In the coming years, a large artificial intelligence (AI) data centre in Wyoming is expected to consume more electricity than all of the state’s households combined. The scale of the project has surprised many observers, not only because of its size but also because the identity of its future occupant remains a mystery. This development raises questions about energy usage, infrastructure demands, and the potential impact on the local economy and environment.
The proposed facility is a collaboration between Tallgrass, an energy infrastructure company, and Crusoe, a developer of AI-focused data centres. It is planned to begin operations at 1.8 gigawatts and could expand to an immense 10 gigawatts. To put that into perspective, this would be more than five times the electricity currently used by Wyoming’s human population. The least populated state in the United States, Wyoming, has roughly 590,000 residents, meaning such an energy draw would dwarf household consumption.
Wyoming is no stranger to large-scale computing facilities. Microsoft has had a presence in the state since 2012, and Meta Platforms, the parent company of Facebook, is nearing completion of an $800 million data centre near Cheyenne. But the new project far exceeds anything seen before in terms of energy demand. It reflects the growing power requirements of AI infrastructure and the trend towards building massive, purpose-built sites to support advanced computing.
Given the extraordinary electricity needs, the developers have confirmed that the facility will not draw its power directly from the public grid. Instead, they plan to construct dedicated energy sources, combining natural gas and renewable energy generation. This approach is partly intended to prevent the project from straining existing infrastructure and disrupting local supply.
However, such a power model still has wide-reaching implications. Building new natural gas plants and renewable installations specifically for the centre will require substantial investment and may alter regional energy markets. While Wyoming is a top energy-producing state—generating about 12 times more energy than it consumes and exporting most of it—projects of this scale can shift priorities in ways that affect local costs and availability.
A single gigawatt can supply electricity to around one million homes, depending on usage patterns. At its maximum planned capacity of 10 gigawatts, this Wyoming facility would have an energy footprint comparable to that of several mid-sized countries. It would be one of the largest such sites in the world, designed for continuous, high-load operation to meet the demands of AI training and deployment.
One of the most curious aspects of the project is that the future occupant of the data centre has not been publicly named. This has led to speculation, with many pointing towards OpenAI, the developer of ChatGPT. Crusoe has already partnered with OpenAI on a separate project in Texas—described by the company as the “largest data centre” in the world—and industry observers note that OpenAI has committed to securing multiple gigawatts of data centre capacity for its AI operations.
When questioned, Crusoe representatives declined to confirm or deny OpenAI’s involvement, saying only that they were “not at a stage to announce our tenant.” This secrecy has fuelled discussions about whether the project could be part of OpenAI’s so-called “Stargate” initiative, a long-term effort to build an extensive network of AI facilities.
The Texas site operated by Crusoe and OpenAI reportedly draws about a gigawatt of power. If the Wyoming project is linked, it would represent a major step up in scale, possibly designed to host large-scale AI models and supercomputing hardware far beyond current standards.
AI research and deployment demand vast computing resources. At the heart of such facilities are advanced processors—often GPUs or specialised AI chips—that perform complex calculations required for machine learning. These chips consume large amounts of electricity and generate considerable heat, requiring extensive cooling systems.
Dense racks of high-performance servers, often stacked in warehouse-like spaces, must run around the clock for tasks such as training large language models. Cooling systems, whether air-based or liquid-based, can consume a substantial share of a facility’s total power. In this context, the choice of Wyoming, with its relatively cool climate, makes practical sense, as cooler air can help reduce cooling costs and improve operational efficiency.
The planned site south of Cheyenne, near the Colorado border, benefits from easy access to energy infrastructure and transportation links. Its rural setting also offers space for expansion and reduced risk of disruption to residential areas.
Wyoming’s political leaders have expressed strong support for the project. Governor Mark Gordon called it “exciting news” for the state’s natural gas producers, viewing the facility as a major opportunity for the local energy sector. Cheyenne’s mayor, Patrick Collins, described it as a “game changer,” highlighting the city’s growing status as a hub for large-scale computing operations.
The facility is expected to create construction jobs during its build phase, as well as ongoing employment for technicians, engineers, and support staff once operational. It could also attract related businesses, such as hardware suppliers and maintenance providers, to the area.
However, not all reactions have been entirely positive. While the facility’s dedicated energy supply may reduce direct competition for public electricity, there are concerns that indirect effects could still impact residents. For example, shifts in energy pricing or supply chain pressures could lead to higher utility bills in surrounding communities. Furthermore, a project of this magnitude may influence Wyoming’s energy export balance, especially if similar facilities follow.
The developers have stated that the facility will use a mix of natural gas and renewable energy. While this may help reduce the project’s overall carbon footprint compared to fossil fuel-only power, it does not eliminate environmental concerns. Natural gas plants emit greenhouse gases, and large-scale renewable installations can require substantial land use and materials.
Experts note that AI data centres can reduce their environmental impact by prioritising renewable energy and implementing highly efficient cooling and power systems. However, the sheer scale of the Wyoming project means that even with high efficiency, its total energy use and associated emissions will be considerable.
This raises broader questions about the environmental cost of large AI models and whether the benefits they bring—such as advances in medicine, engineering, and science—justify their resource demands. It also prompts debate over whether such facilities should be located in areas with abundant renewable resources, such as solar-rich deserts or regions with steady wind supply, to reduce dependence on fossil fuels.
The Wyoming project is part of a wider trend in AI infrastructure. Technology companies are racing to secure large-scale computing capacity to meet rising demand for AI services. This is driven by the increasing size and complexity of AI models, as well as the growing appetite for AI applications in industry, research, and consumer products.
OpenAI, Google, Microsoft, and other firms have all announced substantial investments in new data centres, often in partnership with energy providers. These facilities are becoming larger and more specialised, with designs tailored to AI workloads rather than general-purpose cloud computing.
Wyoming’s advantages—cool climate, ample energy supply, and pro-business policies—make it an attractive site for such investments. If the current project succeeds, it may encourage further development, potentially reshaping the state’s economic profile and its role in the global technology sector.
As of now, the Wyoming project has not yet broken ground, but local officials suggest that construction could start soon. State and local regulators will need to review and approve the plans, including environmental assessments and infrastructure proposals. Given the level of political support, these approvals are expected to move forward without major delays.
If built to its full 10-gigawatt capacity, the facility would set new records for data centre size and power consumption. It would also stand as a symbol of the growing energy appetite of AI systems—a reminder that the digital services we use daily are powered by vast, physical infrastructure with substantial environmental and economic footprints.
