Meta Platforms has significantly expanded its artificial intelligence infrastructure partnership with cloud computing provider CoreWeave, committing to a substantial new $21 billion agreement for AI cloud capacity that extends through December 2032. The expanded arrangement, formalized on March 31 according to securities filings, builds upon the companies' existing master services agreement established in 2023 and adds to a previous $14.2 billion commitment disclosed in September.
Strategic Timing for AI Expansion
The announcement arrives at a pivotal moment for Meta's AI ambitions, coming just one day after the company unveiled Muse Spark as the first product from its superintelligence research group and following the somewhat disappointing launch of its Llama 4 model. For CoreWeave, the agreement represents progress in diversifying its customer base, reducing reliance on Microsoft, which was projected to account for approximately 67% of its revenue next year according to previous filings.
Infrastructure Details and Executive Commentary
CoreWeave plans to distribute the additional computing capacity across multiple geographic locations, marking the first deployment of Nvidia's Vera Rubin platform for some workloads. "This is another example that leading companies are choosing CoreWeave's AI cloud to run their most demanding workloads," stated Chief Executive Michael Intrator. The company will provide reserved capacity through the end of 2032, including exercised options from Meta's earlier orders.
Concurrent with the contract announcement, CoreWeave tapped debt markets aggressively, revealing a $3 billion convertible note offering maturing in 2032 with an additional $450 million option for buyers. Separately, the company upsized a senior notes offering from $1.25 billion to $1.75 billion, pricing the debt with a 9.75% coupon. These financing moves underscore the capital-intensive nature of scaling AI infrastructure.
Aggressive Capital Expenditure Plans
CoreWeave management shows no signs of slowing its spending trajectory. In February, the company projected capital expenditures could reach $35 billion in 2026, substantially above the $14.9 billion planned for 2025. Chief Financial Officer Nitin Agrawal noted at the time that "our revenue backlog grew to $66.8 billion, more than four times where we began the year," referring to contracted sales not yet recognized as revenue.
Investors face a complex assessment balancing growth potential against execution risk. D.A. Davidson analyst Alexander Platt observed in February that CoreWeave was "being punished for either having too little capex or too much capex," while acknowledging no obvious capacity ramp-up issues as a positive signal. The company's shares rose approximately 3.4% to $92 in after-hours trading following the announcement.
Competitive Landscape and Financial Considerations
CoreWeave positions itself as a specialized provider for AI companies requiring high-density Nvidia chip clusters, competing against cloud giants Microsoft and Alphabet's Google. The Meta expansion demonstrates that even technology leaders continue sourcing external computing resources while developing their internal infrastructure capabilities.
Despite the substantial contract win, financing challenges persist. CoreWeave's balance sheet showed nearly $30 billion in debt and lease obligations at year-end, and according to February reports, a significant portion of its signed revenue depends on new data centers becoming operational as scheduled. The company's rapid fundraising efforts highlight the substantial costs of scaling operations, with investors closely monitoring CoreWeave's ability to meet growth targets on schedule.
Meta's latest commitment provides CoreWeave with another major vote of confidence from the social media giant, reinforcing its position among key infrastructure providers supporting the AI boom. However, the simultaneous debt offerings emphasize the substantial financial requirements of maintaining competitive expansion in the capital-intensive cloud computing sector.
