On October 2, 2026, Oracle and We Energies announced a nuclear power subscription deal that marks a significant shift in how hyperscale data centers secure their energy. Oracle has agreed to subscribe to 10-20% of the output from the Point Beach Nuclear Plant, amounting to 125-250 megawatts of power. This energy is earmarked for the Lighthouse Campus, a $15 billion AI data center project in Port Washington, Wisconsin, co-developed by Oracle, OpenAI, and Vantage Data Centers. The facility, which is part of the broader Stargate initiative, is designed to house close to 1 gigawatt of AI capacity within a 1.3 gigawatt total electrical footprint, with completion targeted for 2028.
The mechanism of this deal reveals the escalating costs of the AI compute 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. While Oracle has committed to fully funding the energy costs for its portion of the project, the deal itself 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. The utility projects that the Oracle deal will save customers approximately $300 million in fuel costs between 2027 and 2033 by shifting the burden to the tech giant, but the immediate impact remains a point of contention as the Wisconsin Public Service Commission considers whether to require tech companies to cover 100% of new power plant costs.
This move is not an isolated incident but part of a frantic industry-wide 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 entered a 20-year power purchase agreement for the 835-megawatt restart of Three Mile Island, a $1.6 billion investment expected to come online in 2027. Amazon has acquired a nuclear-adjacent campus in Pennsylvania for over $650 million and invested $500 million in X-energy small modular reactors. Meanwhile, Google has ordered 500 megawatts of capacity from Kairos Power, targeting 2030 for its first reactor. These deals underscore a harsh reality: energy has become the primary bottleneck for the compute supply chain. As data center power demand reached 29.6 gigawatts by late 2025-equivalent to the peak demand of New York state-and is projected by the International Energy Agency to rise by 130% by 2030, the industry is realizing that energy cannot be solved simply by deploying more GPUs.
The compute landlord thesis, previously explored in our coverage of the $42 billion Broadcom financing and the Anthropic S-1 filing, now extends beyond real estate and hardware into the realm of energy procurement. The structural dependency is clear: as AI models grow, the demand for reliable, baseload power grows with them. However, there is a widening gap between these high-profile commitments and physical reality. Despite the surge in contracts, nuclear output in the U.S. has remained largely flat between 2020 and 2025, and there are currently no small modular reactors under construction in the country. Wisconsin’s peak demand is forecast to rise by 40% over the next five years, driven largely by these data centers, creating a direct conflict between the needs of hyperscalers and the interests of local ratepayers.
The question is who ultimately absorbs the risk of this energy hunger. While Oracle’s voluntary subscription is a first-of-its-kind model in Wisconsin, it highlights the tension between private corporate infrastructure and public utility grids. The part that gets hidden in the excitement over AI capacity is the physical strain on existing infrastructure and the regulatory battles over who pays for the necessary upgrades. As the Stargate initiative and similar projects move forward, the industry is moving from a phase of rapid hardware acquisition to a phase of desperate energy acquisition. For investors and operators, the ability to secure power is now as critical as the ability to secure silicon. The Oracle-We Energies deal is a concrete example of this transition, illustrating that the future of AI is not just being built in the cloud, but is increasingly tethered to the aging, finite capacity of the physical grid.
Note: All deal terms and cost figures are from We Energies and Oracle’s October 2, 2026 announcement as reported by the Milwaukee Journal Sentinel and Wisconsin Watch. The subscription requires approval from the Wisconsin Public Service Commission and is not yet finalized. The $300 million savings figure is We Energies’ projection, not an independently verified amount.
