Energy security was the subject of the first Executive Order (EO) issued by US president Donald Trump on 20 January 2025 – the same day he was inaugurated. On 14 February, he issued a second energy EO that required immediate production of a National Energy Dominance Strategy. April saw three further EOs in the energy sector; on reinvigorating the coal industry, limiting the power of states over energy policy; and making the electricity grid stronger and more reliable. A month later, presidential attention was on the US nuclear industry, and four EOs covered new reactor designs, reforming the Nuclear Regulatory Commission, reforming reactor testing, and reinvigorating the nuclear industrial base.
Trump is not the first president to focus on the energy sector, and especially the electricity sector. The reason is the fear of a power shortfall, combined, in recent years, with growing concerns over reliability.
In 2024, the USA had nearly 1.3 TW of electricity generating capacity, but it is estimated that on current demand and supply trajectories, the USA will have a 100 GW capacity shortfall by the mid-2030s, as the building of generating capacity lags behind fast-growing demand. Although electrification of industry and transport plays its part in demand growth, by far the biggest issue is a huge increase in demand from data centres. The investment being made in data centres is illustrated by the US state of Virginia, which has been a major area of data centre construction. The world’s largest data centre operator, Amazon Web Services (AWS) invested $51.9bn in data centres in Virginia in the 10 years to 2021. In January 2025 it announced plans to invest a further $35bn in new fleets of data centres in the state by 2040, although in July it responded to local objections by cancelling one of three ‘campuses’.
But although Virginia was an early leader in hosting data centres – and sometimes dubbed ‘data centre alley’ – many other jurisdictions are catching up. In June alone, immediately before its retrenchment in Virginia, AWS announced plans for a $20bn investment in Pennsylvania and a $10bn investment in North Carolina.
Derek Coleman is a senior manager in Baringa’s US Energy & Power Markets Advisory business. Coleman acknowledged there was a question about “whether the expected load growth going to show up” but he noted that although it was partly driven by electrification of industry, the biggest growth by far is new data centres.
He discussed the uncertainties over that expansion and why investment in generating capacity for data centres will continue, despite those uncertainties. He said, “Data centres will be the largest load, but the projection that also has the most question marks. The hyperscaler businesses [ie AWS, Google etc], have shown they are in an arms race. They are looking for growth at all costs.” They are spending many billions on data centres for AI because, “If the [AI] boom is delivered as predicted, no-one wants to have missed out – that will cost real dollars.” As a result, “they are willing to take the risk of the bubble bursting… They have done the analysis and the energy side is a second derivative”. Even investments of billions of dollars represent a relatively small amount, where the big risk is being left behind in the data centre arms race.
Fast action expected on renewables
The Biden administration’s Inflation Reduction Act (IRA) incentivised the build-out of wind and solar power, seen as the most economical choice to rapidly expand generating capacity, as well as the least carbon-intensive. The administration (and the IRA) also supported expansion of the US nuclear power station fleet and life extensions for plants already in operation.
The Trump administration has replaced the IRA with the One Big Beautiful Act (OBBA), which cuts back radically on supporting renewables, instead stressing the need for “firm, dispatchable” sources of generation. The administration has also cited April’s EO that limits states’ power over energy matters to halt investment in new transmission networks if they are to be used to export renewable energy.
The focus on firm power should be good news for the nuclear industry, and pronouncements from the Trump administration have continued the support already given to the nuclear sector by the Biden administration.
However, in an analysis of the likely build-out of all forms of new generation in the coming decades, Baringa found that while the change from IRA to OBBA had a dramatic effect on some forms of generation, for nuclear it failed to shift the dial at all.
The Baringa analysis found that despite the Trump administration’s EOs and the OBBA, which all included measures intended to quash wind and solar generation, the initial effect of the OBBA would be to speed up deployment of these types of generation.
That is because the production tax credits that were mandated to support them by the IRA do not end immediately. After OBBA, the deadline by which wind and solar farms have to be in operation to qualify for credits has been brought forward and the window now closes in around two years, so wind and solar developers are rushing to install projects that were previously due to start up beyond that date ahead of schedule, before the deadline passes. The result is a short-term boom in deployment, as Baringa’s analysis shows in Figure 1.
As expected, the OBBA also favoured new gas-fired and even coal-fired stations. However, that still leaves a gap in the 2030s. Baringa’s analysis found that “despite improved incentives, gas and firm power expansion is insufficient to meet growing energy demand”. It said growing demand, along with retirements in the existing fleet, would leave a gap of at least 100 GW and “Wind and solar therefore have a clear role in meeting demand growth”.
The analysis found that the impacts of OBBA included higher annual carbon emissions, even when demand was decreasing, because through the 2030s the change in the rate of retirement results in higher levels of coal and natural gas-based generation on the system.
As regards the contrasting legislation, Coleman characterises the IRA as “decarbonise at all costs” and the OBBA as seeking “firm dispatchable power”. He adds, “The latter pushes back at wind and solar, but leaves alone measures for nuclear.” He believes the evidence is that nuclear “has bipartisan support”.
But overall, Baringa found that more wind and solar would be installed in the near term under the OBBA than under IRA and the amount of new nuclear would be unchanged.
Meeting the need for ‘firm’ power
Given the bipartisan support for nuclear, and its status as a ‘firm’ source, why is it not top of the list to fill the looming generation gap? Coleman says “Nuclear’s headwinds are not a lack of policy support”. Instead, he lists some familiar issues: cost over-runs; supply chains, new and existing; employment and skills; permitting and licensing; and public perception.
Coleman says “The OBBA shores up the economics of the existing fleet and lays a groundwork for support for new units, but the headwinds are outside federal government support”.
Behind some of these issues are the extremely long lead time required to build a nuclear unit. That raises questions over maintaining political support: “Look at Vogtle [which has recently started up]. It has taken 17 years and that could have been under up to five administrations. The best case scenario would be two federal administrations in 10 years”. At the same time successive state administrations have to be considered. He adds, “Historically it was easier to build nuclear units under older market designs with fully-regulated utilities, who added the unit to their rate base. In a regulated environment there is more confidence that you can cover your cost.”
In deregulated electricity markets, “Who can invest in such massive capital project when [past] cost overruns have been substantial? … There is a real challenge in the uncertainty and it makes it very difficult to underwrite a project… If you could build in 10 years, instead of 17, you might underwrite it”.
Coleman says scalability – and the possibility of building fast and using standard units – is part of the appeal of so-called small modular reactors (SMRs). This is one reason why the ‘hyperscaler’ data centre operators have expressed interest. But Coleman warns that his list of headwinds remains the same for SMRs. “SMRs still face permitting and perception risks and the technology is not commercialised. The modular nature of the technology is not a benefit if you need a gigawatt of power [eg for a datacentre],” while the alternative – likely to be a gas turbine – can be bought and installed ‘off the shelf’.
He says the hyperscalers and data centres “may be a catalyst” to help get SMRs built, but he says the timeline to build them at-scale is still out of alignment with how hyperscalers want them to be deployed to serve data centres. The deals that have been signed between SMRs and hyperscalers are “not really that firm”, he says, and given the long lead times, if they are realised it may not be at the originally expected site.
Meeting the need for ‘dispatchable power’
If nuclear can overcome the headwinds Coleman describes, it is in a good position to provide the ‘firm’ power at scale that the US is seeking. However, it is less clear that it can help meet the country’s need for dispatchable capacity – capacity that can act in a flexible way to meet the demands of the grid operator, whether that flexibility is required over timescales of seconds, hours or days.
Once again, it is data centres that represent a major challenge. The North American Electric Reliability Council (NERC), highlighted this challenge in its 2025 ‘State of Reliability’ report on the US grid, published in June 2025. It said large data centres are “a significant near-term reliability challenge,” especially as they “can be developed faster than the generation and transmission infrastructure needed in the area to support them, resulting in lower system stability”. That challenge may be because the load disconnects in response to external events – as in an event in 2024, when 1.5 GW of data centres disconnected simultaneously and unexpectedly due to a transmission line fault in 2024. The report said Texas’s ERCOT system operator had experienced similar events, but at the 100–400 MW scale.
NERC noted “rapid changes in load are part of normal operations for these facilities, which raises concerns for balancing, frequency stability, and voltage stability” and said “the voltage sensitivity and rapidly changing, often unpredictable, power usage of these facilities creates new operating challenges”. Fast response is required, and for this, batteries are increasingly becoming the go-to technology. Batteries represented an “ever increasing portion of ERCOT’s ancillary services market, primarily for frequency regulation services, responsive reserve services, fast frequency response, contingency reserve services, and non-spinning reserves,” it said. Installed battery capacity in Texas had reached 10 GW in December 2024, up from 1.3 GW in January 2022, and that was set to increase significantly.
A huge increase in utility-scale battery arrays has been a feature of many electricity systems over the last decade and it is not only Texas where the USA has joined the rush. According to the US DOE’s Energy Information Administration (EIA), annual battery installations leapt from 43 MW in 2003 to 6.8 GW in 2023. Wood Mackenzie figures suggest that utility-scale installations will reach 16.2 GW in 2025, up 49% on 2024. As with wind and solar, battery installers are rushing to complete new projects before OBBA-led changes to tax credit regulations take effect in 2026.
Increasingly, batteries are co-located with solar and wind sites, because the combination allows site exports to be optimised and helps manage renewables’ volatility, while the battery can participate in near-term energy markets or ancillary markets.
Currently nuclear has little track record in joining ancillary services markets, although of course it provides inertia to the system, because it does not access the type of flexibility required, either as a large GW-scale unit or an array of SMRs. Co-located batteries may provide that market opportunity.
What is more, increasingly nuclear will be operating in markets where there is regular surplus and the ability to reduce export to the grid is valuable – something that could be achieved by charging up battery arrays on-site.
If nuclear wants to become as attractive a ‘dispatchable’ source of power as it is a ‘firm’ source, plant operators may have to consider using space on their site to co-locate batteries. That could provide flexibility to complement the firm power produced by the nuclear plant.
