Anthropic has announced a $50 billion data center buildout across the United States through a new partnership with U.K.-based neocloud provider Fluidstack, marking its largest infrastructure investment to date. The company said the facilities, located in Texas and New York, will come online throughout 2026 and will be custom engineered for Anthropic’s compute-heavy AI workloads.
CEO and co-founder Dario Amodei said the expansion reflects the requirements of developing frontier models like the Claude family, noting that more advanced systems capable of accelerating scientific discovery will depend on infrastructure built specifically for high-density training and inference.
Anthropic already relies heavily on cloud partnerships with Google and Amazon, both of which are investors. However, this marks its first large-scale move into building dedicated infrastructure, sized to match internal forecasts that reportedly project $70 billion in revenue and $17 billion in positive cash flow by 2028.
Even at $50 billion, the plan is smaller than competing projects underway across the AI sector. Meta has committed roughly $600 billion to data centers over the next three years, and the Stargate initiative, involving SoftBank, OpenAI, and Oracle, has outlined $500 billion in planned infrastructure spending. The scale of these commitments has prompted analysts to raise questions about an emerging AI investment bubble, citing the possibility of misallocated capital or demand that may not keep pace with projections.
The agreement is a major win for Fluidstack, which has rapidly become a notable supplier in the AI infrastructure boom. Founded in 2017, the firm was selected earlier this year as the primary partner for a 1-gigawatt AI data center project backed by the French government, representing more than $11 billion in funding. Fluidstack also maintains relationships with Meta, Black Forest Labs, and Mistral, and was among the first external vendors to receive access to Google’s custom-built TPUs.
Anthropic said the upcoming U.S. facilities will be optimized for efficiency, built with the company’s long-term scaling needs in mind as model development demands increasingly dense and specialized computing environments.
Featured image credits: Wikimedia Commons (Kimberly White via Getty Images)
For more stories like it, click the +Follow button at the top of this page to follow us.