Artificial intelligence startup Anthropic has announced plans to invest $50 billion in building AI infrastructure across the United States. The company will begin by developing its own data centers in Texas and New York City in partnership with Fluidstack.
The first facilities are scheduled to open in 2026. The project is expected to create thousands of jobs and strengthen the U.S. position in the global technology race. Anthropic CEO Dario Amodei told CNBC that the investment will help build “more powerful AI systems” capable of driving major scientific breakthroughs.
Rising AI Spending Sparks Bubble Concerns
The scale of financial commitments in the industry has reached unprecedented levels. Oracle has reportedly concluded a $300 billion cloud-services contract with OpenAI, while Meta, recognized as an extremist organization in Russia plans to spend $600 billion on infrastructure over the next three years.
Yet these investments remain far ahead of actual earnings. According to the Financial Times, the combined valuations of ten leading AI startups, including Anthropic, OpenAI, and xAI — have surged by nearly $1 trillion in just a year, despite most of them still operating at a loss.
Investor Michael Burry, known for predicting the 2008 mortgage crisis, recently closed his hedge fund Scion Capital, saying its valuation models were “no longer in sync with the markets.”
On X, Burry accused major tech firms of artificially boosting profits by using overly aggressive depreciation schedules on AI hardware, which typically lasts only two to three years. He estimates this could understate depreciation by $176 billion between 2026 and 2028, inflating Oracle’s and Meta’s profits by over 20%.
Jared Bernstein, former chairman of the U.S. Council of Economic Advisers, also warned that an AI bubble is a “likely outcome,” citing extreme valuations and a widening gap between investment and credible revenue expectations. He noted that the share of economic activity flowing into AI investment is now nearly a third higher than what flowed into the internet sector during the dot-com bubble.
Harris Kupperman, founder of Praetorian Capital, estimates that AI data centers built in 2025 will face $40 billion in annual depreciation while generating only $15–20 billion in revenue.
Optimists See a Foundation, Not a Bubble
Not everyone sees a looming collapse. BlackRock CEO Larry Fink told CNBC that massive spending on AI infrastructure is essential for maintaining U.S. competitiveness.
“Investing in AI means not only buying GPUs and chips, but also investing in cooling systems, IT infrastructure, electrical networks, and power supplies,” Fink said.
Physical limits, such as power grid capacity, construction bottlenecks, and chip availability could naturally slow down industry expansion, preventing purely speculative overreach.
Hemant Taneja, CEO of General Catalyst, added a broader perspective: “Of course there’s a bubble. Bubbles are healthy. They channel capital and talent into a new trend, create losses, but also give birth to long-lasting companies that change the world.”
Anthropic’s $50 billion commitment reflects the company’s belief in AI as a transformative force. Whether this marks the beginning of a technological renaissance or the inflation of a new financial bubble remains to be seen.