Particle.news

Meta Expands CoreWeave Deal to $21 Billion for AI Compute Through 2032

The long-term pact secures outside GPU power for Meta’s AI inference push.

Overview

  • A contract announced Thursday between Meta and CoreWeave sets about $21 billion of dedicated AI cloud capacity through December 2032.
  • The agreement centers on inference, which means running AI models for users in real time and requires fast, reliable compute close to where people use Facebook, Instagram, WhatsApp, and Messenger.
  • CoreWeave will deploy capacity across multiple data-center sites and include early rollouts of NVIDIA’s next‑generation Vera Rubin hardware to boost performance for production workloads.
  • To fund the buildout tied to new contracts, CoreWeave is planning more than $4 billion in debt offerings, including $3 billion in convertible senior notes due 2032 and $1.25 billion in senior notes due 2031.
  • The new commitment expands a prior pact of roughly $14.2 billion and includes Meta exercising an option under that earlier deal, so the figures are not a simple sum.