Constellation Plans Nearly 1-Gigawatt Nuclear Uprate to Meet Surging AI-Driven U.S. Power Demand

U.S. electricity demand is climbing again, driven in large part by data centers and the rapid buildout of AI infrastructure. That surge is pushing utilities and policymakers to look for near-term sources of low-carbon power. In recent weeks, one strategy has moved back into the spotlight: getting more electricity out of the nuclear reactors already on the grid.

Constellation Energy, the largest nuclear plant operator in the United States, says it plans to increase output by nearly 1 gigawatt spread across 11 existing reactors at six sites. Days later, on Oct. 5, the U.S. Department of Energy announced a $4.2 billion loan to Vistra to support previously disclosed modernization work at three nuclear plants. The appeal is straightforward: compared with building new generating stations, boosting existing reactors could deliver additional clean electricity faster—potentially on timelines that better match the near-term needs of large data centers.

The approach is known as a “power uprate.” U.S. operators have pursued uprates since the 1970s, but the pace has fluctuated with market conditions. When electricity demand flattened and inexpensive natural gas squeezed nuclear plant economics, investing additional capital to raise nuclear output often didn’t pencil out. Some reactors shut down, and many operators prioritized cost control over expansion.

Regulatory statistics reflect that shift. From 2001 through 2010, the U.S. Nuclear Regulatory Commission approved more than 80 uprates. From 2011 through 2020, approvals fell to 33. Now, with demand forecasts rising, uprates are returning—alongside efforts to revive previously closed sites such as Palisades and Three Mile Island.

Not all uprates look the same. Some are relatively small, involving revised operating setpoints or updated engineering analyses that allow a modest increase in power. Larger projects can lift a reactor’s output by up to about 20%, but they typically require substantial hardware changes. These upgrades generally rely on using slightly higher-enriched fuel or increasing the quantity of fresh fuel loaded into the reactor, which raises heat output and produces more steam. To convert that extra thermal energy into electricity safely and reliably, plants may need new or refurbished turbines as well as upgraded pumps, motors, generators, and transformers.

One of the most recent major U.S. examples is the Browns Ferry nuclear plant in Alabama. A $475 million effort increased the capacity of three reactors there by about 14% during 2018 and 2019.

Constellation’s new plan is much larger in dollar terms: more than $4.3 billion for uprates across 11 reactors. The company says the added generation would be sufficient to supply electricity to roughly 700,000 U.S. homes. A key enabler is long-term demand from the tech sector. Google has agreed to purchase power from the uprated plants under a 20-year contract, a type of arrangement that can provide the price certainty needed to justify major nuclear investments. As Constellation CEO Joe Dominguez told The New York Times, these projects require a higher price than typical wholesale markets provide.

Vistra’s upgrades, supported by the DOE loan, are expected to add 433 megawatts across plants in Pennsylvania and Ohio. Meta has already signed a long-term agreement to buy electricity from those facilities.

Even so, uprates are not cheap. On a cost-per-kilowatt basis, Constellation’s expansion is less expensive than building a new nuclear reactor, but it remains costlier than constructing a new utility-scale solar project paired with storage, or a combined-cycle natural-gas plant.

How much more nuclear power is realistically available? The current U.S. nuclear fleet totals about 98 gigawatts of capacity. Estimates of additional potential from uprates range from roughly 3 to 8 gigawatts. Some of that is already in the pipeline: the NRC expects applications that would add 2.4 gigawatts through 2032.

Why it matters

Uprates could provide a meaningful—though limited—near-term boost in low-carbon electricity, especially for regions trying to power new data centers without increasing emissions. The projects hinge on capital, supply chains, and long-term contracts from large buyers such as Google and Meta. The next signal to watch is how quickly regulators approve applications and how many operators commit to the larger, equipment-heavy upgrades rather than minor adjustments.