Meta enters into new AI computing agreements with data center company Crusoe.

      Meta has recently secured new contracts with Crusoe, a data center developer, to acquire computing power, further expanding the extensive infrastructure the company is building for its AI initiatives. The agreements involve two locations—one in Childress, Texas, and another in Warrenton, Missouri—providing Meta with approximately 1.6 gigawatts of capacity, as reported by Bloomberg citing sources familiar with the situation.

      This capacity figure is significant. A gigawatt represents the output of a large power station, and dedicating 1.6 gigawatts to a single company's computing requirements illustrates the shift in AI infrastructure focus from chips to electricity. Meta is not merely purchasing servers; it is now securing power on a scale akin to national grid planning and is collaborating with multiple developers simultaneously.

      Crusoe, founded in 2018, has quickly become a key player, specializing in the construction and operation of large data centers for AI applications. The company already holds partnerships with Oracle, Microsoft, and Alphabet’s Google for three other data center sites.

      Including Meta as a client places Crusoe in a strong position among developers that major AI enterprises rely on when they need capacity more rapidly than they can develop it themselves. For Meta, these deals are part of a larger, more costly strategy. Similar to its peers in Silicon Valley, the company is fervently working to secure computing power wherever it can, with its most significant project being a nearly 4,000-acre campus in Louisiana aimed at providing up to five gigawatts of capacity.

      In light of this, the agreements with Crusoe are a supplements to what is already an impressive expansion effort. However, this expansion is straining local resources. Large data centers require electricity and water at a level that has begun to raise concerns among grid operators and communities, and the competition for electricity has become a critical limitation for the industry, evident in operators striving to acquire gigawatts on different continents. Meta's strategy has been to diversify its partnerships across various developers and regions instead of depending on a single site.

      The financial commitment behind these endeavors is substantial and continues to rise. The largest cloud and AI companies are investing hundreds of billions into infrastructure, and demand is outpacing even the boldest predictions, prompting some operators to lease capacity from unconventional sources to keep up.

      While Meta's agreements with Crusoe represent a minor entry in this financial landscape, they signal a clear direction: acquire electricity as soon as possible, regardless of cost, and address the returns later.

      The 1.6 gigawatts will be utilized for familiar activities, including training and deploying the AI models and agents that the company has oriented itself around. This capacity is essential for the overall strategy; without it, the models and associated products cannot expand.

      Opting to lease capacity from developers like Crusoe instead of constructing all facilities internally is a strategic move. Building data centers at the gigawatt scale requires years and significant investment; by contracting out to developers who handle the construction and operation, Meta can expedite its efforts and mitigate the risks associated with building, albeit at the expense of a margin paid to the developer. With demand surpassing supply, the value of speed has increased over ownership, and Crusoe is offering just that.

      Neither Meta nor Crusoe has provided detailed comments regarding the terms of their arrangements; the figures have been shared by sources close to the agreements rather than an official announcement. What these deals highlight is a trend: a company reliant on AI is procuring electricity by the gigawatt from anyone who can provide it quickly.