Meta Commits $35 Billion to CoreWeave in Largest AI Infrastructure Deal in Corporate History

Meta Platforms announced on April 8, 2026 that it would spend an additional $21 billion with CoreWeave, the cloud computing company specializing in GPU-accelerated infrastructure, on top of a previously announced $14.2 billion contract — bringing the social media giant's total CoreWeave commitment to more than $35 billion. The announcement, made the same day as the US-Iran ceasefire and a 1,325-point Dow rally, underscored that corporate America's appetite for artificial intelligence infrastructure has not dimmed despite five weeks of geopolitical conflict that had rattled broader equity markets.

CoreWeave, which went public in early 2025 and has grown rapidly as one of the dominant providers of on-demand GPU computing for AI training and inference workloads, confirmed the expanded deal in a statement that called Meta a "foundational partner." The infrastructure will support Meta's next-generation AI models, including large-scale training runs for successors to its Llama family of open-source language models and for AI systems powering Meta AI across Instagram, WhatsApp, Facebook, and the company's Ray-Ban smart glasses platform. Meta Chief Technology Officer Andrew Bosworth described the expanded deal as "the infrastructure layer for the next decade of AI at Meta."

The scale of the commitment reflects a broader pattern among hyperscalers — the handful of companies large enough to shape the global computing market through their capital spending decisions. Microsoft, Google, Amazon, and Meta together are projected to spend more than $300 billion on AI infrastructure in 2026, according to analysts at Morgan Stanley. The CoreWeave deal gives Meta access to clusters of tens of thousands of Nvidia H200 and next-generation Blackwell GPUs without the years-long lead time required to build proprietary data centers from scratch.

Nvidia shares rose 4.2% on the day of the announcement, bringing the chipmaker's year-to-date gains to 34%. Nvidia's H100 and H200 GPUs remain the primary computational substrate for large-model AI training, and any expansion of major AI infrastructure deals directly benefits the company's data center revenue, which has grown at triple-digit rates since 2023. Analysts noted that the Meta-CoreWeave deal also served as an implicit signal to investors that the AI spending cycle would continue regardless of macro uncertainty.

Meta's capital expenditure guidance for 2026, issued in February, had forecast $60 to $65 billion in total spending for the year. The CoreWeave expansion technically falls outside the data center construction line of the company's capital budget, since it is a contracted services commitment rather than owned hardware, but it represents a total obligation that dwarfs what most companies spend on technology in a decade. CEO Mark Zuckerberg has consistently argued that the cost of underinvesting in AI — losing competitive position to Google or to emerging Chinese AI companies — is greater than the cost of overbuilding infrastructure that can be monetized through the company's advertising business.