Google Takes Crusoe’s Wyoming Site With $17B Build Plan

By Amara Osei

Google will invest $17 billion in an unfinished AI data-center campus in southeastern Wyoming after taking it over from Crusoe, which confirmed the sale on Friday, November 6, 2026; neither company disclosed the acquisition price.

What does the $17 billion cover?

The $17 billion is Google's planned campus investment, not its payment for Crusoe's site. Most of it would fund construction, equipment and power infrastructure over the buildout; any estimate of Crusoe's gain or loss would be guesswork.

Markets Insider called it a $17 billion expansion, describing Google's side rather than a payment to Crusoe.

Reported figures put Crusoe's valuation at about $30.9 billion after a $3.9 billion September 2026 round; the 2025 plan targeted 10 GW, or 10,000 MW.

The $17 billion equals roughly 55% of Crusoe's $30.9 billion valuation; the round equals 12.6%. These ratios show scale, not deal economics.

Takeaway: Google has announced the campus investment, but the acquisition value remains private.

Why did Google take over?

Crusoe is an eight-year-old, Denver-based developer of AI infrastructure and data centers, not AI software. The Hans India called it a “$30.9 billion AI startup”; its work covered site selection, government approvals and power strategy, measured in megawatts and delivery dates.

No company supplied an identified on-record quote. Reports citing people familiar with the matter say Google raised concerns about construction costs and timeline and was a prospective customer that did not commit.

TechCrunch called the project “troubled,” Bloomberg said Crusoe had “pitched” the data center, and Yahoo Finance wrote that “hyperscalers move to own AI capacity.” None settles the motive.

Takeaway: The handover is confirmed; cost, schedule and customer-commitment explanations remain unconfirmed.

What changes in the 10-gigawatt project?

Crusoe and Tallgrass Energy announced the campus in 2025 with 10 GW (10,000 MW) of final power, enough to place it among the region's largest consumers. The original model had a developer assemble land and power, an anchor tenant lease capacity and the developer build; Google instead took over.

Reporting gives no construction start date, so progress remains unmeasured. Google has not said whether it will retain, reduce or phase the 10 GW target.

Google will develop it with Tallgrass Energy, Black Hills Energy, American Electric Power and Related Digital. Reports do not split their costs or roles. Tallgrass provides continuity as Crusoe exits; Black Hills and AEP are utilities, Related Digital develops data centers, and the roster points to power before buildings.

Crusoe's approvals may be worth more than land. The reviewed reports cite no local ballot fight or opposition, although 41 data-center ballot measures elsewhere in the United States have pressed the sector on power, water and costs. Permits take time that money cannot compress.

Takeaway: Google inherited early development work and an energy-partner network, but final scale and cost allocation remain open.

How does site control change AI capacity?

New frontier-model generations raise training and serving demand, and labs must secure capacity years ahead. Chips and servers can be ordered; grid capacity takes years to secure.

Training favors large, connected, power-rich clusters; serving favors proximity and flexible capacity. The 10 GW design points toward training, but reporting names no model or workload.

Leasing leaves construction risk with the developer and gives low-to-moderate control. A takeover transfers risk and raises control. Self-build offers the most control but the slowest start; co-development shares risk; direct power finance controls supply, not buildings. Google combines takeover and co-development.

Anthropic reportedly committed $3.4 billion for 1.7 GW of gas power without buying a data-center shell. Google acquired land and approvals and assembled power partners. Both target deliverable power.

Takeaway: Google traded third-party schedule dependence for direct control and accepted construction risk.

What does the sale mean for Crusoe and the market?

Crusoe completed a reported $3.9 billion round in September 2026 at a valuation of about $30.9 billion, then sold the project two months later. One reading says it turned early work into capital for other builds; another says cost or schedule problems would test a delivery-dependent valuation. With no price or proceeds disclosure, neither can be proven.

The sale does not move markets by itself. Developers face tighter milestones and step-in rights after IREN delayed Microsoft AI deliveries while planning $15 billion in capex and Microsoft tightened oversight of several neocloud providers.

Utilities gain a potential 10 GW load; equipment suppliers may gain a direct buyer that can simplify purchasing. A $329.17 per MW-day auction shows power-price pressure. New contracts, other-campus updates and sale terms would separate the two readings of Crusoe's disposal.

Takeaway: One handover will influence contract terms and power planning, but it does not establish an industry-wide shift.

What are the five signals to watch?

1. More pre-development sales may follow as developers sell sites before full build funding.
2. Utility and permit filings should reveal how Tallgrass, Black Hills, AEP and Related Digital divide work and cost.
3. The sale price may remain private unless a filing requires disclosure.
4. New developer contracts should carry stronger milestones and step-in rights.
5. Model launches should sustain capacity spending, though reporting ties the campus to no model.

The price, motive and final scale remain open. A filing, executive interview or joint statement could resolve them; until then, confirmed facts should stay separate from reported explanations.

Takeaway: Public statements confirm the transfer and investment, but not the price, motive or final capacity.