CoreWeave establishes a multi-year agreement with Anthropic, as nine out of ten leading AI model providers become part of its platform.

      In summary: On April 10, 2026, CoreWeave revealed a multi-year agreement with Anthropic, allowing the creator of Claude to utilize Nvidia GPU resources in U.S. data centers for large-scale AI tasks. The financial details of the agreement were not shared. This announcement followed CoreWeave's declaration of a $21 billion expansion of its partnership with Meta and adds Anthropic to a client list that now includes nine of the ten major AI model developers. CoreWeave reported revenues of $5.13 billion in 2025 and is projecting over $12 billion for 2026, supported by a contracted backlog exceeding $66 billion.

      Former crypto miner transitions to AI services

      Established in 2017 as Atlantic Crypto, CoreWeave initially operated as an Ethereum mining business that purchased Nvidia graphics cards in bulk for cryptocurrency mining and rented extra GPU resources to other miners. However, as cryptocurrency profits dwindled in 2019, the company rebranded as CoreWeave and shifted its focus to on-demand GPU cloud services for general computing. This pivot coincided with the AI model training surge that started in earnest in 2023, transforming CoreWeave's cache of Nvidia hardware into a strategically significant asset within the tech sector. The company went public on Nasdaq under the ticker CRWV on March 28, 2025, with an initial share price of $40, raising $1.5 billion and achieving an approximate valuation of $23 billion. CoreWeave currently manages 32 data centers housing over 250,000 GPUs and 1.3 gigawatts of contracted power capacity. Its 2025 revenue of $5.13 billion marked a 168 percent year-on-year increase, and management anticipates revenues of over $12 billion for 2026, backed by a contracted backlog surpassing $66 billion.

      However, this rapid growth brings a substantial concentration risk, with Microsoft contributing around 67 percent of CoreWeave's 2025 revenues—a reliance that analysts and investors highlighted prior to the IPO. Microsoft's initiative to create its own AI models adds another variable, raising concerns regarding how much of Microsoft's computing needs may shift toward in-house solutions instead of third-party GPU cloud services. The deal with Anthropic, coming just after the $21 billion Meta expansion, signifies CoreWeave's concerted effort to develop a diverse customer base that diminishes reliance on any single hyperscaler.

      What Anthropic expects from the partnership

      Anthropic’s computing strategy has evolved alongside its revenue, with its annualized revenue run rate exceeding $30 billion in early April 2026—more than triple the $9 billion reported at the close of 2025. This rapid growth, propelled by enterprise adoption of Claude and the significant progress of Claude Code, necessitates Anthropic to enhance its infrastructure commitments across various chip architectures. Its main training operations utilize Amazon Web Services' Trainium hardware via Project Rainier, a vast network composed of hundreds of thousands of AI chips distributed across multiple U.S. data centers. Just days prior to the CoreWeave agreement, Anthropic secured a deal with Google and Broadcom for multi-gigawatt TPU capacity, enabling access to about 3.5 gigawatts of next-generation tensor processing unit computing anticipated to go live in 2027. The CoreWeave agreement adds a third dimension: Nvidia GPU capacity for production inference workloads, delivering the scale and latency performance necessary for enterprise Claude deployments. Earlier this year, Anthropic also committed $100 million to its Claude partner network, indicating its intent to expand the ecosystem of developers and organizations using Claude—this ecosystem growth is now directly influencing the computing procurement choices behind such agreements.

      Michael Intrator, co-founder and CEO of CoreWeave, emphasized that the deal encompasses more than just infrastructure capacity. “AI is not merely about infrastructure; it involves platforms that translate models into real-world outcomes,” he commented. “We’re eager to collaborate with Anthropic at the forefront of deploying models effectively in production.” Anthropic plans to initiate computing deployment through a phased infrastructure rollout, with potential for future expansion. The specific Nvidia chip architectures involved remain undisclosed, although CoreWeave’s resources include both current and upcoming Nvidia GPU generations. The Vera Rubin GPUs from Nvidia, announced at GTC 2026, represent the next significant architecture on CoreWeave's deployment timeline, with large shipments anticipated for the latter half of 2026.

      Nine out of ten, two deals in 48 hours

      With the Anthropic agreement, CoreWeave's platform is now utilized by nine of the ten leading AI model providers, a market penetration figure referenced in the company's press release. The customer base, developed alongside Microsoft, also includes Meta, OpenAI, Mistral, Cohere, IBM, and Nvidia, along with a sub-leasing arrangement allowing Microsoft to provide some of CoreWeave's capacity to external clients. The relationship with Meta saw a significant enhancement on April 9,