Crusoe Cancels $1.25 Billion Turbine Deal for Its AI Data Centers

Crusoe, one of the best-funded AI data center operators in the United States, walked back its $1.25 billion deal to buy 29 "Superpower" natural gas turbines from Boom Supersonic. Boom CEO Blake Scholl confirmed the cancellation on September 25, 2026, and TechCrunch, Unite.AI, and Startup Fortune reported the details that same weekend.

What the original deal covered

Crusoe had signed the agreement in December 2025 as the launch customer for Boom Supersonic's Superpower turbine business — yes, the same company known for its Overture supersonic jet — committing $1.25 billion for 29 gas turbines with a combined capacity of 1.21 gigawatts. First deliveries were slated for 2027, meant to power Crusoe's AI data center campus in Abilene, Texas, which today serves workloads for OpenAI and Oracle.

Why Crusoe backed out

According to Scholl, the Superpower turbines "are no longer part of Crusoe's short-term primary power mix in Abilene." The 1.2-gigawatt Abilene campus draws mainly on grid electricity, using gas turbines only as backup — suggesting Crusoe managed to lock in more grid capacity than it anticipated when it signed the original contract, reducing its need for dedicated on-site generation.

The other side: what Boom Supersonic loses

For Boom, the cancellation means losing its first launch customer for the Superpower turbine business, barely nine months after announcing it. Scholl said Boom will still deliver roughly 250 megawatts of Superpower turbines to other sites during 2027, with a goal of reaching 1 gigawatt of installed capacity by 2028 — meaning the company is pressing ahead with the business, just without the anchor contract that this deal provided.

What it means for anyone buying AI compute in the cloud

This kind of reshuffling in the AI power supply chain has a second-order effect on the pricing and availability of compute that eventually reaches a company in Guatemala or Latin America renting cloud GPUs: when a large operator changes its energy strategy, it can mean more compute capacity freeing up sooner than expected at a given site, or delays if the change doesn't go as planned. For teams negotiating mid-term AI compute contracts, it's worth asking your cloud provider directly how power supply is resolved at the specific site where your workload will run, not just how many GPUs it's promising.