Google Takes Over Wyoming Data Center Project From Crusoe
Tech giant Google has taken direct control of an unfinished data center project in southeastern Wyoming, purchasing the site from Denver-based developer Crusoe. The transaction follows months of mounting pressure regarding construction timelines and escalating costs at the facility. Crusoe originally unveiled the site last year in partnership with Blackstone-backed Tallgrass, carrying ambitious plans to scale the campus up to 10 gigawatts of power capacity.
Google initially entered the picture as a prospective anchor customer for the campus. However, as development hurdles mounted under Crusoe’s watch, the search giant stepped in to acquire the property outright. While specific financial terms of the asset purchase remain private, Google simultaneously announced a staggering $17 billion commitment to develop the sprawling data center site. Despite this ownership shift in Wyoming, Crusoe and Google maintain an active commercial partnership, continuing joint work on a separate facility near Amarillo, Texas.
📌 Key Takeaways
- Crusoe sold its unfinished southeastern Wyoming data center project directly to Google following construction delays and cost overruns.
- Google announced a massive $17 billion investment plan to build out the acquired Wyoming campus.
- Independent developers face severe hurdles procuring heavy equipment, securing adequate power, and managing local opposition for gigawatt-scale AI infrastructure.
- Crusoe continues to build separate major AI data center campuses for major tech clients including Oracle, OpenAI, Microsoft, and Meta.
The Realities of Scaling Gigawatt AI Campuses
Building modern data centers tailored for artificial intelligence requires navigating engineering hurdles that dwarf traditional cloud computing facilities. Standard cloud servers typically demand density levels of 5 to 10 kilowatts per rack. In stark contrast, modern AI training clusters utilizing advanced graphics processing units require anywhere from 40 to over 100 kilowatts per rack. This extreme thermal and electrical concentration completely redefines facility architecture.
At a mechanical level, developers must integrate advanced liquid cooling distribution systems, high-voltage step-down substations, and uninterrupted power supply arrays capable of sustaining microsecond load swings without grid degradation. Power distribution units must handle hundreds of megawatts locally, necessitating custom substation architecture designed in tandem with regional utility providers. Furthermore, water consumption for evaporative cooling or closed-loop liquid delivery demands secure local water rights, adding a complex regulatory layer to site selection.
Power Bottlenecks and Supply Chain Pressures
Constructing modern infrastructure for artificial intelligence requires solving complex logistical equations involving electrical grid capacity, specialized cooling hardware, and critical electrical components. Securing long-term power agreements often takes years, putting independent developers at a disadvantage compared to hyperscalers with bottomless balance sheets.
Crusoe’s business model spans multiple revenue streams, including leasing infrastructure to clients who supply their own GPUs, renting out proprietary hardware, and selling AI inference compute power. Yet, executing multi-gigawatt construction schedules strains supply chains for transformers, substations, and high-voltage transmission lines. When local community friction compounds these delays, developers frequently find themselves forced to restructure or offload heavy-lift assets.
Shifting Dynamics in Hyperscale Development
The transaction illustrates a shifting dynamic where hyperscalers are increasingly willing to absorb direct control of troubled real estate assets rather than rely solely on independent build-to-suit agreements. As cloud providers race to secure real estate capable of supporting dense clusters of AI accelerators, project execution speed outweighs reliance on third-party intermediaries.
When compared to industry benchmarks, Google’s $17 billion commitment to the Wyoming site ranks among the largest single-site capital expenditures in the tech sector, rivaling massive regional investments announced by Microsoft, Amazon Web Services, and Oracle. While traditional data center buildouts historically relied on regional real estate developers financing construction via commercial debt, the generative AI boom has forced tech giants to internalize capital risk.
Despite relinquishing the Wyoming asset, Crusoe’s pipeline remains exceptionally robust. The company is actively constructing a massive campus in Abilene, Texas, tailored for Oracle and OpenAI, alongside a separate 900-megawatt facility for Microsoft where initial buildings are scheduled to launch in mid-2027. Additionally, Meta utilizes Crusoe’s infrastructure, and trading firm Jane Street reportedly committed to a $13 billion, five-year cloud agreement with the developer.
Frequently Asked Questions
Why did Crusoe sell the Wyoming data center project?
Crusoe sold the unfinished Wyoming project to Google following construction delays, cost overruns, and concerns raised by Google regarding project timelines under Crusoe’s original management.
How much is Google investing in the Wyoming campus?
Google announced plans to invest $17 billion to develop the data center campus after taking over the development from Crusoe.
Does Crusoe still work with Google on other projects?
Yes. Despite the Wyoming asset transfer, Crusoe continues to develop another major data center for Google near Amarillo, Texas.
What technical hurdles do independent data center developers face?
Independent developers often struggle with long lead times for high-voltage transformers, securing multi-gigawatt power purchase agreements with regional utilities, managing high-density liquid cooling requirements, and handling capital expenditure overruns before a facility begins generating revenue.
Sources and further reading
Read the original reporting on TechCrunch.
