Hogan Lovells 2024 Election Impact and Congressional Outlook Report
15 November 2024
Recent announcements pairing Big Tech and nuclear energy have hit the news lately. But where did this interest come from, and what does it mean for nuclear energy in the United States? We walk through the emerging relationships between the two industries below, looking at data centers energy needs, the unique attributes valued from nuclear energy, and the types of relationships emerging in more detail below.
Energy demand in the United States is projected to rise massively over the next decade due to growth in data centers, industrialization and electrification. Of the three, data centers are the leading contributor to this projected demand increase—with their growth predominately driven by the emergence of generative AI, which needs major data processing capabilities and—as a result—consumes a large amount of electricity. Large data centers can consume about the same amount of electricity as a medium-sized city. The Electric Power Research Institute estimates that data centers alone, which currently use about 4% of U.S. electricity, could consume up to 9% of U.S. electricity generation annually by 2030—which, if measured as a percentage of 2023 generation, would be about 376,000 gigawatt hours, or the electricity production of the entire country of Indonesia (for scale). Adding in projected increases from new domestic manufacturing, rise in electric vehicles, and broader electrification, the total energy demand in the United States could grow 25-29% in the next decade—this is double the projections from just last year, and the number continues to rise.
This skyrocketing demand increase is putting pressure on utility companies to meet U.S. electricity needs, something that becomes increasingly more difficult to do as more data centers come online. Fears of increased household electricity rates and grid reliability caused U.S. Department of Energy (DOE) Secretary Jennifer Granholm to urge tech companies to provide their own energy sources for data centers. And tech companies are well poised to meet this challenge—with capital investments by tech companies at $54 billion just last quarter, we are seeing a surge in announcements emerge about new data center and nuclear pairings. The new marriages between Big Tech and nuclear are disrupting the nuclear industry, and bringing new players—with deep pockets—into the sector. These new players are changing the landscape for business arrangements, leaving regulators to figure out how to implement the new changes while still ensuring energy stays affordable and the lights stay on.
Nuclear energy is a natural fit to meet this new demand.
While all these new announcements may seem sudden—leaving some scratching their heads trying to figure out what’s going on—many of the projects have been in the works for a while, just now seeing the light of day in the press. For many people working on nuclear, these announcements seem like natural extensions of a conflux of strong market conditions. Data centers need an immense amount of power; as the most energy dense source of power, nuclear can provide that. Data centers need reliable power; nuclear can provide that, with the highest reliability rating among all power sources. Many data centers want carbon free power; nuclear is the largest source of clean power in the U.S.—by far. Data centers are willing to pay for the externalities of nuclear—such as their reliability and clean power—where the complex U.S. electricity market generally will not.
Nuclear power plants can produce an immense amount of reliable, carbon-free power—necessary for data centers, which operate around-the-clock and are energy-intensive. Currently, the U.S. maintains 54 commercially operating nuclear power plants with 94 reactors total, with each reactor producing roughly about one gigawatt of electricity. Not only is this energy dense, but it is reliable. Nuclear power has the highest capacity factor of any electricity source at 93%, about twice as much as natural gas and three times more than solar and wind. Capacity factor is how much energy a power plant actually produces compared to its “nameplate” or on-paper capacity—essentially a measure of how reliable a plant is. This high capacity factor means that replacing a single one-gigawatt nuclear reactor with an energy alternative, in practice, would need about two gigawatts of natural gas plants or three gigawatts of wind turbines—or about 1,300 utility-scale wind turbines.
Climate change necessitates that any new generation be carbon-free. Tech giants like Amazon and Microsoft, for example, have committed to reaching carbon neutrality in their own operations. Nuclear energy fits the bill to support these commitments. Nuclear power plants produce carbon-free electricity, and currently, nuclear produces about 46% of the United States’ carbon-free energy. For instance, Georgia’s Plant Vogtle, consisting of four units—with the fourth unit beginning commercial operation this year—is currently the largest source of carbon-free electricity in the United States, with a 4,658 MW nameplate capacity.
Nuclear energy today is also flexible—the myriad of different sizes and designs means that nuclear energy can be deployed almost anywhere and to fit nearly any end-user. There are a few main “types” of nuclear reactors, including traditional light-water reactors (LWRs), Gen III+ LWRs, advanced reactors, and microreactors—which can be advanced or LWRs. Each type of reactor has a different strength that can support various needs. Large LWRs like traditional or Gen III+ LWRs have powerful economies of scale and immense power output, while advanced nuclear and micro-reactors could be sited closer end-users like data centers. Some high-temperature advanced reactor designs could even produce process heat for industrial uses. Notably, DOE discussed the commercial potential for various types of nuclear energy in their recently updated Pathways to Commercial Liftoff: Advanced Nuclear report, which we wrote about here.
In today’s political climate, another advantage of nuclear energy is that it is incredibly bipartisan. In July, the ADVANCE Act—the most comprehensive piece of energy-related legislation since the Bipartisan Infrastructure Law—passed Congress with overwhelming bipartisan support and became law. See here for more details on the ADVANCE Act. Before that, the Nuclear Energy Innovation and Modernization Act (NEIMA) passed with bipartisan support to develop a licensing process for new nuclear reactors. Programs like the Advanced Reactor Demonstration Program and funding opportunities like $900 million for new LWRs and DOE contracts for high-assay low enriched uranium (HALEU) have similarly received support across the aisle (written about here).
Given the natural synthesis between data centers and nuclear energy, it makes sense that tech companies are looking to nuclear to meet their electricity needs and climate goals. In particular, data centers are looking at using nuclear energy in a few ways, including co-location with an existing operating plant; restarting a plant in decommissioning; and building new reactors as a dedicated power source. With each type of agreement comes a host of novel business arrangements, as well as regulatory, policy, and other issues to consider.
As data centers look for large sources of reliable electricity, one seemingly simple answer is to co-locate new data centers with existing power plants and purchase electricity directly from the plant. This idea has generated both excitement and concern, with questions about impacts on grid reliability and ratepayer costs still up for debate.
One recent co-location announcement has been paused by the Federal Energy Regulatory Commission (FERC). In March 2024, Talen Energy announced that it had sold a data center campus near Susquehanna Nuclear Power Plant in Pennsylvania to Amazon Web Services. Talen intended to sell power to Amazon from its 2,228 MW stake in Susquehanna, beginning in 120 MW increments with the possibility to grow to 960 MW. Talen shortly thereafter submitted to FERC an amended interconnection service agreement (ISA) to increase Talen’s behind-the-meter connection to Susquehanna from 300 MW to 480 MW to accommodate the deal.
However, last Friday, FERC rejected the amended ISA in a 2-1 vote. The decision rested on the technical justification that the amended ISA did not conform to the pro forma ISA issued by regional transmission organization PJM, but both Chairman Willie Phillips (in his dissent) and Commissioner Mark Christie (in his concurrence) commented on the broader ramifications of co-location, with Commissioner Christie noting the potentially “huge ramifications” on grid reliability and consumer costs. Interestingly, FERC has 5 Commissioners—but the other two Commissioners did not participate in the decision. With a number of other behind-the-meter data center projects pending at FERC, this is a space to watch to see how things pan out. Even if not central to last week’s decision, FERC Commissioners remain concerned about the impact on power prices and grid reliability.
Talen is moving forward with the deal, planning to supply Amazon with the amount of electricity currently allowed under the existing ISA—300 MW. Some are concerned that FERC’s rejection of the ISA could have a chilling effect on similar deals in PJM’s transmission territory, while others question FERC’s overall position on co-location. Time and further progress will reveal the path forward.
Last month, Microsoft and Constellation Energy announced the signing of a 20-year power purchase agreement (PPA) predicated on restarting Three Mile Island Unit 1, which operated from 1974 to 2019. Under the PPA, Microsoft will purchase energy from the restarted Unit 1—to be renamed Crane Clean Energy Center—to supply power for nearby data centers. The power level from the restart would be enough to power approximately 800,000 homes.
If restarted, Three Mile Island Unit 1 would be the second nuclear reactor to be restarted in the United States after Palisades Nuclear Plant in Michigan, which is currently undergoing NRC review and which the licensee Holtec plans to see operating by October 2025. Although Palisades’ electricity will be sold to nearby rural electric cooperatives, its restart could pave the way for other shutdown plants to be restarted to supply targeted end-users.
In addition to Constellation’s proposed restart of Three Mile Island Unit 1, NextEra Energy is considering restarting the shutdown Duane Arnold nuclear power plant in Iowa due to growing interest from data center companies.
Perhaps the biggest announcements lately have come from tech companies investing in new advanced reactor projects—including through direct equity investments in companies deploying nuclear and planned PPAs, or just planned PPAs.
In its latest Pathways to Commercial Liftoff: Advanced Nuclear report, DOE modeling indicates the need for at least ~200 GW of new nuclear—triple the existing capacity. However, new nuclear still faces several hurdles to commercialization. DOE anticipates that to deploy at scale, advanced nuclear in particular first needs a committed orderbook of at least 5-10 deployments of a single reactor design (written about here).
Here, big tech and new nuclear are a match made in heaven. With the announcements coming out over the past few weeks that book order has appeared to be emerging.
Recent announcements put at least two companies—X-Energy and Kairos Power—well on their way to developing a committed orderbook. And big tech companies can afford to pay for new project risk in a way regulated utilities or power cooperatives cannot. These announcements, described more below, have the potential to jump-start the commercialization of new nuclear in the United States.
While large tech companies like Google and Microsoft have made significant strides in partnering with nuclear energy providers to power their data centers and other energy-intensive operations, the tech industry is also actively exploring and investing in fusion energy as a potential long-term solution. Fusion, often referred to as the energy that powers the sun and stars, has the potential to provide a virtually limitless, clean, and safe energy source.
A recent example of tech and fusion is the power purchase agreement that Microsoft signed with Washington-based fusion company, Helion, to obtain fusion power by 2028. While we provided the details of this announcement here, Microsoft stands to benefit from leveraging fusion energy for purposes of sustainability, reliability, cost-effectiveness, and brand reputation – because investing in and using fusion energy enhances Microsoft's image as a forward-thinking company committed to sustainable practices and technological advancement.
And it doesn’t stop there; Helion is heavily active in the tech space. Earlier this year, OpenAI was reportedly looking to buy nuclear fusion energy from Helion to power its data centers. According to a report in the Wall Street Journal, the ChatGPT maker was in talks to buy “vast quantities” of energy from the startup, which is chaired by OpenAI CEO Sam Altman. As another example, in February 2024, SMX, an IT solutions company, and Fusion Technology, a small business, announced a joint venture, which was formed under the Small Business Administration's mentor-protégé program.
As tech companies continue to grow and demand more energy, fusion energy presents an attractive option that could help them meet their sustainability goals while ensuring a reliable energy supply for the future.
The recent projects announced by tech companies upend some of the ways nuclear power is traditionally financed, regulated, and built.
For one, business structures are evolving. Nuclear power plants were traditionally financed under a “build-own-operate” model, wherein a utility (or group of utilities and electric cooperatives) would build, own, and operate the plant. The utility or group would use the nuclear plant to provide power for distributed customers on the electric grid, and the utility would recover the cost of the plant from rates. Now, unique business arrangements are emerging: PPAs with offtakers who are willing to pay a premium for reliable energy (such as Microsoft’s deal for Three Mile Island Unit 1); equity investments into nuclear reactor tech companies with plans to deploy a fleet (such as the Amazon/X-energy deal); and a potential developer model with an ultimate utility owner and operator (such as Amazon’s recent deal with Dominion Energy to build an SMR in Virginia).
These new announcements invoke novel regulatory considerations as well. For instance, the potential restart of Palisades, with Three Mile Island Unit 1 to follow, has never been done before and therefore has no established regulatory pathway at the NRC—although the NRC has many existing regulations that can and do apply. Thus, the NRC has had to innovate a review process to ensure a restarted Palisades conforms with all NRC regulations for operating nuclear reactors. Licensing advanced reactors also involves very different regulatory review than licensing a large LWR—something the NRC has been working on for a while now, offering a range of regulatory options, including the new proposed Part 53 rule. The NRC’s successful and timely licensing of the Kairos reactor, the first of the wave of expected new advanced reactor applications, bodes well for the industry. At FERC, any deal that aims to buy power from a nuclear plant that is connected to the grid may have to develop novel interconnection agreements and other regulatory strategies.
First-of-a-kind projects, like some of these announcements, also pose logistical issues of their own. The new reactors Amazon and Google have ordered will not be operating tomorrow, and the first of any new design to be built incurs unique issues as the supply chain and workforce develop, as Plant Vogtle demonstrated (to some extent). Once the first few reactors are built, though, the quicker, more predictable, and more cost-effective it will be to build them—as described in the DOE Liftoff report. Restarting a shutdown plant like Three Mile Island Unit 1 also poses technical challenges, which can be unique to each plant depending on when the plant shutdown and what condition it has been kept in. Once the first plant is restarted, though, the nuclear industry could learn technical lessons and apply those to other plants when applicable.
As demonstrated, there is a lot of movement in the tech-nuclear space, both publicly and behind the scenes. These projects—those that have already been announced and those yet to come—have the potential to be a game changer for the nuclear industry in the United States. Investments from tech are already challenging traditional ideas about financing, regulating, and building nuclear energy. At the same time, these announcements can jumpstart the use of nuclear energy for more widespread uses, both on the grid and in non-traditional uses like industrial decarbonization. Those following the nuclear industry, the data center industry, or energy in general, should look out to see how this space develops going forward.
For more information, please contact Amy Roma, Partner, Stephanie Fishman, Senior Associate, or Cameron Tarry Hughes, Associate.