← Back to feed

Crusoe and Lancium have announced a 1.0 gigawatt AI data center campus in Childress, Texas — the second execution of a partnership model the two companies proved out in Abilene.

The 270-acre site is Lancium-owned, grid-connected, and will integrate behind-the-meter solar and energy storage. Lancium handles land acquisition, power interconnect, and energy orchestration. Crusoe designs, builds, and operates the compute layer. Construction begins Q3 2026, with thousands of construction jobs and more than 100 permanent positions expected for Childress County.

The partnership model

The Abilene template — Lancium owns and powers, Crusoe builds and operates — has now been reproduced at scale. Abilene became part of the Stargate buildout anchored by OpenAI. Childress follows the same structure but at the next tier of scale.

The division of labor matters. Most gigawatt-scale builds are single-company or hyperscaler-owned. The Lancium-Crusoe split allows each party to do what it is actually good at: Lancium has the land relationships and grid interconnect expertise in Texas; Crusoe has the AI infrastructure design and operations depth. The closed-loop cooling system being deployed at Childress — to reduce water consumption — reflects Crusoe’s operating fingerprint.

Meta as tenant

Data Center Dynamics reports Meta as the probable anchor tenant for Childress capacity. The fit is logical. Meta’s compute expansion in 2026 has been among the most aggressive in the industry: a 5GW Louisiana campus, $21 billion in CoreWeave capacity locked through 2032, and its MTIA 400 inference chip entering production in September. A Crusoe-operated Texas facility would diversify Meta’s geography and give it grid-connected Texas capacity ahead of those MTIA deployments.

Meta has not confirmed. Crusoe and Lancium announced only that the campus is “purpose-built to support the advanced AI computational requirements of a leading hyperscale technology company.”

Scale context

At 1GW, Childress sits at the upper end of single-campus AI infrastructure announcements in 2026. For reference:

• Hut 8 / Beacon Point: 352MW
• CleanSpark mystery hyperscaler: 300MW
• Apollo / Blackstone / Google TPU campus: 500MW

A 1GW campus is a different category. The power interconnect alone — which Lancium is responsible for securing — represents a multi-year grid commitment. That Lancium has already done this in Abilene is the reason Crusoe is betting on a repeat.