As a reminder, you can now buy glorious Better Offline merchandise. There's a link to it in the episode notes. The t-shirts, the tumblers, the tote bags, all that shit. It's lovely. You're going to love it. Buy it today. But if I'm honest today, I'm actually a little bit pissed off. And that's why we've got a two-part episode this week about how fucking stupid the AI boom has become. I wrote farcical in the script. And I'm going to be honest, I need to be a little more pointed. Because I've written tens of thousands of words about this now. I've recorded hours upon hours of podcasts and still to this day, people are babbling about the AI revolution as the sky rains blood and crevices open in the fucking earth, dragging houses and cars and domesticated animals into their moors.

2:57Things are astronomically fucked outside, yet the tech media continues to tell me to get my swimming trunks on and take a long nice dip in the fucking pool. As you can tell, this is going to be a little less reserved than usual. I'm a little bit frustrated. I don't know why I'm the one saying this, and I frequently feel weird that I, a part-time blogger and podcaster, am writing the things that I'm writing. Since I put out the newsletter OpenAI as a systemic risk of the tech industry, and actually a couple of weeks back I did a two-parter about it too, I've heard nothing in response, as was the case with how does OpenAI survive and how OpenAI is a bad business.

3:34There just seems to be little concern or belief that there's any kind of risk at the heart of AI, and OpenAI in particular. And there are companies that spent$9 billion in 2024 to lose$5 billion. Well, I'd love to add a because here, if not because it's important to be intellectually honest and represent views that directly contrast my own, even if I do in somewhat sarcastic and sardonic fashion. Nobody seems to actually have a cogent response to how they write this ship, other than hard fork or Casey Newton throwing a full-scale psycho tantee on a podcast and saying I'm wrong because inference costs are coming down.

4:07inference by the way is when an ai takes an input and produces an output it's the calculations that take place right before google's generative ai assistant attributes a voltaire quote to michael jackson or says that black tar heroin when enjoyed in moderation can help you lose weight newton is a nakedly captured booster that ran an infographic from anthropic a few weeks ago the likes of which i haven't seen since 2013 it was telling you all the ways that people use generative ai it looks like some shit from i don't know early day mashable no offense christina and they essentially treat this company propaganda as gospel but he's really far from the only one with a flimsy attachment to reality the information a publication that genuinely does some great stuff which makes it even more heartbreaking to say this ran a piece in early april that made me even more furious than usual claiming that open ai was forecasting revenue topping 125 billion dollars in 2029 based on selling agents and monetizing free users as a driver to higher revenue.

5:05Agents, I should add, are AI systems that can interact with other systems and do stuff. So an AI that can order pizza from DoorDash for you, that's an example of an agent. And when I say it can order pizza for you, I am talking entirely theoretically, as they cannot do this right now and may never be able to do so. Indeed, the whole agent thing is just what we wish AI was, and it actually doesn't work. And the piece, reported based on things, and I quote, told to some potential and current investors, takes great pains to accept literally everything that OpenAI says is perfectly reasonable, if not gospel, even if said things make absolutely no goddamn sense.

5:41So, according to the information's reporting, OpenAI expects agents and new products, and both of those are quotes, to contribute tens of billions of dollars of revenue, both in the near term, somehow contributing$3 billion in revenue this year, which I will get to in a little bit, and in the long term with an egregious$25 billion in revenue in 2029 even projected to come from just new products. If you're wondering what those new products might be, I am too because the information doesn't seem to know and instead of saying OpenAI has no idea what the fuck they're talking about and is just saying stuff, the outlet continues to publish things with the kind of empty optimism that's indistinguishable from GPT-generated LinkedIn posts.

6:24Must be clear, the information isn't generating their articles, they're writing them fresh. I want to be really, really clear about something. We aren't nearly in May 2025, and indeed one of these will come out actually in May, the second part. I see no evidence that OpenAI even has a marketable agent product they can sell, let alone one that it will make$3 billion goddamn dollars off of. And they definitely are not going to do so in the next six or seven months. Oh my... For context, let's triple the revenue of OpenAI that they made, reportedly at least from selling access to their models via its APIs, essentially allowing third-party companies to use GPT in their apps in the entirety of 2024.

7:05And those APIs and models actually exist in a meaningful sense, as opposed to whatever the fuck OpenAI's half-baked-ass agent stuff is. In fact, no, no, no, no, no, I'm not going to be mean. I'm not. I'm going to be calm. Be normal. I'm going to explain exactly what the information is reporting in an objective way, because writing it out really shows how silly it all sounds. I'm going to rate they believe a lot because I must be clear how stupid this is. Now, according to the information's reporting, they believe that OpenAI will make$3 billion in 2025 from selling access to its agents. This appears to come from SoftBank, which has said it will buy$3 billion worth of OpenAI products annually.

7:41Earlier this year, we got a bit of extra information about how SoftBank will use these products. It plans to create a system called Crystall Intelligence, that's C-R-I-S-T-A-L, and it's one of the most generic names I've ever seen, and it will be a kind of general purpose AI agent platform for big enterprises. The exact specifics of that will shock you in that there are none, but SoftBank intends to use the technology internally across its various portfolio companies as well as market it to other large enterprise companies in Japan. I still do not know what the fuck this is. Crystal Intelligence, billions of dollars, billions of dollars, and they just don't, they can't even describe what it is.

8:20Just saying, yeah, it'll be an agent platform that does stuff with your business. Like, does that sound good? Can I have three billion? I need$40 billion. I need$40 billion. Give me, okay. I also want to add that the information can't seem to keep its story straight on this issue. Back in February, they reported that Open AI would make$3 billion in revenue only from agents with a big, beautiful chart that said $3 billion would come from IT, only to add that IT would be SoftBank using OpenAI's products across its companies. Based on these numbers, it seems like SoftBank will be the only customer for OpenAI's agents.

8:56While this most likely won't be the case, and it isn't because it excludes anyone willing to pay a few bucks to test it out, it nonetheless doesn't signal good things for agents as a mass market product. Not that there were any good signals beforehand, though. Agents do not exist as a product that can be sold at scale. Yes, OpenAI teased Operator its first agent at the start of the year, but it doesn't seem to be able to do anything. The information's own reporting from mid-April highlighted how OpenAI's operator agent struggled with comparison shopping on financial products, and that's a quote, and how operator and other agents are, and I quote again, tripped by pop-ups or log-ins as well as prompts asking for email addresses and phone numbers for marketing purposes, which I think accurately describes most websites.

9:40And just to summarize from everything I've said, the information is saying that the above product will make OpenAI$3 billion by the end of 2025. Sounds very real to me. Sounds extremely real. I love that the business media just prints this. I love this. I love this so much. I'm having so much fun. Jesus Christ. According to the information's reporting, they believe that OpenAI will basically double revenue every single year for the next four years and make$13 billion in revenue in 2025, more than doubling that to$29 billion in 2026, nearly doubling that to$54 billion in 2027, and nearly doubling that again to$86 billion in 2028, and eventually leveling out a ridiculous $125 billion of revenue in 2029.

10:25Said revenue estimates as of 2026 includes billions of dollars of new products that include free monetization, free user monetization either. And if you're wondering what that means, I also am. The information does not explain. Jessica Lesson must have been busy being horrible to people that work for her. They do, however, say that OpenAI will start, and I'm quoting this, won't start generating much revenue from free users and other products until next year. That's 2026. In 2029, and I'm still quoting, however, it projects revenue from free users and other products will reach$25 billion, or one-fifth of all revenue, and then adds that shopping is another potential avenue.

11:04You still probably don't know what they're doing, and neither do I, and I have driven myself insane reading about this. I really cannot express my disgust about how willing publications are to blindly publish projections like these, especially when they're all so stupid. Let me just read this to you, all right? And I quote, OpenAI has already begun experimenting with launching software features for shopping. starting in january some users can access web browsing agent operator as part of their pro chat gpt subscription tier to order groceries from instacart and make restaurant reservations on open table just want to be clear this is a few episodes ago i mentioned casey newton not even being able to like say this worked i just want to be really clear as well what the information is saying so they're saying that this experimental software launched to an indeterminate amount of people that barely works is going to make open ai three billion dollars in 2025 and then somehow this is going to lead to open ai making 29 billion dollars in 2026 and then they're going to eventually be up to 125 billion what the fuck how how what fucking universe are we all living in there's no proof that open ai can do this other than the fact it has a lot of users and a lot of venture capital.

12:23Parking shouldn't slow you down. ParkWiz gives every driver a shortcut. Book ahead, save up to 50 % and skip the hassle of circling the block. Park smarter, park faster. ParkWiz. Download the ParkWiz app today and save every time you park. Run a business and not thinking about podcasting? Think again. More Americans listen to podcasts than ad-supported streaming music from Spotify and Pandora. And as the number one podcaster, iHeart's twice as large as the next two combined. So whatever your customers listen to, they'll hear your message. Plus, only iHeart can extend your message to audiences across broadcast radio.

12:58Think podcasting can help your business? Think iHeart. Streaming, radio, and podcasting. Call 844-844-iHeart to get started. That's 844-844-iHeart.

13:12Jan Marselech was a model of German corporate success. It seemed so damn simple for him. Also, it turned out, a fraudster. Where does the money come from? That was something that I always was questioning myself. But what if I told you that was the least interesting thing about him? His secret office was less than 500 meters down the road. I often ask myself now, did I know the true Jan at all? Certain things in my life since then have gone terribly wrong. I don't know if they followed me to my home. It looks like the ingredients of a really grand spy story. Because this ties together the Cold War with the new one.

13:56Listen to Hot Money, Agent of Chaos on the iHeartRadio app, Apple Podcasts or wherever you get your podcasts.

14:09It may look different, but Native culture is very alive. My name is Nicole Garcia, and on Burn Sage Burn Bridges, we aim to explore that culture. It was a huge honor to become a television writer because it does feel oddly like very traditional. It feels like Bob Dylan going electric, that this is something we've been doing for hundreds of years. You carry with you a sense of purpose and confidence. That's Sierra Teller Ornelas, who with Rutherford Falls became the first Native showrunner in television history. On the podcast Burn Sage, Burn Bridges, we explore her story along with other Native stories, such as the creation of the first Native Comic Con or the importance of reservation basketball.

14:50Every day, Native people are striving to keep traditions alive while navigating the modern world, influencing and bringing our culture into the mainstream. Listen to Burn Sage Burn Bridges on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

15:29and based on its current subscriber base, OpenAI would effectively have to double its current subscription revenue and massively increase its API revenue to hit these targets. These projections rely on one entity, SoftBank, spending$3 billion specifically on OpenAI's services. Really shouldn't have said specifically because they keep changing what it means. Meaning that they'd have to make enough on API calls, so people plug in the models into their products, to generate more revenue that OpenAI made in subscriptions in the entirety of 2024. and something else that I can only describe as an act of God.

16:01And that, I admit, assumes that SoftBank's spending commitment is based on usage and not like a flat fee, where SoftBank just hands them$3 billion and gets infinite levels of access. Assuming it's the former, I'd be stunned if SoftBank's consumption hits$3 billion in 2025, even with the massive cost of the reasoning models that Crystal Intelligence will maybe be based off of. Again, we don't know. And SoftBank announced this deal with OpenAI in February. Crystal Intelligence, if it works, and that is possibly the most load-bearing if of all time, will be a massive, complicated, and ambitious product.

16:34Details are vague, but from what I understand, SoftBank wants to create an AI that handles a bunch of varied tasks that knowledge workers do. I mean, it's just the same marketing bullshit. It's the same thing. It's the thing they've been lying about before. And to be clear, OpenAI's agents cannot consistently do, well, anything right now. What I believe is happening is that reporters are taking OpenAI's rapid growth in revenue from 2023 to 2024, when they went from like tens of millions of dollars a month in the beginning of the 2023 to 300 million in August 2024, genuinely a big leap. They're taking this to mean that the company will always effectively double or triple revenue every single year forever, with their evidence being OpenAI has said that this will happen in projections.

17:14It's bullshit. I'm sorry, it's bullshit. It's bullshit. as i wrote before in a newsletter it's called there's no ai revolution and the accompanying episodes at the time open ai effectively is the generative ai industry and nothing about the rest of the generative ai industry suggests that the revenue exists to sustain these ridiculous obscenes and frankly fucking stupid valuations and projections what do i mean by that by the way okay let me get into it chat gpt is the only real generative ai product with any significant usage or rather their nearest rivals are a fraction of said user base. Or maybe I need to be a little bit blunter.

17:50If anyone held a Google Gemini user conference, all the attendees could probably share a cab. Believing the OpenAI growth myth, and yes, reporting it objectively is both endorsing and believing these numbers, is engaging in childlike logic where you take one event, which is OpenAI's revenue grew 1 ,700 % from 2023 to 2024. Or, wow, to mean another will take place, which is that OpenAI will continue to double revenue literally every other year. Another insane thing to believe. And you're consciously ignoring difficult questions such as, how will they do this? And what's the total addressable market of large language models and their associated subscriptions exactly?

18:28And how does this company even survive when it expects the costs of inference to triple this year to$6 billion alone? Wait, wait, wait. Sorry. Sorry. I really need to be clear with that last one because it's a direct quote from the information the company also expects growth in inference costs the costs of running ai products such as chat gpt and their underlying models to moderate over the next half decade these costs will triple this year referring to 2025 to six billion dollars and rise to nearly 47 billion dollars in 2030 still the annual growth rate will fall to about 30 percent then okay thanks also are you fucking kidding me.

19:04Six billion fucking dollars for fucking inference. Hey, Casey Newton, I thought those costs were coming down. Casey? Casey? Casey? He's not here. He's not here. Anyway, that's not really great at all. That's actually really bad. The information reports that OpenAI will make about$8 billion in subscriptions to ChatGPT in 2025, meaning that 75 % of OpenAI's largest revenue sources eaten up by the price of providing it. This is meant to be the cheap part. This is the one fucking thing people say to me is meant to come down in price. I've had assholes saying to me for the last year, cost of inference is coming down.

19:41Is it? Are we living in different dimensions? Are there large parts of the tech media that have fucking gas leaks? What am I missing? Tell me what I am missing. Now, Ed, you haven't told the people that billions think you don't know what they're... Shut the fuck up. If you were one of these people who says, I need to... Casey, you're included, man. Fuck. I'm so sick of this. Oh, you don't talk to people running these things. I am sick of people like Casey Newton and others too saying, oh, you don't talk to enough AI people. You haven't listened to them. You mean I haven't listened to the pablum of the people that make money off of lying about this dog shit.

20:15You think that's what's missing from my analysis, interviewing people who work at these companies and understanding how the technologies work. I know how the technologies work. I don't need to talk to these fucking people. There are people out there like Simon Wilson and Max Wolfe who know how these things work that I talk to fairly regularly and both of them push back on me because they know how large language models work. Those people matter. What doesn't matter to me, what will never matter to me is what Dario Amadei, Jack Clark and all the other fucking people, Anthropic, think. And I think it's detestable and actively, honestly, malpractice in journalism to pretend that there's something ethical about speaking to these people and listening and taking in their marketing spiel.

20:58It's actually a little bit disgusting that this is even a critique leveled at anyone. But you're going to have to forgive me. I'm going to be a little rude. And I know that seemed like it, but I'm not even getting started. In fact, you know what? I think it's time, okay, everyone? I think it's time that I go through the most common critiques in AI. It's time for me to really sit down and I'm going to do my Kevin Roos voice. And I know a lot of you like my Kevin Roos voice and some of you, not a lot of you, I'm going to say I'm being rude to these people and it weakens my analysis to which I say, kiss my ass.

21:33I will turn you, I will cube you like a car in a garbage dump. But let's start, shall we? The cost of inference coming down. That's one argument. okay source source where is your source if you are someone saying to me that the costs of inference are coming down i want your source i want you to show me the costs i want you to show me the costs at scale because it sure seems like they're increasing for open ai and they're effectively the entire user base of the entire generative ai industry but ed what about deeps you sweet idiot child deep seek is not open ai and open ai's latest models only seem to be getting more expensive as time drags on.

22:11GPT 4.5 costs$75 per million input tokens and$150 per million output tokens. And at the risk of repeating myself, OpenAI is effectively the entire generative AI industry, at least for the world outside of China. On top of that, we actually don't know whether DeepSeek is even profitable to run at scale. It is definitely cheaper to run, but we don't know if it's actually profitable. Indeed, I don't even know how you'd calculate this because running a deep seek modeled as just one person is one thing. The question is whether you could scale it up like open AI. We don't know, but let's get back to the other critiques.

22:47This is a company, it's growth stage. They can just hit the button. It'll all be profitable. You have the mind of a child. If this was the case, why would both Anthropic and open AI be losing so much money? Why are none of the hyperscalers making profit on AI? Why does nobody want to talk about the underlying economics if they're at the growth stage? And also a little side point as well. Why have we been at the growth stage for years and why are hyperscalers at the growth stage? They're not startups. Anyway, on to another one though. These are the early days of AI. It's just like the early days.

23:17Wrong. Wrong. We have all the king's horses and all the king's men, the entire tech industry, and more money that has ever been invested into anything piled into generative AI and the result has been utterly mediocre. Nobody's making money on AI other than nvidia and maybe turing a consultancy but ed they're already showing signs that the ai is going to be powerful no it's not no it's not like i'm if anyone brings these critiques to you just say no no they're not show me show me show me why is it the only people i'm giving simon wilson credit here he's one of the only people who will show you anything cool and it's cloud compute stuff it's like relatively boring enterprise stuff it's exciting for the niche cases like software generally is, but it's really not showing any power.

24:03We talk about this powerful AI thing. Is it in the room? Like, where is this? Where is this powerful AI? But then I have actually had a few emails saying, Ed, Ed, look at OpenAI's O3 model. And I just want to be clear that this new and extremely expensive reasoning model also hallucinates more. Is that AGI, by the way? is this AGI is the AGI in the room with us did the AGI tell you it loved you did it tell you to leave your wife did it did it offer you sex I hope you're okay