Artificial intelligence firm Anthropic has announced a sweeping $50 billion investment to build a network of data centers across the United States, signaling a shift toward owning more of the infrastructure that powers its growing family of AI models.
The initiative, developed in partnership with U.K.-based cloud provider Fluidstack, will see Anthropic construct new data centers in Texas and New York, with the first facilities expected to come online in 2026. The centers are being purpose-built to optimize efficiency for Anthropic’s large-scale AI training and deployment workloads.
Co-founder and CEO Dario Amodei said the move is part of the company’s long-term strategy to reduce reliance on major cloud partners such as Google Cloud and Amazon Web Services, which have so far handled much of Anthropic’s compute demand. Building proprietary infrastructure, he noted, is essential for supporting the next generation of Anthropic’s “Claude” AI systems and for expanding access to powerful, safer AI tools. 
Anthropic projects the investment will align with its growth forecasts, including an expected $70 billion in annual revenue and $17 billion in positive cash flow by 2028.
The $50 billion outlay places Anthropic among the top AI infrastructure spenders but still behind some of its rivals. Meta plans to invest roughly $600 billion in data centers over the next three years, while the “Stargate” consortium—a collaboration between SoftBank, OpenAI, and Oracle—is preparing an estimated $500 billion buildout. Industry analysts say the escalating costs highlight both the capital intensity of frontier AI research and growing concern that the sector’s infrastructure race could be overheating.
For Fluidstack, the deal represents a major milestone. The company, founded in 2017, has become a key player in AI infrastructure projects across Europe and North America. It previously partnered on a French government-backed 1-gigawatt AI facility valued at over $11 billion and has worked with leading AI developers such as Meta, Mistral, and Black Forest Labs. Fluidstack was also among the first external firms to deploy Google’s custom Tensor Processing Units (TPUs), underscoring its position as a trusted partner in high-performance AI computing.
Anthropic’s investment marks one of the clearest signs yet that leading AI developers are moving beyond rented cloud capacity to secure long-term control of their computing foundations — a trend likely to reshape the economics of the AI industry over the next decade.
