Microsoft will lease Crusoe's 900 MW data center in Abilene, Texas, after Oracle and OpenAI reportedly withdrew, with the first building expected by mid-2027
Microsoft Corp. will occupy a data center project previously earmarked for Oracle Corp. and OpenAI, scooping up 900 megawatts …
Context & Ripple Effects
The Abilene project has moved rapidly from Oracle's abandoned expansion plans to Microsoft's advanced lease talks, as reported earlier in March. The latest report raises the planned commitment from the roughly 700 MW cited in the earlier Microsoft-Crusoe lease report to 900 MW.
The tenant change matters because Crusoe had previously raised $11.6B to expand a Texas facility being built for OpenAI. It shows how a large powered-data-center build can be reassigned when an intended customer exits.
First-order effects
- Microsoft gains a path to occupy 900 MW of planned Crusoe capacity in Abilene, with the first building targeted for mid-2027.
- Crusoe replaces the reported Oracle/OpenAI demand at the site with Microsoft, preserving a customer for the planned project.
Second-order effects
- Oracle and OpenAI's withdrawal makes their future capacity plans a separate procurement question, while Microsoft becomes the immediate buyer attached to this Abilene build.
- For data-center developers and infrastructure financiers, the switch puts greater weight on securing creditworthy replacement tenants when large AI capacity commitments change.
Third-order effects
- If such reassignments become common, AI data-center projects may increasingly be financed and operated as flexible powered-capacity assets rather than facilities tied permanently to one model developer or cloud customer.
- The limiting competitive asset shifts further toward deliverable power and construction execution: announced capacity matters less than whether a site can be completed and leased on schedule.
The trend: AI infrastructure is evolving into a market for transferable, long-duration powered capacity, with major cloud and AI buyers reshuffling commitments across large projects.