VAST Data has signed a $1.17 billion commercial agreement with CoreWeave — a deal that feeds storage and software into one of the most aggressive AI cloud buildouts on the planet.

Inside CoreWeave's data halls, 250,000 NVIDIA GPUs hum today. By mid-2026, that number crosses 400,000. Each new accelerator needs data plumbing. That plumbing is now VAST's job.

The agreement, announced this week, carries no disclosed end date. It lets CoreWeave "manage and secure hundreds of petabytes of data at a time," the company said, with POSIX compliance that supports shared access across multiple compute instances.

What does the $1.17B actually buy?

VAST's AI Operating System — built on its DASE (Distributed and Shared Everything) control plane — bundles six services under one roof:

• DataSpace
• DataBase
• DataStore
• DataEngine
• AgentEngine
• InsightEngine

CoreWeave's broader storage layer still leans on multiple vendors. Dedicated clusters pull from VAST, WEKA, DDN, IBM Spectrum Scale and Pure Storage. The S3-compatible object tier uses CoreWeave's own Local Object Transport Accelerator, or LOTA, to cache data directly on GPU nodes.

Who says what

Renen Hallak, VAST's founder and CEO, framed the pact as roadmap alignment. "At VAST, we are building the data foundation for the most ambitious AI initiatives in the world," he said. "Our deep integration with CoreWeave is the result of a long-term commitment to working side by side at both the business and technical level. By aligning our roadmaps, we are delivering an AI platform that organizations cannot find anywhere else in the market."

Brian Venturo, CoreWeave's co-founder and Chief Strategy Officer, was equally direct. "The VAST AI Operating System underpins key aspects of how we design and deliver our AI cloud," he said.

Why CoreWeave matters

CoreWeave launched as an NVIDIA-powered crypto mining outfit. It pivoted to GPU cloud services in 2019. It hit the public markets in March 2025, raising roughly $1.5 billion at $40 a share. The stock traded recently at $114.42, lifting market cap to $56.65 billion.

The operator runs 28 US data centers and five more in Europe. It claims customers can spin up 10,000 GPUs in under eight minutes. By mid-2026, CoreWeave projects more than 400,000 GPUs online — a 60% jump from current capacity.

Independent benchmark firm SemiAnalysis keeps noticing. Its latest ClusterMAX 2.0 ranking again placed CoreWeave as the only Platinum-tier cloud, the second consecutive Platinum nod.

Competitive context

CoreWeave sits atop the neo-cloud pile — the specialized GPU providers carved out since 2019 to challenge AWS, Microsoft Azure and Google Cloud. Its rivals include Lambda Labs, Crusoe, Nebius, Together.AI and Core42. VAST counts Lambda Labs, Crusoe and Core42 among its customers, per the company's disclosures.

That footprint turns the $1.17B commitment into something closer to an industry vote than a single-vendor contract.

What it signals for storage

VAST's customer list now spans three of the most aggressive GPU-first clouds in the West — CoreWeave, Lambda Labs and Crusoe — plus Core42 in the Middle East. Each one runs NVIDIA silicon at scale and ships POSIX-compliant storage to AI labs and frontier model builders. By landing the largest of the bunch, VAST tightens its grip on the high end of the AI storage market, where Pure Storage, DDN and IBM Spectrum Scale still compete inside dedicated clusters.

That positioning reframes the headline figure: storage spend is now tracking GPU spend, and the vendors that ride the GPU cycle hardest are selling the biggest tickets.

Forward look

Neither VAST nor CoreWeave has disclosed the deal's length, pricing ramp or volume milestones. What is on the record: CoreWeave is committed to roughly 60% more online GPUs by mid-2026, and every additional accelerator needs data plumbing behind it — which is precisely what this contract now guarantees.