Anthropic PBC plans to invest USD$50 billion to build large-scale data centers for artificial intelligence across the United States, including sites in Texas and New York.
The company said Wednesday that construction will begin soon, with facilities coming online in 2026. The project marks Anthropic’s first major effort to build and operate its own computing infrastructure, rather than renting capacity from partners like Amazon (NASDAQ: AMZN) and Alphabet Inc.’s Google (NASDAQ: GOOG).
The OpenAI rival said the new data centers will support the Trump administration’s goal of maintaining U.S. leadership in AI by strengthening domestic technology infrastructure. Anthropic expects the projects to create 800 permanent jobs and 2,400 construction jobs. Its development partner, UK-based Fluidstack Ltd., will provide “gigawatts” of energy capacity to power the sites.
Chief Executive Officer Dario Amodei said the new facilities will help the company build more powerful AI systems capable of driving major breakthroughs in science and technology. He added that the investment would create skilled jobs.
Founded in 2021 by former OpenAI employees, Anthropic has quickly become one of the strongest challengers in the AI industry. The company’s Claude chatbot and underlying AI models have attracted enterprise customers in sectors such as finance, health care, and technology. In September, Anthropic raised $13 billion at a $183 billion valuation and now serves about 300,000 business clients.
The company’s $50 billion build-out follows a broader wave of massive data center investments by major technology firms. OpenAI, Meta Platforms Inc. (NASDAQ: META), Microsoft Corp (NASDAQ: MSFT), NVIDIA Corp (NASDAQ: NVDA) (ETR: NVD), and Google all plan to spend hundreds of billions of dollars to expand their AI infrastructure.
Read more: CleanSpark outruns Microsoft to secure Wyoming AI and Bitcoin mining site
Read more: Can Quantinuum program Helios transform enterprise quantum computing?
Meta develops two gigawatt data facility in Louisiana
Earlier this year, OpenAI announced Project Stargate, a USD$500 billion plan to develop a network of advanced U.S. data centers that will later expand internationally. Meta is also developing a two-gigawatt data facility in rural Louisiana. Furthermore, CEO Mark Zuckerberg said he expects to invest USD$600 billion in U.S. infrastructure over the next several years.
However, not all investors are convinced that this massive spending is justified. Some have raised concerns about whether the enormous costs will lead to sustainable profits. OpenAI Chief Financial Officer Sarah Friar responded to those doubts last week, suggesting that enthusiasm for AI’s potential remains too limited given its transformative impact on work and society.
Anthropic’s partnership with Fluidstack shows how smaller, specialized firms are helping to build the infrastructure behind the AI boom. Fluidstack belongs to a group of “neocloud” providers that lease high-powered computing systems to AI developers. The company recently made headlines for two unconventional deals involving Google and cryptocurrency miners TeraWulf Inc. (NASDAQ: WULF) and Cipher Mining Inc. (NASDAQ: CIFR). In those arrangements, Google agreed to reimburse creditors if Fluidstack defaulted or filed for bankruptcy.
In addition to its work in the U.S., Fluidstack has also emerged as a key player in Europe’s AI ambitions. Earlier this year, it announced plans to build a €10 billion (USD$11.5 billion) “supercomputer” in France. This is part of French President Emmanuel Macron’s pledge to expand Europe’s AI capacity. That system, which will deliver one gigawatt of power, is also scheduled to open in 2026.
