The race for artificial intelligence computing capacity intensified this week with news that reveals both the velocity and the fractures of the independent data center development model: Google has acquired a data center campus project in Wyoming that Crusoe had developed from scratch, committing to invest US$ 17 billion in construction. The announcement, made on Thursday the 5th, was confirmed by Crusoe on Friday the 6th in a detailed press release. It makes this deal a real-time case study in the tension between the ambitious vision of independent AI infrastructure developers and the reality of who actually controls capital and risk.
Google did not disclose the amount paid for the acquisition of the Wyoming site and development rights — but the total investment announced for campus construction far exceeds any upfront payment. According to Google's own announcement, the project represents Google's first data center in the state, on land in southeastern Wyoming near Cheyenne. The campus was planned to eventually grow to 10 gigawatts of power capacity, a number that alone helps understand the scale of the undertaking: 10 gigawatts is enough to power a city of millions.
But what makes the story more interesting than the size is how it happened. Crusoe, a Denver-based company with eight years of existence and recently valued at US$ 30.9 billion, had announced the project the previous year in partnership with Tallgrass Energy, an infrastructure company backed by Blackstone. Crusoe was the first to select the site, obtain government approvals, and develop a hybrid energy strategy — combining grid power, natural gas turbines, and a carbon capture plan — all before securing a customer. Google was precisely that missing customer. Instead of signing the leasing agreement, the search giant bought the project.
According to TechCrunch, Google had raised concerns about the cost and construction timeline under Crusoe's management. The company had unsuccessfully tried to attract Google as a customer for the project — meaning the direction of the commercial relationship had flipped: rather than Crusoe selling compute capacity to Google, Google itself took over the development. This raises questions about the business model of Crusoe and other independent data center developers that emerged to serve AI demand.
Crusoe was founded by Chase Lochmiller and Cully Cavness, starting as a cryptocurrency mining business powered by natural gas that would otherwise have been flared. Under Lochmiller's leadership, the company repositioned itself as an "AI factory company," focused on activating gigawatts of energy for artificial intelligence. Since then, Crusoe has signed contracts for more than 6 gigawatts and has 40 gigawatts in its pipeline — numbers that sound ambitious, but become clearer when you understand that the money comes from future contracts, not current revenue. In September, the company raised US$ 3.9 billion in a funding round, adding members including JB Straubel, founder of Redwood Materials and former Tesla CEO, to the board, and invested in the energy storage business of which Crusoe itself was the first customer.
Crusoe's largest projects serve some of the biggest names in the industry. Its campus in Abilene, Texas, is being built in phases for Oracle and OpenAI, and the company is also constructing a 900-megawatt campus next door for Microsoft, with the first building expected to come online in mid-2027. Meta is a customer, and Bloomberg reported that trading firm Jane Street signed a US$ 13 billion, five-year cloud deal with Crusoe. In Wyoming, however, the model hit its limit.
What the Wyoming case reveals is the structural tension between independent developers and the hyperscalers who hire them. Crusoe and similar companies do the heavy lifting of origination, regulatory approval, energy engineering, and site de-risking — basically turning a vacant field into a place where a data center can be built. In return, they sell space or compute capacity to companies like Google, Oracle, OpenAI, and Meta. But when the construction schedule extends and costs exceed expectations, the hyperscaler client can simply buy the project and take control. The independent developer keeps the pre-construction value — which is substantial — but loses the long-term revenue that was the model's promise.
This is not new in absolute terms — data center development has always involved this asymmetric relationship. But the AI boom has brought unprecedented scale: 10 gigawatts, US$ 17 billion, multiple simultaneous campuses. Numbers like these require not just capital, but direct control over schedule and costs. For Google, taking over development in Wyoming may be the most pragmatic decision possible: instead of trusting that Crusoe will deliver on time, Google simply takes control. For Crusoe, it is an operational setback, but not a failure — the company is still building a data center for Google in Texas and continues to have a 40-gigawatt pipeline.
The irony is that while Google invests US$ 17 billion in Wyoming, the very infrastructure that sustains this investment depends on technologies Crusoe had developed: the hybrid energy strategy, the environmental approvals, the integration with Tallgrass Energy and its regional CO2 sequestration hub. In other words, Google is buying the technical intelligence that Crusoe built — and then executing what was planned. It is not the end of a business model, but it is a lesson in who actually holds power in the AI infrastructure value chain.
What remains to be seen is whether, with AI compute demand continuing to grow exponentially, how many other Wyoming-like projects will take shape? Or will other independent developers simply accept that their role is pre-development — doing the heavy lifting and then selling the site to those who truly control capital? The answer may define the architecture of the next decade of artificial intelligence infrastructure.
Sources: Crusoe, TechCrunch, Google Blog
✓ Independent sources cross-checked and verified before publishing
