Microsoft Slashes Spending on Anthropic AI Claude - AI is Too Expensive

SPEAKER_1 (0:00)

Coming from the information, Microsoft slashes internal Claude spending by a third. Meta has halved its internal users of Anthropics AI as both companies work to shift internal AI use to its own models and tools. And so this is gonna be a big problem coming up for Anthropic and also for OpenAI, basically where companies and people realize that AI or these models are really cool, but they don't necessarily want to use Anthropic or OpenAI. With these companies, with Dario and with Sam Altman, they essentially had the idea that they could monopolize these models for anybody who wanted to have an intelligence layer, and then people essentially will be forced to come to these companies. The thing is, as I've talked about for a long time now, that's not how systems architecture works. It's not how systems architecture has ever fucking worked. And so that's one of the reasons I argue against the whole business case for these companies being worth trillions and trillions and trillions of dollars, because it just simply doesn't make sense. I think this is also interesting too with Anthropic trying to IPO shortly. So Anthropic is trying to IPO in November now. And I was listening to the big short guy, right? Burry, Michael Burry, I think his name is. And anyways, one of his arguments is that they have to basically IPO in November before the third quarter results fully have to be documented or released. Because basically, what he's been arguing is that a lot of Anthropic's revenue has been boosted over the TokenMaxing idiocy, right? So we had this whole thing. I thought, I literally back in January, February, I thought TokenMaxing was a parody. I heard it. I thought it was like, this is so stupid. It's got to be a parody. It wasn't, right? A lot of companies out there, they just wanted their employees to burn cash for some reason, which is so weird for me being a technology professional. In what other era of technology have CEOs literally come to the IT department, the tech department, with fucking wheelbarrows full of cash saying, you have to spend it?

You're not going to the CEO begging for a rack of servers. You're not going to the CEO trying to explain why you need another engineer. Like the CEO is coming to you with just all the cash that they can put in a wheelbarrow, saying, you have to spend this by the end of the day. If you don't spend this by the end of the day, you're going to get punished. Like what? That's not how any of this normally works, right? But we had the whole token maxing thing, and even token maxing right now fully hasn't washed out of the system. And basically, there's the argument by Michael Burry is that that token maxing and that kind of wave of spending, that is what Anthropic is basically using to prove how successful they are, because they're using ARR. So either annualized revenue or annualized recurring revenue, right? The idea is to take the amount of money you earn in one period of time, multiply that by X to get to 365 days, and that gets you your annualized revenue. And one of the things that Michael Burry is saying is that he thinks coming into October, November, even December, we're going to see more and more cuts to AI spend. So the problem is that annualized revenue is going to start to go down. And so if they want a $2 trillion valuation, they need to get the money in. The argument is the IPO now pull in as much money as possible. And then I'm not sure what happens after that, because it seems like that would get up. So think about this. If you IPO based off of bloated annualized revenue now, so you IPO, you go on the market, you get your $2 trillion valuation, you sell a shit ton of equity, but then after that, you actually have to file everything quarterly with the SEC, and your revenue goes off a fucking cliff afterwards. What the hell is going to happen to the value of your stock?

I don't know. Like I said, all of this is so, all of this is just, this is ugly. This is just all ugly, ugly stuff. But the curious thing, again, as we see, as people decide to use more and more artificial intelligence, the reality is that they're going to use options other than open AI or anthropic, whether it's using things like Olamma, whether it's using Chinese models, whether it's using American open weight model. So NVIDIA has put money. So NVIDIA Neutron? So many names. Anyways, I think they have Neutron. So I think NVIDIA has the Neutron models. I think the Neutron. Anyways, they have open weights models that you can use. Not only that, but they've invested, I think it's into a reflective AI.