Why it matters
The strategic read
Turns the prior Meta/CoreWeave relationship into a much larger long-duration AI capacity backlog. Meta is reserving external specialist GPU-cloud infrastructure through the Rubin-era buildout while CoreWeave uses contracted demand to support a rapidly expanding data-center and financing stack.
Source-supported terms
What was disclosed
- 01
CoreWeave announced an expanded long-term agreement with Meta Platforms to provide AI cloud capacity through December 2032 for approximately $21B.
- 02
The agreement was filed as an exhibit to a CoreWeave Form 8-K and builds on the companies’ prior September 2025 order form.
- 03
CoreWeave’s Q1 2026 Form 10-Q says the March 31, 2026 order form includes access to new computing capacity through December 20, 2032.
- 04
The same disclosure says the commitment also includes Meta exercising an existing option to access additional computing capacity through April 10, 2032.
- 05
CoreWeave says the expanded agreement contributes to revenue backlog, which reached $99.4B as of March 31, 2026.
- 06
Near-primary market reporting says the dedicated AI cloud capacity will support Meta model training and deployment across multiple locations.
Disclosure boundary
What is known—and what is not
SEC-filed press-release exhibit plus quarterly filing summary; full order form, capacity schedule, data-center sites, GPU/server mix, SLA terms, termination rights, financing links and pricing mechanics are not public.