CoreWeave inks multiyear capacity deal with Anthropic, tapping a range of Nvidia chips as it grows to 43 data centers

Deal details

It has been reported that CoreWeave has signed a multiyear agreement with Anthropic to provide GPU capacity across its U.S. data centers, including a variety of Nvidia chips. The deal, allegedly intended to help Anthropic run its Claude family of models, adds to a growing roster of partnerships that position CoreWeave as a go‑to specialist for AI compute. Quick and quiet wins like this keep the lights on — and the GPUs humming.

Why it matters

CoreWeave now operates 43 active data centers, a sign that the niche GPU cloud players are scaling fast to meet ravenous demand for model training and inference. Nvidia’s architecture continues to dominate the market, and companies that can aggregate and sell that capacity cheaply and flexibly are suddenly in the spotlight. Who wants to wrestle with long lead times for hardware when you can rent tuned clusters by the month?

The bigger picture

This deal underscores an industry shift: some AI developers are diversifying away from hyperscalers and toward specialized providers that promise lower latency and more predictable pricing for GPU-heavy workloads. For CoreWeave, landing a multiyear customer like Anthropic feels like validation — a payoff for an all‑in bet on AI infrastructure. For the rest of the market, it’s a reminder that compute is the battleground now.

What to watch next

Keep an eye on capacity availability, Nvidia’s chip roadmap, and whether more model builders follow Anthropic’s lead. Pressure on GPU supply could make these partnerships strategic lifelines — and fast. In short: the race for compute is heating up, and CoreWeave just picked up another runner.

Sources: bloomberg.com