CoreWeave, via a public SEC (Form 8-K) filing, disclosed that it has locked in a deal with Meta worth up to $14.2 billion. The deal will span into 2031 with options to stretch further. CoreWeave will grant Meta access to some of its most advanced hardware, including Nvidia’s GB300 systems (NVL72 architecture). The arrangement strengthens Meta’s ability to train, deploy, and scale large AI models without building all the supporting infrastructure itself.
Interestingly, this deal came just days after CoreWeave expanded its partnership with OpenAI by $6.5 billion, pushing its total collaboration to $22.4 billion. CoreWeave also has an arrangement with Nvidia to buy any unused compute capacity for up to $6.3 billion through 2032, another buffer to smooth demand swings.
Stock soars 13% in trade
Markets reacted swiftly to the announcement, CoreWeave’s stock jumped ~13% in trade at $138.57. The surge reflected optimism that the contract provides long-term revenue visibility and validates CoreWeave’s position in AI infrastructure. Meanwhile, Meta’s stock (META) saw little direct upward pressure.
Analysts are bullish on the deal, with some initiating coverage of CoreWeave with an “outperform” rating. Seeing upside potential amid robust AI demand, they have revised their target to $175 per share.
Other mega AI infrastructure deals
Nvidia recently committed $100 billion in a sweeping collaboration with OpenAI to build next-generation AI data centers, often referred to as “AI factories.” The deal aims to showcase OpenAI’s rising compute needs and further entrench Nvidia at the heart of AI infrastructure.
In another development, Microsoft sealed a multi-year pact with Nebius to lease AI compute power from its upcoming data center in New Jersey. This deal is valued at roughly $17 billion and is a clear indicator that tech giants are outsourcing a large portion of their compute burden to specialized infrastructure providers.
Earlier in 2025, Applied Digital saw its stock soar after agreeing to a 15-year lease deal with CoreWeave, estimated at $7 billion, for data center capacity.
The AI cloud race intensifies
In the larger AI infrastructure race, this deal signals that hyperscalers will increasingly lean on external cloud providers specialized in GPU/accelerator compute. Don’t be surprised if more such deals and consolidation among infrastructure players occur in the near future.