Meta locks in 1.6 GW of AI capacity in Crusoe data center deal

Meta Platforms locked in 1.6 GW of AI computing capacity through a deal with Crusoe Energy, as the social media giant accelerates infrastructure spending to support its artificial intelligence ambitions.

The agreement covers computing power from two Crusoe data centers in Missouri and Texas, according to a person familiar with the matter who asked not to be identified discussing private terms. Meta will purchase capacity from both facilities, the person said.

The 1.6 GW commitment is among the largest single AI computing deals by a hyperscaler this year. One gigawatt can power roughly 750,000 homes, according to industry estimates. The deal comes as Meta and other tech giants compete for data center capacity during a global shortage driven by surging AI demand.

The agreement highlights the enormous power requirements of AI infrastructure and the tensions they create in local communities. In Montgomery County, Missouri, Google and Amazon data center plans have divided residents over tax benefits versus resource strain, with a local advocacy group taking the county to court over water and power concerns. In Louisiana, a Public Service Commission consultant warned that a $1.8 billion gas plant acquisition by Entergy is needed mainly because of Meta's data center in Richland Parish.

The deal strengthens Crusoe's position as a dedicated AI infrastructure provider. The company specializes in modular data center construction and has become a key partner for hyperscalers seeking to bypass traditional colocation bottlenecks. Crusoe's model offers faster deployment timelines compared with conventional data center operators, a critical advantage as tech companies compete for limited capacity.

Meta's commitment comes as the company projects its AI-related capital expenditure will reach $60 billion to $80 billion annually by 2026, according to company filings. The Crusoe deal provides a portion of that capacity without requiring Meta to build and operate the facilities itself, a strategy that preserves balance sheet flexibility while still securing the compute power needed to train and deploy large language models.

The arrangement carries implications for the broader data center industry. Equinix and Digital Realty, the two largest data center REITs, face potential competition from newer entrants like Crusoe that offer dedicated AI infrastructure rather than shared colocation space. At the same time, utilities across the US are grappling with demand forecasts that have surged beyond expectations, driven almost entirely by data center construction. In Missouri, Google and Amazon's combined "Mega Site" spans about 1,900 acres in Montgomery County, where residents have formed an advocacy group to challenge the projects in court over water and power concerns.

The deal also affects the AI chip market. Meta's growing compute capacity reduces its dependence on Nvidia's graphics processing units, which have been in short supply. While Meta remains one of Nvidia's largest customers, securing dedicated data center capacity gives the company more flexibility in how it deploys its AI workloads across different hardware configurations.

For Meta, the deal provides a hedge against rising power costs. The company's energy consumption has climbed as it trains and deploys large language models, and securing long-term capacity helps manage what is becoming one of its largest operating expenses. Meta shares have gained 42 percent over the past 12 months, outpacing the S&P 500's 18 percent gain, as investors reward the company's AI investment strategy. The stock trades at about 24 times forward earnings, according to data compiled by Bloomberg.

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