What will it take for hyperscalers to revive a stalled nuclear industry?

The status of nuclear power appears to be changing dramatically. Since 2012, 12 U.S. nuclear reactors have permanently closed, and the only new facility arrived years late and billions of dollars over budget. But in the last two weeks, nuclear has reemerged as a promising option for clean firm power, especially to fuel data centers.

In late September, Microsoft signed a 20-year power purchase agreement to help restart the remaining reactor at Three Mile Island. And on Monday the Department of Energy’s Loan Programs Office closed a $1.52 billion loan guarantee to restart an 800-megawatt nuclear generating station in Michigan. 

The increase in demand for clean, firm energy, particularly for data centers, and the high-profile interest from the tech industry, is creating a new market signal for nuclear. That said, putting more nuclear power on the grid — whether through reviving shuttered plants or building new ones — is complicated. At present, only two nuclear power plants have signed agreements to power data centers. New projects, from licensing to engineering, promise to be costly and lengthy. 

And logistical challenges aside, it remains to be seen whether the signals that Microsoft and other hyperscalers are sending regarding their willingness to finance energy projects are enough to rebuild a nuclear market that has been declining for decades. As DOE put it in its liftoff report on the technology, nuclear has faced “a commercial stalemate.”

The agency, for its part, sees a major role for tech companies to play in pushing the market forward, just as they’ve done for other, more nascent technologies like carbon removal and advanced geothermal. As in the case of those technologies, the cost of the first few plants is likely to be “higher than is economically competitive,” DOE acknowledged.

Unlike those technologies though, the nuclear industry has been around for quite some time, and has the potential to get down the cost curve to around $60 per kilowatt hour in relatively short order, according to DOE's report, by building just eight reactors.

But, as the agency put it, “no one can be fifth if no one leans in to be first, second, third, and fourth."

Diffusing risk

Getting nuclear to scale — meaning around ten projects — would generate so much power and require so much capital investment that it would be extremely difficult for a single company to offtake it alone. That’s where industry offtakers could come in, to share early costs and lower barriers to entry, and to guarantee a market.

One option for diffusing risk and getting nuclear to scale, is what DOE calls a “buyers’ consortium.” That could take several forms, including aggregated offtake by tech companies, who could also step in to own or invest equity in projects. That would allow buyers “to all be ‘one fifth’ of the orderbook versus ‘waiting to be fifth,’” DOE said. 

The key difference between a potential nuclear consortium and something like the Frontier Coalition — an advance market commitment to purchase carbon dioxide removal, funded by major corporations — is that Frontier members committed to investing $925 million over a set period of time.

That’s a familiar structure for hyperscalers, but getting nuclear to “liftoff” will require a different approach.

Microsoft’s deal with Constellation to reopen the remaining Three Mile Island reactor, for example, isn’t a fixed amount of money or even a fixed cost per kilowatt hour. Because Microsoft signed that deal alone, it’s likely to pay a premium, but a coalition of tech offtakers could offset early costs.

But there’s another potential market benefit to the consortium approach: the ability to bring in big banks. At Climate Week in New York in September, a group of banks including Citigroup, Goldman Sachs, and Morgan Stanley, voiced their support for efforts to triple nuclear energy capacity by 2050.

However, it may be difficult for those institutions to move forward without another party stepping in to cover risk, said Nuclear Innovation Alliance senior fellow Stephen Greene. “It was great to see the commercial banks be part of that commitment,” Greene said on a webinar hosted by NIA this week. “But now the rubber meets the road and the question is what do the actual transactions look like?”

Commercial banks are likely to seek out a position on those transactions where someone else is taking on the bulk of the risk, Greene added. “Then the question is, who’s able to step into that position?”

Challenges remain

One initial technical hurdle facing would-be consortium members, however, is that various stakeholders have different preferences for which type of reactor they’d prefer.

According to DOE, utilities are likely to be most comfortable with Generation 3 nuclear reactors, which were developed in the 90s. Tech companies and other industrial offtakers may prefer Generation 4 reactors, which are still being developed, but which are modular and more efficient. They’ll have to reach consensus, however, for a consortium to work.

Steve Swilley, vice president of nuclear research and development at the Electric Power Research Institute, said small utilities are unlikely to take the risk of developing nuclear without an opportunity to share that risk with, for example, hyperscalers. But the structure of those deals may be non-traditional, he added. 

One issue that may be of particular concern to utilities is the PPAs themselves. Hyperscalers have been willing to embrace the financing model for new technologies, but those agreements generally span 20-year time frames — significantly shorter than the estimated 100-year lifespan of a new nuclear asset, Swilley said. The analysis of whether to restart a canceled unit will therefore be based in part on the remaining lifetime of the asset. 

“Is a 20-year PPA enough to justify the cost to restart and operate it with the plan that there’s probably going to be a market for that electricity?” Swilley asked. Given load growth even outside the data center industry, it’s likely that utilities wouldn’t have a problem selling power past the end of a PPA, he added, “but that’s still a business risk for utilities.”