Oracle signs a nuclear power partnership to supply its data-center facilities with reliable, carbon-free baseload power.

On October 2, 2026, Oracle and We Energies announced a nuclear power subscription deal marking a significant shift in how hyperscale data centers secure energy. Oracle has committed to subscribing to 10-20% of the output from the Point Beach Nuclear Plant, representing 125-250 megawatts of power for the Lighthouse Campus, a $15 billion AI data center in Port Washington, Wisconsin, co-developed by Oracle, OpenAI, and Vantage Data Centers as part of the Stargate initiative. The facility is designed to house approximately 1 gigawatt of AI capacity within a total electrical footprint of 1.3 gigawatts, with completion targeted for 2028.

The deal's financial structure reveals the escalating costs of the AI supply chain. Point Beach, Wisconsin's only operating nuclear plant, has seen its cost per megawatt-hour rise from $45.94 in 2016 to $75.51 in 2026, with projections reaching $122.45 by 2033. Although Oracle is funding the energy costs for its allocation, the subscription is a primary driver of a proposed $176 million electric rate hike for We Energies customers in 2027, accounting for roughly 20% of that increase. We Energies projects the Oracle deal will save customers approximately $300 million in fuel costs between 2027 and 2033, but the immediate impact remains contentious as the Wisconsin Public Service Commission weighs whether tech companies should cover 100% of new power plant costs.

This arrangement reflects a broader industry pivot toward nuclear energy. Big Tech has contracted over 10 gigawatts of new nuclear capacity in the United States over the past year. Microsoft has secured a 20-year power purchase agreement for the 835-megawatt restart of Three Mile Island, a $1.6 billion investment expected online in 2027. Amazon has acquired a nuclear-adjacent campus in Pennsylvania for over $650 million and invested $500 million in small modular reactors from X-Energy. Google has ordered 500 megawatts from Kairos Power, targeting 2030. These commitments underscore a central reality: energy has become the primary bottleneck for the compute supply chain. With data center power demand reaching 29.6 gigawatts by late 2025—equivalent to New York State's peak demand—and projected to rise 130% by 2030, the industry faces a constraint that deploying more GPUs cannot resolve.

The compute landlord thesis now extends beyond real estate and hardware into energy procurement. The structural dependency is clear: as AI models scale, demand for reliable baseload power grows proportionally. Yet a widening gap separates these commitments from physical reality. Despite announced contracts, U.S. nuclear output has remained largely flat between 2020 and 2025, and the country currently has no small modular reactors under construction. Wisconsin's peak demand is forecast to rise 40% over the next five years, driven substantially by data centers, creating direct conflict between hyperscaler needs and local ratepayer interests.

The central question is who absorbs the risk of this energy hunger. While Oracle's subscription represents a novel model in Wisconsin, it exposes the tension between private corporate infrastructure and public utility grids. The physical strain on existing infrastructure and regulatory battles over cost allocation remain obscured amid AI capacity announcements. As projects like Stargate advance, the industry transitions from rapid hardware acquisition to urgent energy acquisition. For investors and operators, the ability to secure power has become as critical as securing silicon. The Oracle-We Energies deal illustrates this transition: the future of AI is increasingly tethered to the finite capacity of the aging physical grid.