The obvious question is why Meta — which builds its own data centers and has been designing custom AI chips — would spend $35 billion with an external cloud provider. The answer is speed and flexibility: Faster deployment: CoreWeave can bring GPU capacity online faster than Meta can build new data centers from scratch Nvidia access: CoreWeave has secured significant Nvidia GPU allocations, including early access to next-gen architectures like Vera Rubin Flexibility: Cloud capacity can be scaled up or down more easily than owned infrastructure Bridge capacity: Meta needs computing power now while its own data centers are still under construction

Meta Commits Additional $21 Billion to CoreWeave for AI Cloud Infrastructure

By Jaspal April 9, 2026 Updated: April 9, 2026

Meta just doubled down on its AI infrastructure bet in a massive way. The company has committed an additional $21 billion to CoreWeave for AI cloud infrastructure, on top of a prior $14.2 billion arrangement. That brings the total deal to a staggering $35.2 billion through December 2032.

The Deal Details

The new $21 billion commitment covers the period from 2027 to 2032, during which CoreWeave will provide Meta with dedicated AI computing capacity across multiple locations. The infrastructure will feature early deployments of Nvidia's Vera Rubin platform — the next-generation GPU architecture that succeeds Blackwell.

CoreWeave, which essentially rents out Nvidia GPU capacity at scale, has become one of the most important infrastructure companies in the AI ecosystem. This expanded Meta deal, announced on April 9, sent CoreWeave shares up 3% in pre-market trading. The company is also raising an additional $3 billion in fresh debt to fund the expansion.

Meta's AI Spending Is Staggering

This deal needs to be understood in the context of Meta's overall AI investment. In its last earnings report, Meta said it plans to spend between $115 billion and $135 billion on capital expenditures in 2026 — above Wall Street estimates and nearly double its 2025 capex. The CoreWeave deal represents just one piece of this massive infrastructure push.

Mark Zuckerberg has been clear that he sees AI infrastructure as a competitive moat. With Muse Spark now powering Meta AI across its family of apps, the company needs the compute capacity to serve AI to over 3 billion users.

Why CoreWeave?

The obvious question is why Meta — which builds its own data centers and has been designing custom AI chips — would spend $35 billion with an external cloud provider. The answer is speed and flexibility:

Faster deployment: CoreWeave can bring GPU capacity online faster than Meta can build new data centers from scratch
Nvidia access: CoreWeave has secured significant Nvidia GPU allocations, including early access to next-gen architectures like Vera Rubin
Flexibility: Cloud capacity can be scaled up or down more easily than owned infrastructure
Bridge capacity: Meta needs computing power now while its own data centers are still under construction

The AI Infrastructure Arms Race

Meta isn't alone in this spending spree. The enterprise AI consolidation trend is driving unprecedented infrastructure investment across the industry. OpenAI's $500 billion Stargate project (though paused in the UK) continues in the US, while Google and Amazon are also investing tens of billions in AI compute.

The TSMC packaging bottleneck means that access to advanced Nvidia chips remains constrained, making deals like this CoreWeave partnership strategically critical. Companies that lock up compute capacity now will have an advantage for years.

What This Means

The $35 billion total commitment to CoreWeave underscores a fundamental reality of the AI era: the companies that can afford to build or rent the most computing infrastructure will have the strongest competitive position. Meta is betting that the returns from AI-powered advertising, commerce, and user engagement will more than justify this extraordinary spending.

Frequently Asked Questions

How much is Meta spending with CoreWeave total?

Meta's total commitment to CoreWeave is approximately $35.2 billion — $14.2 billion from the original deal plus $21 billion in the new expanded agreement announced April 9, 2026, running through December 2032.

What is CoreWeave?

CoreWeave is an AI cloud infrastructure company that rents out Nvidia GPU computing capacity at scale. It has become one of the most important infrastructure providers in the AI ecosystem, with major deals from Meta, Microsoft, and other tech giants.

What will the infrastructure include?

The expanded deal spans multiple locations and features early deployments of Nvidia's Vera Rubin platform — the next-generation GPU architecture that succeeds Blackwell — aimed at boosting efficiency, reliability, and scale.

How much is Meta spending on AI total in 2026?

Meta plans to spend between $115 billion and $135 billion on capital expenditures in 2026, nearly double its 2025 spending. The CoreWeave deal is one component of this massive AI infrastructure investment.