Ever wonder where all the compute power behind Claude comes from? As of this week, there’s one more answer: CoreWeave.
The Deal
On April 10, CoreWeave officially announced a multi-year agreement with Anthropic. The gist: Anthropic will use CoreWeave’s cloud platform to run Claude models at production scale. Infrastructure is set to come online later this year, starting with a phased rollout — with room to expand.
CoreWeave didn’t disclose the deal’s value. But the market spoke for itself: CoreWeave’s stock jumped 11% on the news. This was the company’s second major deal in just 48 hours — Meta had just announced a $21 billion expansion of its own CoreWeave partnership.
Why This Matters
With Anthropic on board, nine of the ten leading AI model providers now run on CoreWeave’s platform. That’s a remarkable concentration. CoreWeave is firmly positioning itself as the GPU cloud for AI workloads — alongside the big hyperscalers like AWS, Azure, and GCP.
For Anthropic, this deal means more infrastructure flexibility. Until now, Anthropic was heavily tied to Google Cloud and AWS. Adding CoreWeave diversifies that dependency and gives Anthropic more negotiating leverage.
The Bigger Picture
The timing isn’t accidental. Anthropic just launched Managed Agents, Cowork is generally available, and demand for Claude is growing fast. More users mean more compute needs. And CoreWeave has consistently topped MLPerf benchmarks for inference workloads — which is exactly what Claude needs in production.
What I find notable about this deal: Anthropic is building a multi-cloud setup rather than locking itself into a single provider. That’s a smart strategic move — and shows that Claude demand is big enough to keep multiple infrastructure partners busy at the same time.
Sources: