CoreWeave (Nasdaq: CRWV) has signed a multiyear cloud agreement with Anthropic to provide GPU-accelerated infrastructure for the development and deployment of Anthropic's Claude family of artificial intelligence models. The compute capacity will come online later in 2025 in a phased rollout, with potential for expansion over time. Financial terms of the deal were not disclosed.
Expanding the AI Client Roster
With this agreement, Anthropic joins a client base that now includes nine of the top 10 AI model providers globally on CoreWeave's platform. Founded in 2021 and headquartered in San Francisco, Anthropic is among the most prominent AI research companies in the United States. Its Claude models directly compete with offerings from OpenAI and Google, and the company has been aggressively expanding its infrastructure partnerships as demand for its models grows among developers, startups, and enterprise customers.
CoreWeave's existing clients include Microsoft, OpenAI, and Meta. OpenAI holds a contract valued at over $12 billion, while Meta signed an expanded infrastructure agreement worth $21 billion. Historically, more than 60% of CoreWeave's revenue came from Microsoft, but the company has been working to diversify that concentration.
Financial Strength and Growth Trajectory
CoreWeave went public on the Nasdaq on March 28, 2025, raising approximately $1.5 billion in one of the largest AI-related listings to date. The company reported $5.13 billion in revenue for the full year 2025 and holds a backlog of $66.8 billion. For 2026, CoreWeave expects revenue between $12 billion and $13 billion.
To fuel its expansion, CoreWeave plans capital expenditures of $30 billion to $35 billion in 2026, targeting an annualized revenue run rate above $30 billion by the end of 2027. The company currently operates 32 to 43 data centers across the United States and Europe, deploying hundreds of thousands of Nvidia GPUs, including the H100, H200, and the Blackwell GB200 NVL72 series. Nvidia is a strategic investor in CoreWeave and committed an additional $2 billion in 2026.
From Ethereum Mining to AI Cloud Leader
Founded in 2017 as Atlantic Crypto, an Ethereum mining operation, CoreWeave pivoted its GPU inventory toward rendering and machine learning workloads after the crypto market downturn from 2018 to 2019. The company rebranded to CoreWeave and fully transitioned into AI cloud infrastructure as generative AI demand accelerated.
CoreWeave has earned the highest Platinum rating in both Semianalysis ClusterMAX 1.0 and 2.0 benchmarks, which evaluate AI cloud performance, efficiency, and reliability. It also holds an industry-leading MLPerf benchmark score for AI workloads. The company targets 1.7 gigawatts of power capacity by the end of 2026, with a longer-term goal of multi-gigawatt expansion in partnership with Nvidia by 2030.
The Anthropic deal further cements CoreWeave's position as a premier provider of AI cloud infrastructure and underscores the intensifying demand for compute capacity to power cutting-edge large language models.