An order that doubled as a funding story
When the deal was announced last December, it did not mean the same thing to both companies. For Crusoe, 29 turbines and 1.21 gigawatts were one self-generation option within its "AI factories from the power up" pitch. For Boom, a company that has yet to deliver an airliner, it was **its first source of commercial revenue**: Superpower uses the engine core developed for a supersonic jet, so selling power-plant turbines both spreads engine development costs and shows the engine can be built at volume. Scholl said at the time that with this financing and the first Superpower order, Boom was funded to deliver both its engine and its airliner. **So for Crusoe the cancellation is a procurement adjustment; for Boom it removes the most solid brick in its funding story.**
The deleted line says more than what remained
According to TechCrunch, Scholl's post first said that turbines are no longer part of Crusoe's "near term primary power mix" at Abilene and elsewhere, so a launch partnership didn't make sense. That line was later deleted. Crusoe's own framing is different: it stays flexible, choosing the right energy solution for each site, and turbines remain an option alongside wind, solar, batteries and the grid. Both can be true, but the emphasis differs. **The first says turbines are no longer needed in the primary power mix; the second says turbines are still one of the options.** With the first line gone, only the second framing remains on the record. TechCrunch's site details help here: the 1.2-gigawatt Abilene campus built for Oracle and OpenAI runs mainly on the grid, with gas turbines only as backup, while the 900-megawatt campus built for Microsoft is powered by on-site gas turbines. In other words, Crusoe has not given up on gas generation; it has given up on a supplier that has never produced at volume.
Read it alongside two other stories this week
When we covered Crusoe's funding on September 20, the point was that it reports more than $140 billion in total contract value and over 6 gigawatts of gross contracted capacity, against 1 gigawatt delivered, and that the real constraints are power and construction time. On September 25, Oracle issued a force majeure notice on Project Jupiter, held up by a delayed natural gas pipeline. This is the other side of the same problem: **when power is the bottleneck, buyers favor options that can deliver on time and have an operating record over the most imaginative one.** A turbine adapted from a supersonic jet engine, with first deliveries planned for 2027, being the first thing cut is not surprising.
What this means for readers
For anyone tracking AI infrastructure, the lesson is that a new power equipment supplier's announced "first order" should be treated as intent until delivery, especially when the order is also carrying the supplier's funding story. As for Boom, Scholl says it will still deliver about 250 megawatts to other sites next year and is targeting 1 gigawatt in 2028, but he did not name the new customers. Until the customer list is public, those two figures are targets. Limits of what is known: neither side has disclosed the terms of the cancellation, including any penalties or payments already made; whether Crusoe has shifted turbine purchases to other suppliers could not be verified from first-hand sources.
via: TechCrunch, TechCrunch on the original December 2025 deal, Data Center Dynamics on the original order