Initial facilities are slated to go online in 2026, with more sites to follow, supporting enterprise growth and long-term research.

Fluidstack will build and manage large-scale GPU clusters to support Anthropic’s workloads, enabling scalable model training and deployment.

Fluidstack, a London-based neocloud provider, partners with Anthropic to supply the necessary infrastructure for Claude.

Internal projections, cited by The Wall Street Journal, show Anthropic aiming to break even by 2028 as it scales.

WSJ-reported projections also position Anthropic’s path to profitability ahead of OpenAI’s expected losses in the same year.

Anthropic has released Claude Sonnet 4.5, claiming top coding-test results and the ability to produce fully ready applications, not just prototypes.

Building proprietary data centers will give Anthropic greater control over performance, costs, and hardware, with plans for additional sites beyond Texas and New York.

The project is framed as meeting rising Claude platform demand while emphasizing cost-effectiveness and sustainability.

The data centers are designed to run at scale for Claude workloads, aligning with Anthropic’s research agenda and enterprise ambitions.

Industry observers note rivals are pursuing massive AI infrastructure bets, with Meta's rumored $600 billion data-center push and a SoftBank–OpenAI–Oracle Stargate project valued at about $500 billion, signaling a potential AI infrastructure bubble.

Anthropic is moving from cloud reliance to owning core infrastructure, signaling a strategic push for vertical integration of compute for training and deploying Claude AI models.

Industry data shows cloud providers leasing unprecedented data-center capacity in the third quarter, underscoring strong demand for AI-ready infrastructure.