Bloomberg reported that Meta Platforms contracted roughly 1.6 gigawatts (1.6 GW) of AI computing capacity from data center developer Crusoe, spanning two planned campuses in Childress, Texas, and Warrenton, Missouri. Reuters, which carried the Bloomberg report, said neither Meta nor Crusoe immediately responded to requests for comment and that the report could not be independently verified. Bloomberg and subsequent coverage said financial terms and delivery timing were not disclosed. Separate reporting by MLQ and other outlets notes that Crusoe has said its contracted capacity approaches 5 GW with a broader pipeline above 20 GW, which would make a 1.6 GW commitment a material share of Crusoe's contracted base if confirmed. Observers will watch for formal confirmation, timing, and energy supply details.
What happened
Bloomberg reported that Meta Platforms contracted roughly 1.6 gigawatts (1.6 GW) of AI computing capacity from data center developer Crusoe, covering two planned sites in Childress, Texas, and Warrenton, Missouri. Reuters, which carried the Bloomberg account, said neither company immediately responded to requests for comment and that the report could not be independently verified. The Bloomberg coverage and follow-on reports stated that financial terms and timelines were not disclosed.
Technical details
Per MLQ and other reporting, Crusoe has previously said its contracted AI infrastructure capacity approaches 5 GW, with a development pipeline exceeding 20 GW; the reported Meta commitment would represent a significant portion of Crusoe's disclosed contracted base if those figures are accurate. MLQ also reported site-level details, including that the Childress campus uses Lancium-linked clean-campus infrastructure with a 1 GW ERCOT-approved interconnect, and prior coverage noted Crusoe has announced large commitments to other hyperscalers on its Texas campus.
Industry context
Industry-pattern observations: large-scale AI training and inference require high-density power and grid interconnect capacity, which has driven some cloud and hyperscaler buyers to secure long-term, large-capacity arrangements with third-party developers. Public reporting frames this deal as another example of firms tapping external 'neocloud' or developer-built capacity to accelerate GPU deployment beyond in-house construction timelines.
What to watch
Observers will look for formal confirmation from Meta or Crusoe, the split of capacity between the two campuses, contract duration and pricing, and the timeline for equipment delivery and commissioning. Energy sourcing, grid interconnect upgrades, and local permitting or tax arrangements at the Childress and Warrenton sites are additional indicators of when capacity could be usable for large-scale model training.
For practitioners
Editorial analysis: procurement, infrastructure, and ML platform teams should note that third-party capacity leasing can change deployment timelines and cost structures. Teams evaluating workload placement should track announcement-to-delivery lag, power availability, and interconnect characteristics when estimating throughput for large-scale training or serving.
Scoring Rationale #
Securing grid-scale power commitments matters to ML infrastructure planning because 1.6 GW materially increases available compute for large model training. The story is notable for infrastructure teams and procurement but not a frontier-research shock.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
[Active Search Campaigns by BudgetEasy](/problems/sql/active-search-campaigns-by-budget)
[High CPC Clicks & Poor Landing PagesMedium](/problems/sql/high-cpc-clicks-poor-landing-page)
[Campaign ROAS by Attribution ModelHard](/problems/sql/campaign-roas-by-attribution-model)
250 free problems · No credit card