Investments by private equity and venture capital are pouring into advanced nuclear technologies, financial executives said at the Atlantic Council Nuclear Energy Policy Summit on Sept. 23, though challenges to carrying out these projects persist.
In 2024, advanced nuclear companies saw about $800 million in investments, up more than 13 times from the prior year. That growth is expected to continue, following signals of support and policy changes from the Trump administration, said Kara Succoso Mangone, managing director and head of the sustainable finance group at Goldman Sachs.
Data centers have created a strong demand signal for nuclear, and hyperscalers have demonstrated a willingness to pay a premium for the generation.
"That's a pretty significant tailwind," said Mark Sowinski, an executive director at Morgan Stanley.
But while the AI boom is driving demand for nuclear power, the financing must follow, said Mangone.
"Microsoft getting Three Mile Island [Unit 1] reopened is a tremendous template," said Karen Fang, managing director and global head of sustainable and infrastructure finance at Bank of America Corp. "If we can do that, if we can actually have [power purchase agreements], we can start financing on the project level, and that will make the cost of capital even cheaper."
However, few opportunities exist to restart retired units in the US, and the costs associated with building new units from scratch, even with offtake agreements, could be much higher, finance experts on the panel said.
New projects are making financial progress, though, even from one or two years ago, said Alexei Viarruel, executive director of power and renewables investment banking at JP Morgan.
"The projects that we're seeing crossing our desks now are at a much higher level of maturity than they were before," Viarruel said.
If the finance industry is going to bite on nuclear projects, they need to meet certain thresholds, according to Viarruel. Those include "high-quality offtakes on ironclad PPAs with, ideally, an investment-grade counterparty" as well as "high visibility to the equity portion of the financing pool" and reliable contractors, Viarruel said.
Many more projects — ranging from large-scale reactors such as AP1000s to small modular reactors (SMRs) — now have some of the necessary elements to appeal to the finance industry, Viarruel said. "I haven't seen a perfect package just yet, which is why we haven't done one of these deals."
Bank project financing also has never been done for nuclear projects, panelists said.
"This is ... very new for the industry and we're sort of trying to meet the challenge that the industry has put to us," Viarruel said. "But what we're seeing is increasingly approaching what the ideal looks like."
Panelists forecast a decade before the US sees another large-scale reactor such as the AP1000 units that came online in 2023 and 2024 at Southern Co.'s Vogtle Nuclear Plant expansion in Georgia after years of delays and major cost overruns. Experts were more optimistic about SMRs, which they predicted would likely come online earlier.
But the advanced technologies carry their own challenges, with none yet commercially available in the US.
"That is a risk that would require at least some operating history before the bank financing market really steps in," Viarruel said. "Before that, it's probably going to be a lot more equity-heavy and a lot more quasi-government lending."
The US has advantages over other countries, including a large existing fleet with extended operating lives, an approved design in the AP1000 and advanced technologies such as SMRs with support from the US Energy Department.
The DOE has also recently pledged to deploy three new reactor technologies by July 4, 2026, said Aleshia Duncan, deputy assistant secretary for international cooperation at the department's Office of Nuclear Energy.
"That gives me hope that maybe that's also more near around the corner than we think," said Fang. "If DOE really gets their moonshot and we start seeing these [SMRs] coming online, maybe this is a three- to five-year game, not a five- to seven-year game."
