Dive Brief:
- AI cloud provider CoreWeave expanded its existing relationship with Meta through 2032 in a $21 billion deal as the social media giant seeks more cloud compute capacity to power its AI deployments, CoreWeave shared in a Thursday announcement.
- The increased capacity will be spread across multiple locations, with some initial deployments of Nvidia’s Vera Rubin AI computing platform included. Distributing compute improves the resiliency and scalability of Meta’s AI operations, the announcement said.
- The CoreWeave partnership expansion is the latest in a string of high-profile deals for specialized AI cloud providers. In March, Meta entered into a long-term compute agreement with AI cloud platform Nebius, indicating the tech company’s growing capacity needs. CoreWeave also struck a multi-year strategic partnership with AI search engine Perplexity last month.
Dive Insight:
Big tech’s massive AI infrastructure buildouts to support emerging projects such as Muse Spark, the first in a series of large language models purpose-built to power Meta products, have contributed to explosive growth among specialized AI cloud providers, also referred to as neoclouds.
Surging AI infrastructure demand is expected to propel the neocloud market to $400 billion in revenues by 2031, representing a 58% compound annual growth rate, according to a report from Synergy Research Group. Neocloud revenues exceeded $25 billion in 2025 and reached $9 billion in Q4 alone, a 223% increase year over year.
Neoclouds are purpose-built to deliver high performance computing capabilities for AI workloads, which makes them a draw for hyperscalers and other tech industry giants starving for compute power. Market leaders include CoreWeave, Crusoe, Core Scientific, Lambda, Nebius and Nscale, the Synergy Research Group report said.
Under Meta’s five-year agreement with Nebius, the neocloud provider will enable $12 billion of compute capacity starting in 2027. Similar to the CoreWeave deal, added capacity will be spread across multiple locations and will be based on large-scale deployments of the Vera Rubin platform. Meta also committed to purchasing up to $15 billion in additional compute capacity over the five-year period for a total contract value of $27 billion.
As AI workloads continue putting pressure on companies to deliver necessary compute, neoclouds could result in a structural realignment of compute systems, said Jeremy Duke, founder and chief analyst at Synergy Research Group.
“Traditional hyperscale systems were conceived around a form of generalized elasticity, whereas AI workloads impose far more rigid constraints — particularly around parallelism, locality and the concentration of compute,” Duke said in the report. “Neoclouds are, in effect, an architectural response to those constraints.”
