Wisconsin faces a dramatic energy crisis driven by artificial intelligence infrastructure, with three major data center projects set to consume more electricity than every home in the state combined. A draft report from Wisconsin's Public Service Commission (PSC) projects that the state's peak energy needs could surge more than 40% by 2032, with hyperscale data centers responsible for over 70% of that increase.

The three projects driving this unprecedented demand are Microsoft's campus in Mount Pleasant, Oracle's facility in Port Washington, and Meta's data center in Beaver Dam. These aren't small computing facilities; they represent the physical infrastructure required to train and run the artificial intelligence models that power everything from ChatGPT to enterprise AI applications. According to a separate report from Clean Wisconsin, just the Port Washington and Mount Pleasant projects alone will need more energy than all residential homes in Wisconsin combined.

Why Is Data Center Power Consumption Exploding?

The global AI data center market is experiencing explosive growth. The market was valued at USD 19.6 billion in 2025 and is projected to reach USD 262.7 billion by 2035, growing at a compound annual rate of 29.6%. This growth is driven by surging demand for generative AI training and inference workloads, which require massive amounts of computing power concentrated in single facilities.

The International Energy Agency projects that global electricity consumption from data centers will more than double to around 945 terawatt-hours (TWh) by 2030, representing just under 3% of total global electricity consumption, with artificial intelligence identified as the most important driver of this increase. In practical terms, this means data centers are becoming some of the largest electricity consumers on the planet, rivaling entire industries.

Hardware spending dominates data center investment, accounting for 72.6% of the market in 2025. Graphics processing units (GPUs), which are specialized chips designed for AI computation, are the primary driver of this spending. NVIDIA, the leading GPU manufacturer, reported record data center revenue of USD 75.2 billion in its first quarter of fiscal 2027, up 92% year over year, demonstrating how intensely companies are investing in AI infrastructure.

How Are Utilities Planning to Meet This Demand?

Wisconsin utilities are pursuing an "all of the above" energy strategy to meet the projected demand surge. According to reporting on the PSC findings, utilities plan to deploy approximately 14.2 gigawatts of new energy generation capacity, with natural gas providing the largest share, followed by solar power. Natural gas plants are being prioritized because they can provide consistent baseload power when renewable sources like wind and solar aren't generating electricity.

However, this approach has sparked significant concern among environmental advocates. Clean Wisconsin and other groups have raised alarm about the reliance on fossil fuel generation at a time when climate change impacts are becoming increasingly visible. Some utilities are exploring nuclear options; for example, We Energies' parent company is working with a Utah-based firm to explore potentially restarting the Kiwanee nuclear plant for new generation capacity, though such projects could take years to develop.

What Factors Are Driving Data Center Site Selection?

The rapid growth of hyperscale data centers, which held 48.3% of the market share in 2025, is reshaping how companies evaluate real estate. Site selection is no longer primarily about real estate costs or proximity to customers. Instead, companies are securing what utilities call "megawatts," along with GPU access, cooling capacity, fiber connectivity, and future expansion rights.

• Power Access: Reliable, abundant electricity supply is now the primary constraint limiting data center expansion, making proximity to power plants and transmission lines critical.
• Cooling Infrastructure: AI servers generate enormous amounts of heat, requiring sophisticated cooling systems and access to water resources for liquid cooling systems.
• Grid Capacity: Utilities must have sufficient transmission capacity to deliver power to the facility without overloading existing infrastructure.
• Regulatory Environment: Local zoning approvals, state utility commission reviews, and environmental assessments can delay or block projects by years.
• Land Availability: Hyperscale facilities require large plots of land, typically hundreds of acres, which limits viable locations.

In Wisconsin, the challenge is becoming acute. Cloverleaf Infrastructure, the early developer of the Port Washington project, has expressed interest in developing an additional data center campus in Northeast Wisconsin that would require roughly one gigawatt of power, but no project is currently moving forward in that region. The question of whether Wisconsin can even approve additional data centers remains uncertain, with communities showing "immense pushback" whenever new projects are proposed.

What Does This Mean for the Broader AI Infrastructure Market?

The Wisconsin situation reflects a global pattern. North America leads the world in AI data center deployment, holding 41% of the global market share in 2025, supported by strong hyperscale cloud infrastructure, high demand for AI model training, advanced semiconductor access, and large-scale investments by technology companies. The U.S. market alone was valued at USD 6.3 billion in 2025 and is projected to expand at a 29.1% compound annual growth rate through 2035.

Cloud-based deployment accounts for 56.9% of the market, with companies preferring to access AI compute and storage through cloud providers rather than building private data centers. This model allows enterprises to avoid massive upfront infrastructure costs while gaining immediate access to GPUs and managed AI platforms. Microsoft's cloud results illustrate this trend; in the third quarter of fiscal 2026, Microsoft Cloud revenue reached USD 54.5 billion, up 29%, while Azure and other cloud services revenue increased 40%.

The capacity constraints visible in Wisconsin are appearing nationwide. In North America's primary data center markets, supply reached a record 8,155 megawatts in the first half of 2025, up 43.4% year over year, while vacancy dropped to a record-low 1.6%. More significantly, 74.3% of under-construction capacity was already preleased, mainly by cloud and AI providers, indicating that demand is outpacing supply.

The convergence of AI compute demand, power availability constraints, and capacity scarcity is reshaping the energy and infrastructure sectors. Companies are no longer simply buying data center space; they are competing for access to megawatts of power, and that competition is intensifying as AI adoption accelerates globally.