Crusoe Ends $1.25 Billion Plan to Power AI Data Centers With Boom Turbines
Crusoe will no longer use Boom Supersonic stationary power plants at its AI data centers. Boom CEO Blake Scholl disclosed on September 25, 2026 that the two companies are no longer moving forward with the turbine launch partnership, and Crusoe confirmed to TechCrunch that it is no longer doing business with Boom. TechCrunch
Crusoe had agreed to spend $1.25 billion on 29 of Boom's 42-megawatt Superpower turbines, with first deliveries set for 2027. Crusoe was to be the first customer for Boom's stationary power business.
Superpower is a natural-gas turbine built for AI data centers, based on Boom's supersonic engine work. Boom Supersonic The design shares about 80% of its parts with Boom's Symphony aircraft engine. That aero-derivative approach, meaning an aircraft engine adapted for ground power, is meant to reuse combustor design, turbomachinery and factory experience from jet propulsion.
Boom introduced the Superpower line on December 9, 2025, alongside $300 million in new funding. That 2025 raise was aimed largely at turning the stationary power business into a commercial product. Supplier Baker Hughes secured a 1.21-gigawatt generator order to support Boom's AI data center solution. Boom Supersonic
For Crusoe, part of the power was linked to large-scale AI capacity. Crusoe is building a group of data centers for OpenAI. Fast Company
Scholl said Boom still plans to deliver about 250MW of Superpower turbines in 2027 to other sites, and is targeting 1GW in 2028. That plan now goes ahead without its anchor project at Crusoe.
The broader context here is the risk of buying primary power for large AI training and inference clusters from a single new source. A 29-unit block at 42 MW each is more than a gigawatt of rated capacity with one operator, which puts delivery, installation, gas supply, emissions permits and maintenance on one turbine design. Using one design simplifies spare parts and daily operations. It also concentrates schedule and performance risk until the turbines build up real operating hours.
In my view, losing a first customer tests Boom's aircraft-to-power idea without settling it. Sharing 80% of parts can shorten development and tooling time, but data center operators pay for uptime, fuel efficiency, service intervals and service contracts they can take to a bank. Those are earned by hours in the field, not by shared part numbers. Worth flagging for planners is the new sequence to watch: about 250MW placed elsewhere in 2027, then a fourfold increase to 1GW in 2028. If Boom places that first capacity, gathers operating data and keeps its generator and supporting equipment supply steady through Baker Hughes and others, it still has a route to sell a repeatable gas turbine package where the grid cannot keep up with AI demand. If that placement slips, other turbine makers will likely fill the gap while Crusoe powers its OpenAI buildout another way.
