CoreWeave's stock popped after a deal to power Anthropic's Claude. The neocloud announced a multi-year agreement Friday to supply Nvidia GPU capacity across U.S. Data centers for Anthropic's production-scale workloads. Shares jumped on the news — moving noticeably in both morning and later trading.

Deal details and market reaction

The companies did not disclose the price of the deal. CoreWeave didn't disclose the value of the Anthropic agreement in its Friday release, and Anthropic hasn't published financial terms either. Still, investors treated the announcement as a meaningful win: CNBC reported shares climbed as much as 11% after the news, while Yahoo Finance noted an earlier 5% uptick with shares trading around $97.92 in morning hours.

The timing sharpened the market's interest. The Anthropic news arrived just a day after CoreWeave revealed a $21 billion expansion of its Meta partnership, following an earlier commitment that the company described as $14.2 billion. That pair of announcements pushed CoreWeave further into the spotlight as one of the industry's most in-demand GPU cloud providers.

What Anthropic gets — and why it matters

Anthropic will tap CoreWeave's Nvidia GPUs to run Claude at production scale, the companies said. The rollout will begin in phases, with CoreWeave expecting to bring compute online for Anthropic later this year, according to the company release. Anthropic uses multiple suppliers for capacity. It runs its main training workloads on AWS Trainium hardware and is now adding more GPUs to meet growing enterprise and developer demand.

The move reflects how Anthropic's infrastructure needs have ballooned. Source reporting shows Anthropic's annualized revenue run rate topped roughly $30 billion in early April 2026, leaping from about $9 billion at the end of 2025. That surge in demand for Claude and Claude Code has forced the company to diversify compute partners and architectures so it can keep scaling without hitting capacity or vendor limits.

CoreWeave’s growth story — from crypto miner to AI landlord

CoreWeave started life as a crypto miner in 2017, buying large volumes of Nvidia GPUs to mine Ethereum before shifting into GPU-on-demand cloud services when crypto margins tightened. The pivot paid off in the AI boom that took off in 2023.

CoreWeave went public on Nasdaq on March 28, 2025, at $40 a share, raising about $1.5 billion and valuing the company at roughly $23 billion, according to reporting.

Numbers released by the company and reported in industry coverage show CoreWeave now operates 32 data centers, runs more than 250,000 GPUs, and holds about 1.3 gigawatts of contracted power capacity. The company reported $5.13 billion in revenue for 2025 — up 168% year-over-year — and guided for more than $12 billion in 2026, backed by a contracted backlog that sources put above $66 billion.

Customer concentration and diversification

However, rapid growth has introduced risks. In 2025 Microsoft accounted for roughly two-thirds of CoreWeave's revenue, a concentration investors flagged before the IPO. Microsoft is also building more in-house capacity and developing its own models, so how much of its future compute needs will stick with third-party providers remains a question.

Adding Anthropic — along with the expanded Meta commitment — is clearly part of CoreWeave's strategy to broaden its customer mix. The company now counts nine of the top ten foundational AI model providers among its customers, the company said, with reports noting xAI as the notable exception. By spreading demand across multiple hyperscalers and large model makers, CoreWeave aims to reduce the single-customer dependency that critics worried about.

Industry context: GPUs, competition and scale

Demand for Nvidia GPUs has outpaced supply for years, and firms building large language models are racing to secure capacity. CoreWeave's inventory of GPUs and power capacity is the backbone of its pitch: customers get access to specialized hardware in multiple U.S. Data centers without having to build their own farms. That matters because training and serving modern models requires both raw GPU compute and dense, reliable power and networking at scale.

At the same time, hyperscalers and cloud rivals are competing fiercely. Amazon, Microsoft, and Google have vast internal resources and are increasingly offering proprietary chips and services. Anthropic itself runs sizable parts of training on AWS Trainium, while also looking to GPU providers for inference and other high-throughput workloads. The multi-vendor approach reduces supplier risk for large model providers and lets them optimize for cost and performance across different types of compute.

Operational takeaways and near-term outlook

Right now, CoreWeave says it plans a phased rollout of the Anthropic deployment, which gives both companies time to scale capacity and tune operations. For CoreWeave, that means integrating Anthropic workloads into its U.S. Data centers and bringing additional GPUs online later this year. For Anthropic, the deal adds another major source of GPU capacity that can support peak demand for enterprise Claude instances and Claude Code deployments.

From a revenue perspective, the deal could add materially to CoreWeave's 2026 outlook if it converts to long-term contracted usage, but without disclosed terms it's hard to quantify the impact precisely. Still, the pattern is clear: large AI developers are signing multi-year deals with specialized cloud providers while also maintaining relationships with hyperscalers and chip vendors. The result is a more crowded, more competitive market for AI infrastructure.

What the move means for customers and the market

Customers building on Claude are expected to benefit from growth.eater capacity and geographic diversity in the infrastructure layer. More back-end options mean lower risk of bottlenecks during big launches or spikes in demand. For enterprises buying Claude services, that translates into fewer outages and potentially faster response times when usage soars.

Analysts and investors will be watching whether CoreWeave can keep converting big-name partnerships into steady, recurring revenue without losing margin to rapid infrastructure buildouts. The company has already signaled ambitious growth goals — more than $12 billion in revenue guidance for 2026 — and the Anthropic and Meta deals are the kind of headline contracts that back those targets. Whether CoreWeave can maintain margins while scaling power and data-center capacity will be a central operational test.

Final operational detail

CoreWeave said it didn't disclose the value of the Anthropic deal and that, among the top ten foundational AI model providers, xAI remains the only one not on its platform. The company expects to bring compute online for Anthropic later this year, per its release.

Related Articles

CoreWeave declined to disclose the value of the Anthropic deal.