80% of AI Neoclouds Will Fail by 2030 - AI Isn't Profitable Enough for $3.5 Trillion Investment

Eli the Computer Guy (0:00)

Coming from Gartner, over 80% of Neo Clouds may cease independent operations by 2030 as 3.5 trillion of AI infrastructure faces impairment risk. So that's right, when we talk about Neo Clouds, we're talking about companies like CoreWeave, basically these are cloud infrastructure providers that just provide AI services. So we're not talking about AWS here, we're not talking about Azure here, we're not talking about DigitalOcean or Linode. Does Linode still exist independently? I don't know, whatever, right? We're not talking about those particular companies, we're talking about the companies like CoreWeave that have sprung up, basically to provide rentable AI hardware for the big AI boom. Basically, what Gartner is saying, they are looking at the buildouts, they are looking at the lack of demand for AI services and solutions, and they're saying 80% of these neoclouds may not be able to survive by 2030 And I will remind the GenXers out there, 2030 is not 50 years away. Again, every time I hear a date like 2030, I'm like, well, I mean, come on now, come on, in 50 years, who knows what companies will still be alive or will be dead? And you realize, oh, it's not the 1980s, it's 2026, 2026, 2030 is three years away. So literally 80% of these Neo clouds may go either belly up or be gobbled up by somebody else because they will become economically non-viable. And this is the big issue that I have with this whole AI boom, is we haven't proved economic viability. We just haven't. And again, and it's not simply that we have, we have proved that there is some value to AI. So Anthropic, I guess, came out and they just did like an investor briefing, and they said they earned like $16 billion or something for the quarter. And so don't get me wrong, and there is demand to a degree for these AI products or services. But also look at $16 billion or whatever it is for the quarter and realize that Anthropic wants to be valued at $2 trillion.

You know what I'm saying? Like that's the problem that we're running into. Anthropic wants to be valued at $2 trillion.

OpenAI wants $1.5 trillion on the private market, which means when they IPO, they're going to need $2 or $2.5 trillion. Again, we've got SpaceX, they're like $2.3 trillion right now. And that's simply the problem that we get to with so much of the stuff, is it's not that there is no value here. It's not simply that there's not a lot of value. It's just not that much. It's just not that much value. It's a problem now with CoreWeave, CoreWeave and any of these Neo clouds. The idea is they build out these AI data centers, so that they can then rent the services out to everybody who needs it. But then the question that becomes is that if the demand is not there, then what happens to these companies? As I've been talking about many times at this point in time, there's this idea of resource constraint. The only thing that's keeping AI from growing faster, than it currently is, is because there's simply not enough resources, right? Resource is constrained. But my big argument, my big argument with all of this, is if somebody wants to use an AI service, simply have them swipe a credit card and pay $1 per month. And $1 per month will not pay for inference costs for most people. But just put that wall up, just put that one barrier, just add that teeny tiny ounce of friction to the transaction process.

And I bet you, the whole concept of resource constraint dis-fucking-peers overnight. Because I can tell you, when I use Claude, or I use Anthropic, or I use Gemini, I don't pay for it. I'm actually using them every day right now, because I'm doing the education classes again. I do use them. I don't pay for them. When I go to Google, they shove AI down my throat, force people to pay a buck, and I do wonder what the hell happens. And so again, think about how tenuous this market is.

A dollar per month of friction, and I put $1,000 on the table, this entire fucking thing collapses overnight. And so think about it, all these Neo clouds. Basically, the idea, and that's one of the things to think about, like these kind of companies, is I had a buddy of mine who sold a startup company to Yahoo. So, he had a failed startup company. Really good idea. BizDev, the technology was amazing. The BizDev was non-fucking existent. Anyways, that's his own story. Anyways, really kind of cool company.