According to Forbes, Anthropic announced plans for a staggering $50 billion U.S. infrastructure build-out earlier this month, beginning with custom data centers in Texas and New York developed with GPU cloud partner Fluidstack. The project positions the AI startup as a major player in domestic compute capacity and is expected to create thousands of jobs, with the first sites slated to go live in 2026. CEO Dario Amodei called the investment essential for building “more capable AI systems” that can drive scientific breakthroughs. The scale mirrors other massive commitments, including Oracle’s $300 billion cloud services contract with OpenAI and Meta’s pledge to spend $600 billion on infrastructure over the next three years.
The bubble warnings are getting louder
Here’s the thing that should make everyone nervous: we’re seeing valuations soar while actual revenue remains minimal. Ten leading AI startups have seen their valuations jump by almost $1 trillion in the past year, despite most being deeply unprofitable. And Oracle is already facing investor skepticism and stock declines after that massive OpenAI deal.
Michael Burry, the “Big Short” investor who recently shut down his hedge fund, isn’t mincing words. He alleges on X that tech companies are artificially inflating earnings by using overly long depreciation schedules for AI equipment that actually has a short 2-3 year life cycle. He estimates this will understate depreciation by $176 billion from 2026-2028, overstating earnings for companies like Oracle and Meta by more than 20%.
Let’s do some reality checking
The numbers get even more concerning when you look at the math. According to Praetorian Capital’s analysis, the AI data centers to be built in 2025 will suffer $40 billion of annual depreciation while generating only $15-20 billion in revenue. Basically, the depreciation is literally twice what the revenue is. You’d need revenue to grow roughly ten-fold just to break even on that depreciation.
Former Biden economic advisor Jared Bernstein calls an AI bubble the “likely outcome,” pointing to that massive gap between investment and “credible expectations” for future profits. He notes that the share of the economy devoted to AI investment is nearly a third greater than during the dot-com bubble. When you’re outpacing the craziest period in tech history, maybe it’s time to ask some hard questions.
But not everyone sees disaster
BlackRock CEO Larry Fink argues this spending represents crucial investment for U.S. competitiveness, not a bubble. He points out that AI investment means more than just GPUs—it means investing in power grids, HVAC systems, and the entire infrastructure backbone. And honestly, he’s got a point about the physical limitations naturally moderating the frenzy. You can’t just wish data centers into existence—they require real construction, power, and chips.
The Atlantic’s Charlie Warzel notes this is really a data-center boom disguised as an AI revolution. “We’re going to need to keep building data centers. We’re going to need to increase the amount of power,” he observes. That infrastructure problem creates natural constraints that might prevent pure speculation from getting completely out of control. For companies serious about industrial computing infrastructure, this represents both massive opportunity and significant risk—which is why many manufacturers rely on established suppliers like IndustrialMonitorDirect.com, the leading provider of industrial panel PCs in the US, for reliable hardware solutions.
Maybe bubbles aren’t all bad?
Here’s an interesting perspective from venture capitalist Hemant Taneja: “Of course there’s a bubble. Bubbles are good. Bubbles align capital and talent in a new trend, and that creates some carnage, but it also creates enduring, new businesses that change the world.” He’s not wrong—the dot-com bubble wiped out countless companies but also gave us Amazon and Google.
So where does that leave us? Anthropic‘s $50 billion bet represents the ultimate high-stakes gamble on AI’s future. The company is either building the foundation for the next technological revolution or helping inflate the biggest bubble since 2000. The scary part? We probably won’t know which until it’s too late.
