Google and Constellation Strike $4.3 Billion Nuclear Uprate Deal, Adding 890 MW to the U.S. Grid

Why it matters: The deal illustrates a model in which hyperscalers finance capacity gains at existing nuclear plants rather than wait for new construction, sidestepping multi-year permitting and interconnection queues. For the PJM grid, which serves 67 million people across 13 states, 890 MW of firm, carbon-free baseload is significant. Google’s own 2025 environmental report showed electricity demand rising 37% year-over-year, yet direct emissions fell 2%, a decoupling the company attributes to large-scale clean energy purchases. The tension, however, is visible: value-chain emissions kept climbing, and Google acknowledged its 24/7 carbon-free energy ambition is becoming harder to reach as AI infrastructure scales faster than grid decarbonization.

What uprates mean for the nuclear expansion debate

Building a new nuclear plant in the United States takes a decade or more from licensing to commercial operation. Uprates offer a shortcut. By modernizing turbines, steam generators, and digital control systems at plants that already hold operating licenses, Constellation can deliver incremental megawatts without greenfield construction or new interconnection requests. The 890 MW target by end of 2032 amounts to roughly the output of a mid-sized new reactor, achieved at a fraction of the time and regulatory complexity. The approach also avoids the siting controversies that slow new projects.

This is not an isolated bet. In September, Google signed a 22-year PPA with Finnish energy company Fortum for a nuclear plant life extension and uprate, its first such deal. Constellation itself announced a 20-year agreement with Amazon the week before the Google deal. Microsoft’s 2024 agreement to restart Three Mile Island Unit 1 follows the same logic. The pattern suggests that U.S. tech companies view existing nuclear assets, not new-build reactors, as the fastest path to firm clean power at scale.

Who gains and who loses

The immediate winners are nuclear plant operators with aging but licensable fleets. Constellation, the largest U.S. nuclear operator, now has long-term offtake contracts that de-risk billions in capital expenditure. Google gets firm, carbon-free electrons it can count toward its sustainability targets without waiting for new plants. Union workers at the six plant sites benefit from the construction and modernization work tied to the uprates.

The losers are less obvious but real. Renewable-plus-storage developers competing for the same PJM interconnection capacity and corporate PPA dollars face a rival product (nuclear uprates) that offers 24/7 output without intermittency. Merchant generators without long-term contracts may see their competitive position erode as uprated nuclear plants push more baseload power into the market, potentially suppressing wholesale prices in PJM during off-peak hours. Communities hoping for new-build nuclear projects in their regions get a mixed signal: capital is flowing to upgrades, not to new sites.

What remains undecided

The strategic framework the companies announced to evaluate new clean generation, storage, and demand response is intentionally vague. It mentions identifying “potential projects, commercial structures, and policy pathways” but commits to nothing beyond study. Whether this framework leads to actual new-build nuclear, advanced reactors, or large-scale storage depends on outcomes that neither company controls: federal production tax credit extensions, NRC licensing timelines for new designs, and PJM’s own capacity market reform process. The 5-year technology alliance deploying Google Cloud and Gemini across Constellation’s operations is similarly open-ended in its deliverables. Until uprate construction begins and megawatts reach the grid, the 890 MW figure remains a contractual target, not an operational fact.

via ESG Today