Meta secures 1.6 GW of AI computing from data center firm Crusoe

Meta Platforms Inc. has secured agreements to buy roughly 1.6 gigawatts of AI computing capacity from data center developer Crusoe across two sites, as the social media giant accelerates infrastructure spending to support its artificial intelligence ambitions.

The capacity — enough to power about 1.2 million US homes at any given time — will come from data centers in Childress, Texas, and Warrenton, Missouri, according to people familiar with the matter who asked not to be identified because the discussions are private. Meta and Crusoe declined to comment.

"Meta is making a long-term bet that AI compute demand will keep outstripping supply, and they're willing to commit massive capacity years in advance to secure it," said Rachel Kim, an analyst covering AI infrastructure at Edgen. "The question is whether the revenue from AI products will eventually catch up to the scale of this buildout."

The Crusoe deal adds to Meta's sprawling infrastructure program. The company has pledged to invest $600 billion in US infrastructure and jobs over the next three years, including a nearly 4,000-acre campus in Louisiana designed to deliver up to 5 GW of capacity. Meta recently began selling subscriptions to its AI chatbot for the first time, a key step toward monetizing the technology.

Crusoe, founded in 2018, specializes in building and operating large-scale data centers for AI workloads. The company said earlier this month it had contracts for 4.9 GW of capacity, with more than 40 GW in its total project pipeline. Its other tenants include Oracle Corp., Microsoft Corp. and Alphabet Inc.'s Google, though some partnerships have hit snags — Bloomberg reported last week that Crusoe was being pushed out of a planned Wyoming data center after earlier deal talks with Google fell through.

The agreements underscore the breakneck pace of hyperscaler infrastructure spending. Meta joins Amazon.com Inc., Microsoft and Google in committing tens of billions of dollars annually to data center construction, driven by the computational demands of training and running large language models. The four companies collectively spent more than $230 billion on capital expenditures in 2025, with analysts projecting further increases this year.

Meta shares, trading at about 18 times forward earnings, rose 1.7% to $577.22 on Thursday. The stock has gained roughly 30% over the past 12 months, though investors remain split on whether the AI infrastructure buildout will generate returns proportional to its cost. Meta's digital advertising business remains its primary revenue engine, contributing more than 98% of total sales in the most recent quarter.

This article is for informational purposes only and does not constitute investment advice.