The contracts cover facilities in Childress, Texas, and Warrenton, Missouri as hyperscalers race to lock down power capacity through 2030
Meta just locked in roughly 1.6 gigawatts of AI computing capacity through new agreements with Crusoe Energy. The deals, signed around June 18, cover Crusoe facilities in Childress, Texas, and Warrenton, Missouri.
From flared gas to AI factories #
Crusoe has pivoted hard toward what it calls an “energy-first AI factory” model. The company operates GPU-based cloud infrastructure built on NVIDIA and AMD hardware, catering specifically to the kind of large-scale training and inference workloads that companies like Meta are scaling up at a breakneck pace.
The client list tells you everything. Crusoe now counts Google, Microsoft, Oracle, and OpenAI among its customers. Adding Meta to that roster cements Crusoe’s position as one of the go-to infrastructure partners for hyperscalers who need power and compute.
Why hyperscalers are in a power arms race #
Training frontier AI models requires staggering amounts of sustained compute. Each generation of model gets bigger, and the inference workloads that come from deploying those models to billions of users, which is exactly what Meta does with its AI features across Instagram, WhatsApp, and Facebook, consume power around the clock.
Traditional utility infrastructure in the US simply wasn’t built for this. Grid interconnection queues stretch years into the future. That’s why companies like Meta are turning to specialized partners like Crusoe, who have built their entire business around solving the energy-to-compute pipeline. The race to secure capacity through 2030 is real.
What this means for investors #
Crusoe is privately held, so there’s no direct stock play here. The crypto angle, or rather the absence of one, is worth noting. There were no mentions of digital assets or tokens in the Meta-Crusoe transaction. Crusoe’s evolution from a crypto mining operation to a pure-play AI infrastructure company mirrors a larger capital reallocation away from proof-of-work mining operations, which compete for the same scarce resource: cheap, reliable electricity.
The risk to watch is execution. Building out 1.6 gigawatts of capacity on schedule is an enormous engineering and logistics challenge. Grid connections, cooling infrastructure, GPU supply chains, and local permitting all need to align.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our