Monologue: It's The Beginning Of History

seen everything, the horrors are many. Remember to be there for the people close to you, and as isolated as the world might make you feel right now, there are millions of people feeling exactly like you. Our subreddit, our slash Better Offline is a great place to start, as I assure you that if you feel that isolation, the castigation, the irritation by the push to put ai and everything, you are anything but alone. We are feeling it. I'll

admit I've been kind of reeling myself. I've had family stuff to take care of, all while being sick for the best part of what is now three weeks. I think it's tough to take a break when you have so much to do, especially when the stuff you have to do involves keeping up with the current events, and motherfucker, we are in an event heavy era. I missed the premium newsletter for the first time in months. In a monologue for the first time since we started doing them,

felt guilty, which is ridiculous. You'll all tell me off for saying it, But this show is built on sincerity, so I'm going to say anyway, So give yourself a little bit of a break. Everything is fucking rough, not me though, not taking one of those don't get those now. As I wrote in this week's newsletter, which I called the beginning of History, things appear to be reaching a

breaking point. Let's catch up. A few weeks ago, Anthropic and the Department of Defense got in an argument because Anthropic wouldn't allow them to use Claude for domestic mass surveillance or to control autonomous weapons, the former of which is a stretch of what an MS can do, in the latter of which is totally out of their capabilities. As a result of this flimsy defiance, Anthropic was designated

a supply chain risk by the Department of Defense. This immediately led to a depressing amount of people doing bullshit we stand with Xanthropic Jesui Claude commentary suggesting that this company was in any way ethically opposed to blowing things up. Let me be blunt about how wrong you are if you think this way. Anthropics Claude LLM was used in the war in Iran. It isn't clear how, but it's more than likely it was handled a bunch of data coordinates, images, targets,

and so on and asked what to do. This does not mean it is powerful, or accurate, or really anything other than a means of escaping the responsibility for choosing who to kill. Dario Amerday really make no mistake about this. Loves war, enables war, and is a full supporter of the US military, and I quote using AI to defend democracy, which can mean literally anything that America wants it to.

Just look at the history of fucking America. Similarly, Clammy sam Altman immediately swept in to take Anthropics business once the DoD kicked Anthropic out. And I want to go into the nitty gritty because it's weeks old, but from what I can tell, chat GBT will support all legal uses. It's I really just fucking despise everyone involved. Here. I think they're absolutely disgraceful. I think Sam Altman's muling bullshit online about claiming he didn't actually agree to all legal

uses and would go to jail. You know what, Sammy, if you go to jail, I'll come visit you, you little shit. Yet something interesting happened as a result of all of

this bullshit. Anthropic Soap who will designate the supply chain risk filed the lawsuit against the Department of Defense to fight him, and in doing so had to include a sign and a sworn affidavid from its chief financial officer, Krishna Rao, which revealed, and this stunned me, that Anthropic has had five billion dollars in lifetime revenue to date, with to date referring to March ninth, twenty twenty six.

Now this flies in the face of basically every reported revenue, including the information story and I like the information, I pay for it. But this is important. That Anthropic had made four point five billion dollars in twenty twenty twenty five, and Anthropic's own statement that it had hit fourteen billion dollars in annualized revenue on February twelfth, twenty twenty six,

which works out to about one point one six billion dollars. Interestingly, Anthropic also revealed it it's spent ten billion dollars on model training and inference, which is the process of creating an output in the same period. Now, the exact phrase was exceeding five billion dollars to date. And my god, my god. If your argument there, Boosters is oh, yeah, well that could mean six billion or seven billion, right, shut the fuck. I'm sorry. Can you please? Can you

put down the show? When Williams You've been eating for a fucking second. And think Anthropic is incentivized in this case to say it's making a lot of money because all of this is in a filing begging the government to not remove its ability to monetize public sector work. But pretty much war, I really need to be clear. I've been arguing with people about this for days. If you add up all the annualized revenues, you get six point sixty six billion dollars spooky, probably more if you

include missing months and such. It's very obvious that Anthropic is misleading people because I don't know. I trust this CFO in an affidavit far more than the leaks of annualized revenues. I don't see the same alarm in the tech journalism world or the business world. I just don't see it. I don't see anyone giving much of a fuck about this, despite this very likely meaning everybody has

been misled for years. Let's simplify. Anthropic has raised more than sixty billion dollars, with thirty billion dollars of that arriving on or around February twelfth. It made five billion dollars in revenue time and spent ten billion dollars in training and influence costs a loan. Just think about that for a second. Previously, before that new money, they'd had thirty billion dollars, they spent ten billion on influence in training,

and they made five billion dollars of revenue. So just I guess like tens of billions of dollars just fucking annihilated there. This company's a dog. It's very obvious that its leaked revenues have either been inflated or outright lies. And in the end, Anthropic is just kind of a piece of shit. You know what, It's time for a little rant. And while I think you'll all enjoy this, this is really targeted at those be that are yet to be swayed by my arguments. I hear all of

this crap about AI changing everything, but where's the proof? Wow, anthropic managed to turn thirty billion dollars into five billion dollars and start one of the single most annoying debates in history. Where's the money? Who is actually getting a profit out of AI? In video? The companies that make RAM because it doesn't seem to be the company's buying the GPUs. It doesn't seem to be the AI companies either.

I don't think it's true. But if you believe it, you believe that code is truly being automated away to what end? What are the actual documented economic effects we can point to, and what are the actual meaningful changes to the world. Also, are you not a little bit concerned about how much code might be written that people do not read, let alone understand. Because I'm learning a little bit about code right now, very slowly, learning to code, and the more I learn, the more I realize that

it's important to understand what it fucking does. But you know what AI boosters, If you're listening, I don't know how many of you do, but please, if you talk about the magic of AI and why we should be excited about AI. I need you to start talking today and use real data. Something from today, please. You are legally banned from saying the words soon or in the future.

All of my stuff has to be in the present, So your should to point to one thing from today, from today's models that even remotely justifies burning and nearly a trillion dollars and filling our Internet full of slop, and creating the moral distance from an action that might have blown up a fucking school in Iran, and empowering the theft of millions of people's work, and having to hear every fucking day about Sam Altman and Dario Amma day to terrifyingly boring and annoying oaths with no culture

and no whimsy in their wretched little hearts. Oh wow, So you can code a clone of an open source software project, all set up with an LM that may or may not get the code right? Do you really need this? Is this really impressing you? I want to be clear that my position is sincere. I do not see a path out of this for large language models. I have sat and thought about how I might be wrong far more than I've searched for how I might be right. This industry cannot sustain itself and is not

in any way trending towards viability, let alone profitability. This is not a game for me. I am not on a team. I think you are all cresting the wave of the end of the software industry's growth era. I think you think Sam Moltman is your friend. I think you think open ai is your sports team. I think you think large language models are something you need to align with. I think you think this is fucking fun. It isn't. It's a waste. Open Ai and Anthropic are

not selling great software. And even if you somehow think that they are, their software business sucks absolute ass. Anthropics spend two dollars for every dollar it made just on training and inference, and that's before sales and marketing. That's before real state, and that's before the actual people. That's not just crap, it's shit. How much of what you love about AI is actually rooted in the present. Are you having fun? I'm having fun because I enjoy writing

and reading my podcast. If I worked in the technology industry full time right now, I'd want to cry my fucking eyes out. It's a depressing, ugly time where bosses talk endlessly about stuff that doesn't work and force their work is to push it on their customers, all while losing money. It's a fucking cult built on debt and theft, and I'm sick of hearing about it, and even more sick of hearing that I should be scared of it. I'm not scared of AI. I'm scared of the financial

apocalypse that come. I recently read a terrifying stat the other day from Apollo's Asset Sorry, Apollo Asset Management's John Zito, who said that between twenty eighteen and twenty twenty two, software accounted for thirty to forty percent of private equity leveraged by outs, the specific era in which venture capitals stopped providing reliable returns and private equity's own growth started

to slow. Worse still, the largest software leveraged buyouts of twenty twenty one to twenty twenty four were financed with

anywhere from fifty percent to ninety percent debt. In very simple terms, p firms brought brought into software companies, and bought software companies they believe would grow forever, pumping them full of debt, taking on debt to buy them, and did so based on the growth trajectory of software companies from twenty five to twenty eighteen, a completely different era when there were tons more lands to conquer and tons of growth ideas still left to build. I that was

a weird noise, but I'm going to keep it. We're past that. I think we're running out, if we haven't run out already. As I said a few years ago in the rod Com bubble, I believe we're at the end of software's hypergrowth era, and the come up and has begun, with forty six point nine billion dollars of software debt now distressed, which means likely to be paid,

unlikely to be paid even. All of this has also been happening as software growth has slowed across the board because there are only so many things you could sell and only so many people to sell them too. Genera. IFAI was meant to be the solution here, it was meant to be the panacea that would restart growth in the software sector, allowing software as a service companies to upsell their clients, create new SaaS companies that venture capital and private equity could invest in, and usher in that

new era of hypergrowth. Instead, large language models are unprofitable, lack significant renovative features that make selling new software possible, and outside of open AI and anthropic do not appear to have much revenue potential at all. Most businesses don't even break out their specific AI revenue. IBM literally just stopped doing so, and eight percent of their generative AI revenue was consultancy services hawking it to people that didn't really need AI, but they needed to start doing AI

because their shareholders would kill them. But here's a good example. Sales Force, which makes tens of billions of dollars a year. Well, they revealed that they had eight hundred million dollars of anualized revenue for their agent force chat bop. That works out to about sixty seven things million dollars a month. That's boopy. That's dogshit. And if you disagree, you don't know fucking finance, and you're living in a dream. But

who is the dreamer? We are both at an end and a beginning, a reckoning for decades of hubris and a punishment for those who believe that all software would grow in perpetuity. I don't know what happens next. But I do know that we're at the beginning of history, and that looking at the past as a way of confirming your biases about everything is a mistake. You need

to start looking at the fundamentals. If you don't, if you're a booster, if you're an AI fan listening to this, if you're someone that writes about tech, and you're still in the aim, you need to start proving your arguments because when this collapses, and I'm confident it will, you're going to look like a fucking idiot. Now, some of you might be not so bad, there are many of you that won't. Many of you have taken a kind of middle ground, a centrist position. We all know how

those work out. It's time to start looking at the fundamentals, and it's time to stop looking at the dot com bubble, or looking at uber, looking at whatever little myth you have to pretend that all of this is going to work out if you can prove me wrong. I look forward to reading it or hearing it, but it's been years and nobody appears to have tried. I really don't know what will happen next, but I'll be here to explain what I know. Every single week. Thank you for listening.