Data as of: 2026-10-04 Issued: 2026-10-04 (America/Toronto) Evidence key: CONFIRMED (primary source), REPORTED (multiple outlets), RUMORED (unverified market reporting), ANALYTICAL (inference).
Oracle and We Energies announced on October 2 that Oracle will subscribe to a portion of the existing Point Beach Nuclear Plant’s electricity generation — 10 to 20 percent of the plant’s output, or roughly 125 to 250 megawatts. The power 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 under the Stargate initiative, designed for close to 1 gigawatt of AI capacity within a 1.3 gigawatt electrical footprint and targeted for completion in 2028.
The mechanism is as notable as the megawatts. Oracle committed to absorb approximately $300 million in rising Point Beach energy costs, a move expected to save Wisconsin utility customers about $300 million in fuel costs between 2027 and 2033. The deal is also a driver of a proposed $176 million electric rate increase for We Energies customers in 2027 — about a fifth of the increase — and the Wisconsin Public Service Commission is weighing whether technology companies should be required to cover 100 percent of new power-plant costs.
The subscription extends a year-long Big Tech pivot toward nuclear. Microsoft signed a 20-year power-purchase agreement for the 835-megawatt Three Mile Island restart in a $1.6 billion deal; Amazon acquired a nuclear-adjacent Pennsylvania campus and put $500 million into X-energy small modular reactors; Google ordered 500 megawatts from Kairos Power. Oracle’s structure goes further: rather than a PPA for new capacity, it is taking a slice of an operating plant and funding the cost escalation itself.
Point Beach’s economics show why. The plant’s cost per megawatt-hour has risen from $45.94 in 2016 to $75.51 in 2026, with projections reaching $122.45 by 2033. With US data-center power demand surging and grid-connection queues stretching for years, hyperscalers are discovering that energy — not chips — is the binding constraint on compute. The Wisconsin rate case will determine whether the subscription model becomes a template or a cautionary tale.