Crusoe Drops $1.25B Boom Supersonic Turbine Deal, Shifting Its Data Center Power Strategy
The AI data center builder walked away from 29 jet-derived natural gas turbines, its biggest single power bet, days after closing a $3.9 billion raise and retreating from a planned Wyoming campus.
Crusoe, the Denver-based AI data center builder, has canceled a $1.25 billion agreement to buy 29 of Boom Supersonic's 42-megawatt Superpower turbines, ending what had been one of the stranger bets in data center power procurement.
The news broke Friday when Boom CEO Blake Scholl posted on X that the turbines are no longer part of Crusoe's near-term primary power mix. According to TechCrunch, which spoke to Crusoe's PR team, the company confirmed the decision, with spokesperson Andrew Schmitt saying Crusoe intends to stay flexible and choose energy solutions site by site.
The turbines themselves aren't conventional generator sets. Boom adapted the engine technology from its Overture supersonic passenger jet to burn natural gas on the ground. When Crusoe signed on as launch customer in December 2025, the pitch was straightforward: a fast, modular path to gigawatt-scale power without waiting years for utility grid connections. Twenty-nine units at 42 megawatts each adds up to roughly 1.21 gigawatts, which is meaningful headroom for the kind of GPU-dense builds Crusoe is doing for customers including Oracle and Microsoft.
That logic didn't hold. Crusoe's Abilene, Texas campus, where the turbines were originally slated to land with first deliveries in 2027, now uses the grid as its primary power source, with natural gas turbines in a backup role. The adjacent 900-megawatt site Crusoe is building for Microsoft follows a different design, combining on-site generation with dedicated data center buildings, with the first building targeted for mid-2027. Crusoe has also ordered roughly a gigawatt of generation equipment from GE Vernova and holds a contract with Bergen Engines. The Boom deal wasn't the only iron in the fire; it just turned out to be the wrong one for how Crusoe's portfolio evolved.
The timing tells a more specific story. Crusoe closed a $3.9 billion Series F at a $30.9 billion valuation just weeks before pulling out of the Boom agreement, and also stepped back from a large-scale planned campus in Wyoming. Taken together, as reported by Startup Fortune, the moves point to a company that's choosing modular, flexible infrastructure over large fixed commitments made years in advance. That's a real architectural choice with real financial logic behind it: locking in a gigawatt of novel turbine capacity from a company that hasn't yet shipped at volume is a risk profile that looks different at $30 billion than it did at series C.
The harder hit is to Boom. Scholl's post on X tried to soften the announcement by noting that Boom is targeting delivery of about 250 megawatts to other customers next year and aiming for one gigawatt in 2028. But Crusoe was the launch customer. That status matters commercially: it's how you get your first real operational data, your reference architecture, and your next pitch. Without it, Boom's Superpower business is still a product looking for its first deployment at scale. Baker Hughes, which was contracted to supply generator components for the original Crusoe order, is also now in an uncertain position.
The collapse lands the same week Oracle reportedly filed a force majeure notice over power delays at its own Stargate site, which means this isn't an isolated Crusoe story. Power is still the binding constraint on AI infrastructure build-out, and the AI industry's attempts to work around slow grid interconnection with purpose-built or novel generation capacity keep running into the same problem: procurement reality catches up with procurement ambition.
Crusoe's retreat from the Boom deal isn't a vote against on-site generation. It's a vote against betting a specific unproven technology as your primary power source before the architecture is settled. The difference matters for how you read the rest of the market.
Sources cited:
- TechCrunch (https://techcrunch.com/2026/09/25/crusoe-abandons-1-25b-plan-to-use-boom-turbines-at-ai-data-centers/)
- XenoSpectrum (https://xenospectrum.com/en/crusoe-boom-turbine-deal-ended/)
- Startup Fortune (https://startupfortune.com/crusoe-walks-away-from-its-125-billion-jet-turbine-deal-with-boom/)
- Gokhshtein (https://gokhshtein.com/news/2026-09-26-crusoe-scraps-125b-turbine-deal-with-boom-supersonic)
- Crypto Briefing (https://cryptobriefing.com/crusoe-abandons-boom-supersonic-turbine-deal/)
This release was originally distributed via ETL Newswire. Visit TechCrunch for the full story, related releases, and contact information.
Visit TechCrunch →