Crusoe Wyoming Site Sale to Google, November 2026

Crusoe has sold its stalled Wyoming data-centre site to Google, which plans to build a $17 billion campus there, as the neocloud pivots from site developer to cloud-services platform after closing a $3.9 billion Series F at a $30.9 billion valuation.

The sale of Crusoe's stalled Wyoming data-centre site to Google — and Google's announced intention to build a $17 billion campus there — provides the clearest articulation yet of the strategic pivot this neocloud has been executing through 2026. Rather than competing head-to-head with hyperscalers on balance-sheet scale, Crusoe is repositioning as a development-and-services platform: one that originates sites, secures power, pre-develops infrastructure, and transfers the resulting asset to a counterparty with deeper pockets, while retaining cloud-services revenue from its own compute fleet.

Multiple reports published in November described the Wyoming project as one Crusoe could not complete independently. The site had been identified in October reporting as a 2.7-gigawatt campus carrying the internal designation "Project Tembo," and had by that point already been confirmed by Google as its own intended development. When Crusoe made its formal announcement, management framed the transaction as the transfer of a "pre-developed" asset — permits, power agreements, and preliminary infrastructure in place — rather than a distressed disposal. November coverage reported that Google committed to covering 100% of electricity costs and incorporating carbon capture and storage.

The Wyoming handover did not arise in isolation. By September, Google had been named as the anchor customer for Crusoe's development in Armstrong County, Texas — the Goodnight Campus — where Crusoe had filed permits for additional buildings in July. In October, Crusoe and Lancium jointly announced a one-gigawatt AI data-centre campus in Childress, Texas, with construction reported to have commenced in the third quarter of 2026. Taken together, these moves reveal a consistent operational pattern: Crusoe originates and de-risks infrastructure; Google absorbs the resulting capacity. The heavy concentration of that relationship in a single hyperscaler is a structural question the company has not yet fully answered — whether Crusoe is a strategic co-developer or an infrastructure contractor, and what that distinction implies for long-run margin and pricing leverage.

Capital formation has tracked the ambition. The company's Series F, formally announced in September and confirmed across multiple sources in October, closed at approximately $3.9 billion, with post-money valuation reaching roughly $30.9 billion; some earlier dispatches had cited a $3 billion target, suggesting the round was upsized. Nvidia joined Mubadala and QIA as investors, lending the round strategic significance beyond pure financial capital. On the demand side, a reported five-year, $13 billion cloud-compute contract with Jane Street, a $65 million inference deal with Thinking Machines Lab, and a $700 million power-facility financing with NBSF collectively provide revenue-visibility arguments that few infrastructure startups at this valuation can marshal — though the Jane Street figure comes from media reports rather than audited disclosures, and standard source caveats apply.

Beneath the campus layer, Crusoe has been assembling vertical-integration capabilities that could ultimately matter as much as any individual site. The company debuted the Spark Liquid in November — a 3.5-megawatt, factory-prefabricated modular data-centre unit designed to compress deployment timelines. It has backed transformer maker Heron Power's first large-scale California factory and signed a five-gigawatt AI-UPS agreement with ON.energy for hyperscale campuses. In September, Crusoe walked away from a $1.25 billion turbine supply deal with Boom, reportedly redirecting toward alternative power sources; Blue Energy separately filed to build a nuclear reactor at a Texas plant to serve Crusoe's demand, and Aalo's microreactor — in which Crusoe is involved — reportedly achieved criticality, pointing toward a 2027 nuclear-powered AI factory demonstration. IPO preparations were reported as under way as of August. Individually each of these moves is incremental; collectively, they describe a company attempting to control the chokepoints — power, cooling hardware, transformers, grid stability — that constrain the entire buildout.

The outlook is substantively two-sided. The bull case rests on locked-in hyperscaler relationships, a closed multi-billion-dollar round with sovereign and strategic investors, long-dated revenue commitments, and a modular product that could command a deployment-speed premium in a supply-constrained market. The bear case is equally grounded: the Wyoming episode illustrates the execution risk embedded in greenfield development; the portfolio is concentrated in Texas and Wyoming, compressing geographic diversification; and the developer-to-hyperscaler model, while capital-efficient in concept, may erode long-run economics if Crusoe absorbs origination risk while counterparties capture the stabilised yield. Crusoe's chief executive noted in October that the CoreWeave selloff was "overblown" — a defence of a rival that was simultaneously an implicit acknowledgment of sector-wide valuation scrutiny. Three signals will be most informative in the period ahead: whether and at what multiple Crusoe files for an IPO relative to its $30.9 billion Series F mark; how quickly the Armstrong County and Childress campuses reach operational scale; and whether the Wyoming-style development-and-transfer model proves repeatable and margin-accretive or resolves as a one-time clearing of a troubled asset.