Crusoe Drops Its Boom Turbine Plan, Upending a $1.25 Billion AI Power Bet

Crusoe has dropped its Boom turbine plan from its near-term roadmap, reversing a proposed $1.25 billion commitment announced less than a year ago. The planned purchase covered 29 Superpower gas turbines with 1.21 gigawatts of combined generating capacity.

Boom Supersonic CEO Blake Scholl told TechCrunch that the stationary power plants were no longer part of Crusoe’s near-term plans. That wording leaves several contractual questions unanswered. However, it removes the launch customer from the immediate deployment schedule that supported Boom’s original commercial pitch.

The reversal matters beyond one supplier agreement. Boom presented the Crusoe order as proof that technology developed for a supersonic aircraft could enter the demanding data center power market. Crusoe now has other generation agreements involving experienced manufacturers, including GE Vernova and Bergen Engines.

The dispute is therefore not about whether AI data centers need electricity. Crusoe’s development pipeline makes that need clear. The real contest is between a promising, unproven turbine and equipment that operators can evaluate using established manufacturing and service histories.

What Changed in the Crusoe Boom Turbine Plan

A proposed anchor order has disappeared from Crusoe’s near-term deployment strategy, although neither company has publicly explained every contractual consequence.

Boom introduced its Superpower turbine in December 2025 with Crusoe as the launch customer. According to Boom’s original Superpower announcement, Crusoe had ordered 29 units representing 1.21 gigawatts of capacity.

Each Superpower unit was designed to produce 42 megawatts. Boom said the complete system would fit within a shipping-container-scale package and maintain full output in temperatures above 110 degrees Fahrenheit.

The company also promoted waterless operation as an advantage for campuses in hot or water-constrained areas. Those specifications remain company claims until operating hardware produces independently reviewed performance data.

Boom described the order as the foundation of a backlog exceeding $1.25 billion. It planned to deliver the first units during 2027 before expanding annual production through the end of the decade.

That customer relationship served several purposes. It gave Boom a commercial application for technology related to its Symphony aircraft engine. It also gave the company a potential revenue path while it developed the Overture supersonic airliner.

Crusoe, meanwhile, appeared to gain access to capacity within a market facing long turbine lead times. The data center developer has repeatedly emphasized speed to power, meaning the time required to energize computing equipment after selecting a site.

Scholl’s new statement changes that arrangement. The Boom turbines are no longer in Crusoe’s near-term plans, according to the reported reversal.

The available reporting does not fully establish whether the agreement was terminated, deferred, restructured, or replaced by another commercial arrangement. Neither company has publicly detailed cancellation payments, deposits, delivery obligations, or other contract terms.

That distinction matters. A canceled binding purchase and a removed planning assumption can have different financial consequences. Both outcomes still weaken Boom’s original claim that a major customer would support its first commercial production ramp.

Crusoe had publicly promoted the relationship after the December announcement. The company described itself as Boom’s launch customer and highlighted the size of the proposed purchase.

The change is therefore more than the expiration of an informal conversation. A publicly announced customer has moved the technology outside its immediate infrastructure plan before the first Superpower unit entered commercial service.

Boom can continue developing Superpower and pursuing other buyers. Yet it must now prove that demand extends beyond the original anchor customer.

For Crusoe, the decision shows that an energy-first strategy does not require loyalty to one generation technology. It requires selecting equipment that fits a specific campus, customer, delivery window, and regulatory environment.

Why AI Data Center Power Buyers Keep Changing Course

Data center developers face urgent electricity demand, but urgency does not remove construction, reliability, permitting, or financing constraints.

AI campuses require far more than racks of accelerators. They need substations, switchgear, cooling systems, backup equipment, transmission access, and enough continuous generation to support dense computing workloads.

Utilities cannot always provide that capacity on a developer’s preferred schedule. Grid interconnection studies and transmission work can take years, particularly for campuses measured in hundreds of megawatts.

That delay has encouraged a bring-your-own-power model. Under this approach, developers assemble generation, storage, and grid connections around the data center instead of waiting for a single utility solution.

Crusoe has made that strategy central to its identity. The company says its power pipeline reached 45 gigawatts during 2025. It also says the first buildings at its 1.2-gigawatt Abilene campus were energized one year after groundbreaking.

Those figures show why Crusoe evaluates multiple power routes at once. A developer managing several large campuses cannot assume that one turbine model, reactor design, battery system, or utility agreement will fit every project.

Natural gas turbines offer dispatchable generation. Operators can run them when computing loads require power, unlike weather-dependent generation that changes with sunlight or wind.

Aeroderivative turbines adapt aviation engine designs for stationary generation. Their relatively compact packages and faster installation can suit campuses that need large blocks of electricity without a conventional power station’s footprint.

However, a turbine package does not eliminate project work. A developer still needs gas delivery, electrical connections, emissions controls, permits, noise management, maintenance systems, and trained operators.

Boom’s design added another layer of execution risk. Superpower was based on technology being developed alongside the Symphony engine rather than a mature commercial fleet with years of operating data.

The proposed benefits were attractive. Boom said the turbine would retain its rated output in high temperatures, avoid water use, and deliver substantial capacity from a compact package.

The timing was less certain. The first deliveries were planned for 2027, leaving Crusoe dependent on Boom’s ability to complete engineering, testing, manufacturing, and site integration.

Established vendors also face supply constraints. Yet their customers can review operating fleets, service organizations, spare-parts networks, maintenance schedules, and performance histories.

That difference becomes important when a turbine supports costly computing equipment. A generation failure does not only reduce electricity sales. It can strand servers, interrupt customer workloads, and undermine contractual availability commitments.

Crusoe’s recent expansion increases those consequences. The company announced a $3.9 billion financing round in September 2026 at a $30.9 billion post-money valuation. It also reported more than $140 billion in total contracted value across its infrastructure platform.

Those are company-reported figures, not guarantees of future revenue. They still illustrate the scale of the commitments surrounding Crusoe’s construction program.

The larger its customer obligations become, the harder it is to treat power generation as an isolated equipment experiment. Delivery confidence, service coverage, and project-specific risk become central purchasing criteria.

That pressure explains why an urgent market can produce rapid reversals. Demand encourages developers to explore new suppliers. Execution risk pushes them back toward equipment with clearer delivery and operating evidence.

Proven Turbines Now Have the Advantage

Crusoe’s broader procurement record suggests that Boom was one option within a diversified power strategy, not the foundation of every planned campus.

Crusoe had already ordered 29 LM2500XPRESS turbine packages from GE Vernova before announcing the Boom arrangement. Ten units were ordered in December 2024, followed by another 19 booked in June 2025.

The combined GE Vernova order was expected to provide nearly one gigawatt of electricity. Each LM2500XPRESS package can produce approximately 35 megawatts, according to the companies’ turbine agreement.

GE Vernova says each package arrives 95 percent factory-assembled. It also supports a five-minute start without relying on an operating grid connection.

The units include selective catalytic reduction, an emissions-control system that reduces nitrogen oxides in exhaust. GE Vernova says the configuration produces substantially lower emissions than traditional gas or diesel reciprocating engines.

Those claims still depend on site conditions and operating practices. However, the underlying LM2500 engine family has an established service history that Boom’s Superpower turbine does not yet possess.

Crusoe also announced an approximately 750-megawatt agreement with Bergen Engines in June 2026. That agreement further reduced the likelihood that one supplier would dominate the company’s power portfolio.

At other sites, Crusoe is pairing generation with energy storage. Its Nevada deployment combines a 12-megawatt solar installation with 63 megawatt-hours of repurposed electric-vehicle batteries supplied through Redwood Materials.

Storage cannot replace continuous generation indefinitely. It can still smooth short interruptions, manage changing loads, and reduce dependence on immediate grid supply.

Crusoe has also pursued a gas-to-nuclear project with Blue Energy in Port of Victoria, Texas. The companies plan to use gas generation as an initial bridge before transitioning to nuclear power.

Their Texas energy plan targets initial power as early as 2028 and expected nuclear generation during 2031. Those dates remain forward-looking development targets.

The range of agreements reveals Crusoe’s working model. The company is assembling a portfolio that includes established gas turbines, batteries, solar generation, grid connections, and prospective nuclear projects.

Boom’s Superpower turbine would have joined that portfolio as a higher-risk, potentially high-value option. It promised more output per unit than the LM2500XPRESS packages and specific benefits in hot conditions.

Once removed from the near-term roadmap, it no longer addresses Crusoe’s immediate construction schedule. Existing agreements can advance independently while Boom completes its engineering and validation work.

This is the main opponent in the story: promised performance versus deployable, proven capacity.

GE Vernova does not need to show that its equipment originated from an ambitious new engine program. It needs to manufacture, deliver, commission, and support systems derived from an established fleet.

Boom faces a different task. It must simultaneously validate a new engine core, package it for stationary generation, create a production system, and convince infrastructure buyers that service will remain available.

The company says Superpower and Symphony share technology, allowing operating turbines to generate reliability data for the aircraft engine. That overlap can improve engineering feedback.

It can also concentrate risk. A technical delay affecting the common engine architecture could influence both the stationary turbine program and the aircraft schedule.

Crusoe’s change does not prove that the Superpower design is unsound. No public evidence establishes that conclusion. It demonstrates that projected specifications alone were not enough to preserve the original near-term customer plan.

The Reversal Exposes Boom’s Financing Logic

The lost launch plan matters because Superpower was supposed to support more than data centers. It was also designed to help finance and validate Boom’s aircraft ambitions.

Boom raised $300 million when it announced Superpower and the Crusoe relationship. Darsana Capital Partners led the round, with participation from several venture and investment firms.

The company said the financing would fund Symphony development. It also said revenue from Superpower would help finance certification and delivery of the Overture passenger aircraft.

Scholl described the stationary turbine as an accelerator for both artificial intelligence and supersonic aviation. His argument connected two capital-intensive markets through a common engine core.

Data center deployments could generate far more operating hours than a flight-test program. Each commercial turbine could provide information about durability, manufacturing, maintenance, and performance.

Boom also expected turbine sales to create cash flow before Overture carried passengers. That offered investors a commercial narrative beyond waiting for a new airliner to complete development and certification.

The Crusoe order made that story concrete. Twenty-nine planned units represented a customer, a production target, and a visible bridge between engineering work and commercial revenue.

Without Crusoe in the near-term plan, Boom must replace more than nominal backlog. It needs customers willing to accept the risks surrounding early production and commissioning.

That challenge is common in complex hardware. A manufacturer needs meaningful orders to justify production capacity. Buyers often want completed hardware and operating evidence before accepting delivery risk.

An anchor customer can break that cycle by committing early. Losing one can return the supplier to a difficult position where production and demand must advance together.

Boom still says it plans to manufacture Superpower in Denver and expand annual output substantially by 2030. The company’s Superpower specifications continue to advertise 42 megawatts of power, waterless operation, and strong hot-weather performance.

Those specifications do not establish manufacturing yield, maintenance cost, availability, fuel efficiency across real duty cycles, or emissions performance at a permitted site.

A complete validation program must address more than whether the core can run. Customers need evidence about starts, shutdowns, load changes, component life, inspection intervals, controls, and integration with generators.

Baker Hughes announced an order for 31 generators designed to support Boom’s 1.21-gigawatt commitment. Generator deliveries were scheduled across multiple years.

The removal of Crusoe from the immediate plan raises questions about that supply chain. Boom has not publicly explained whether those generators will support other buyers, test units, inventory, or a revised schedule.

The pressure does not necessarily threaten the entire company. Boom can redirect units, recruit another customer, or continue development at a slower commercial pace.

However, its earlier financing logic depended on reinforcing loops. Turbine orders supported manufacturing, manufacturing supported engine validation, and turbine profits supported the aircraft program.

A delayed commercial ramp weakens each connection. It can increase the amount of capital or time needed before the strategy becomes self-supporting.

Boom’s investors must therefore watch customer replacement more closely than marketing interest. Expressions of demand do not carry the same value as deposits, binding schedules, and disclosed project sites.

They must also watch whether a new customer accepts the same configuration and delivery targets. A heavily redesigned product or extended timetable would make the original Crusoe plan a weaker indicator of market readiness.

What the Boom Superpower Claims Still Need to Prove

The central uncertainty is not whether Boom can produce a working turbine, but whether it can deliver a dependable fleet on commercial data center schedules.

Boom says Superpower can maintain full rated output at ambient temperatures above 110 degrees Fahrenheit. That feature would address a real limitation of many turbines, whose output declines as inlet air becomes hotter.

The company also says its system does not require water. That could reduce competition for local water supplies and simplify certain site designs.

Yet waterless turbine operation does not make an entire data center waterless. Server cooling, heat rejection, landscaping, construction, and other site systems can still consume water.

Likewise, a compact turbine does not guarantee a compact power plant. Developers need fuel treatment, exhaust systems, emissions controls, electrical equipment, safety clearances, maintenance access, and noise mitigation.

Permitting remains another major variable. Gas turbines emit carbon dioxide and regulated pollutants, even when modern controls reduce specific emissions.

Local residents and environmental organizations have challenged turbine deployments at other AI data centers. Noise, air quality, transparency, and cumulative emissions can all affect approval timelines.

Crusoe has acknowledged this tension in its wider energy strategy. The company supports natural gas as a near-term supply source while pursuing batteries, renewable generation, and nuclear power.

That mixed approach recognizes an uncomfortable tradeoff. AI customers want rapid capacity, but communities and regulators still evaluate the environmental cost of producing electricity onsite.

The Superpower design also needs independent operating evidence. Boom’s public materials describe intended capabilities, not performance from a commercial fleet.

A preproduction test can validate core engineering. It cannot immediately establish fleet availability over thousands of operating hours or show how quickly technicians can complete field repairs.

Manufacturing scale adds another uncertainty. Boom originally planned to increase Superpower production from early deliveries to multiple gigawatts per year.

That ramp would require qualified suppliers, repeatable machining, quality controls, generator integration, transportation systems, commissioning teams, and spare-parts capacity.

A single technically successful unit would be an important milestone. It would not prove that Boom can manufacture dozens of identical systems on customer deadlines.

The shared architecture with Symphony creates both an opportunity and a constraint. Stationary operation can produce useful engine data, but data center requirements differ from flight requirements.

A stationary turbine prioritizes continuous output, service access, electrical integration, and operating economics. An aircraft engine must meet demanding weight, safety, noise, and certification requirements.

Common components can spread development costs. They do not erase the need to validate each finished system for its intended use.

The original customer announcement also left questions about contract structure. Boom characterized the Crusoe commitment as an order and included it in its backlog.

Public reporting has not disclosed whether the deal contained performance milestones, termination rights, refundable deposits, or conditions linked to testing.

Those details determine how much financial protection either party retained. Until the companies disclose them, observers should avoid treating the full headline amount as recognized revenue or a confirmed loss.

The cautious conclusion is narrower. Crusoe no longer expects Boom turbines to serve its near-term needs, and Boom has lost the immediate deployment path attached to its launch announcement.

That is significant without assuming technical failure, contractual damages, or the end of Superpower.

Three Signals Will Show What Happens Next

Boom’s first full-scale test, a replacement customer, and Crusoe’s actual power deployments will determine whether this was a delay or a deeper commercial setback.

The first signal is a fully integrated Superpower unit operating under a documented test program. Boom needs to show sustained output, heat performance, efficiency, emissions behavior, and maintenance requirements.

Independent or customer-observed results would carry more weight than updated specifications. A successful test would strengthen Boom’s claim that the technology remains commercially relevant without Crusoe.

A delayed test, major redesign, or limited performance disclosure would weaken that claim. It would also make aggressive production targets harder to defend.

The second signal is a new anchor customer tied to a real site. Boom has described substantial interest in data center power, but interest must become an executable order.

A credible replacement should include unit count, capacity, delivery timing, project location, and clear responsibility for generators, emissions equipment, fuel, and electrical integration.

If another experienced developer accepts those terms, Crusoe’s exit will look more like customer-specific reprioritization. If no replacement appears, the reversal will look like a broader warning about technology or delivery risk.

The third signal is how Crusoe energizes its next campuses. The company has agreements spanning GE Vernova turbines, Bergen Engines equipment, storage projects, utility power, and prospective nuclear generation.

Actual commissioning dates will reveal which technologies solve the speed-to-power problem rather than merely occupying announcements and development plans.

Crusoe’s BYO power strategy says the company wants control over generation, infrastructure, and computing delivery. The Boom reversal tests how flexible that strategy becomes when a supplier’s timetable no longer matches immediate needs.

For developers and enterprise AI buyers, the lesson is not to dismiss emerging turbine companies. New entrants can expand constrained supply and introduce useful engineering choices.

The lesson is to separate nameplate capacity from dependable capacity. A proposed gigawatt does not run models until equipment is manufactured, permitted, installed, fueled, tested, and supported.

Boom now has to establish that chain without the customer that originally validated its data center pivot. Crusoe must show that its alternative portfolio can energize campuses on the schedules promised to customers.

Watch the hardware, the replacement contracts, and the commissioned megawatts. Those signals will determine whether the Crusoe Boom turbine plan was an early detour or a warning about financing AI infrastructure with unproven power systems.