Can Nuclear Energy Help Big Tech Meet Data Center Needs?

The surging need to feed power-hungry data centers is reviving interest in nuclear energy. As the momentum for artificial intelligence accelerates, Big Tech is turning to next-generation technology such as small modular nuclear reactors, or SMRs.

Part of the reason is that other carbon-free options won’t provide enough juice. “Wind and solar are not going to get it done,” says Travis Miller, energy and utilities strategist for Morningstar Research Services. Microsoft MSFT, Alphabet GOOG/GOOGL, and Amazon.com AMZN have inked nuclear energy deals, and their interest has sparked questions about the extent to which they’re influencing the direction of America’s power industry.

“Despite all the big headlines and all the discussion about tech, data centers really are a fairly small part of the total US electricity demand,” says Miller. They are making headlines because they are the fastest-growing source of demand. However, electric vehicle demand also stands to grow significantly. Meanwhile, bringing SMR power stations online is a lengthy process; this isn’t a near-term solution.

This all bodes well for utilities, which Miller expects to show higher growth than in the past. Granted, valuations are also on the high side after the sector’s strong run in 2024. That said, there were still some attractive opportunities when we spoke on Dec. 3. Our conversation reflects valuations as of that date and has been edited for length and clarity.

Ivanna Hampton: Help us picture how much energy a data center needs to run 24/7. What is that equivalent to in our everyday lives?

Travis Miller: Some of these numbers are just incredible, and it’s kind of fun as a utilities analyst now because we haven’t had electricity demand growth projections and opportunities like this in a long time—even back to the Industrial Revolution. Probably the best bring-it-home literally is a typical home, which uses about 10 kilowatts when everything is on in the middle of the day or early afternoon. Some of these large data centers are going to do 150,000 kilowatts at their peak.

Expanding that, a large data center in Virginia uses about 50 times as much electricity per square foot as an equivalent square foot of office building space.

Hampton: Wow. Where will all this electricity come from?

Miller: In terms of the amount of electricity needed to bring just a single data center online, wind and solar are not going to get it done. But really that’s all utilities are building right now, and frankly that’s all the governments, either state or federal, want utilities to build for various environmental reasons. Not only are wind and solar too small, but they only run when the sun shines and the wind blows. That’s not when data centers want to run. Data centers want to run 24/7.

Utilities and data center developers such as Microsoft, Amazon, and Google are trying to find some way to get all of the electrons that they need to run these data centers. The fastest, cheapest option right now is natural gas power generation, but they’re also looking for some other source of power. Small modular reactors are a possibility. The technology has been around for a long time, and utilities have studied using SMRs in certain applications. Utilities never had the amount of demand and challenges serving that demand as they might in the next decade with data centers. Utilities are looking at any solution, and SMRs could be a solution, although they’re very expensive.

Hampton: When could these reactors come online—a decade from now?

Miller: That’s the big challenge. Utilities are facing a lot of pull demand from data center developers and the tech companies who want their data centers online and electrons right now. And the utilities are saying there’s too much risk around SMRs. You’ll need an SMR technology provider to work with the utilities to integrate them into the grid to serve data centers. Nothing happens quickly in the utilities industry. You’re absolutely right that we could be talking five-plus years before a single SMR were to come online, at the fastest.

Hampton: Data centers have a big environmental impact, and these companies have pledged to become carbon-free at some point. Is that what’s driving them toward nuclear energy?

Miller: Yeah. Solar and wind are obviously a zero-carbon solution, but data centers can’t use those as their primary source for electricity. There’s natural gas, which is an economic solution. It’s easy to build. It’s the best near-term solution for generating electrons to power data centers. But it does come with carbon emissions. That’s one knock against it.

Nuclear is carbon-free. It’s the only thing left in the toolbox for utilities. Either large-scale or small-scale nuclear is going to have to be a solution if those data centers want to be carbon-free. I think data center developers and tech companies are really struggling in terms of meeting some of their clean energy goals.

Hampton: Which utilities are signing deals?

Miller: Just about every utility has received some type of interest from data center developers, except for certain areas of the country where electricity is already expensive. All of these utilities have some substantial electricity demand needs over five to 10 years. I don’t think we’ll see utilities directly investing in either SMRs or large-scale new nuclear. Natural gas seems to be the preferred source of generation. I think analysts are searching for which utilities will find a solution to meet this demand growth.

Southern Company SO in Georgia is talking about 8 gigawatts of electricity demand commitments from new large customers, including data centers. They have 25 gigawatts of potential data center demand. For reference, the largest nuclear plant in the US is just over 4 gigawatts. SMRs can’t come close to that. SMRs are less than a gigawatt. You need a lot of SMRs, gas, solar, and wind to make any of these data centers viable.

Hampton: What are some people getting wrong about nuclear energy?

Miller: Definitely the cost and the timeline. In over 50 years of US history in the nuclear industry, nothing happens fast, on time, and on budget. I think that’s the concern that we, other analysts, and the industry have. If any utility were to go forward with a nuclear project, whether it’s connected to a data center or not, it could be facing a long timeline, cost overruns, and ultimately shareholder losses.

Hampton: Are we witnessing a seismic shift when it comes to nuclear energy, or is this really tailored to data center demand?

Miller: I think this is really tailored to data center demands. We think that core electricity demand growth is only going to be about 1%. As much as the headlines are reflecting the whole tech and data center buildout, data centers—not just the hyperscale, but data centers in general—are only 2.5% of US electricity demand. Even in the most aggressive scenarios, that grows to 5% or 6% in the next five-plus years. It really is a fairly small part of the total US electricity demand. The key reason data centers are getting headlines is that they’re the fastest-growing source of electricity demand in the US.

Yet the other wild card here to think about is electric vehicles. Data center excitement and some slack EV sales have kind of pushed EVs out of the headlines. But in our calculations, EVs could ultimately in the next decade or more be more of an electricity demand issue than data centers.

Data centers have massive numbers, but there are issues with getting electrons to them. Getting electrons to EVs isn’t a problem for most people who are charging at home or in store parking lots. You don’t need a lot of new generation in places serving EVs.

We think EV electricity demand could quadruple just in a pretty normal environment during the next decade. EVs are a very small part of total electricity demand, less than 1%. If they quadruple, they still only get to 3% or 4%. But that’s in line with data center demand in some scenarios.

There’s this real debate about what will be the key electricity demand growth driver during the next decade. Data centers and EVs are really driving new growth. I don’t know that utilities can serve both if they grow at superfast rates.

Hampton: What happens if utilities struggle to supply both?

Miller: The big question is what type of power generation and what fuel will be necessary to supply data centers. What we’re hearing is that it’s going to take many years to figure out that new supply source. Gas plants are the easiest, fastest, and cheapest way to generate power. Utilities are telling us it could be three years or more before they can get those online. Solar and wind are quick and readily available but don’t serve the 24/7 needs for data centers. Nuclear is a solution, but a long-term and high-cost solution at this point.

Hampton: Utilities’ other customers are residents like you and me. What does this surge in competition mean for residential service and monthly bills?

Miller: There are two different ways of thinking about this. All else equal, utility bills for everyone will go up if the utility has to invest in new infrastructure for new customers. However, data centers could bring in a huge amount of new demand that allows utilities to spread costs over a much larger number of kilowatt-hours, which would flow down to savings at the residential level. As these data centers come online and the utilities assess the amount of capital investment necessary, then the utilities will have to work with regulators in terms of who pays that ultimate cost and the rates for the different customer classes.

Hampton: What is your five-year outlook for the utilities sector?

Miller: It has been a couple of interesting years. Utilities are up almost 30% in 2024. However, people might forget that 2023 was one of the worst performances in several decades for utilities: They were down 7% and underperformed the market by 33 percentage points, a massive underperformance. Smoothing those returns out over the last three years, utilities have averaged an 8% total return. This is the same as the average annual return for utilities during the last 40 years. Even though we’ve had these two very different years—and very volatile years relative to how utilities typically perform— the total return over the last three years has been a pretty pedestrian 8%, just like what we think utilities investors should expect.

We think that utility returns will come back to that 8% type of range when you consider dividend yields around 3.5% and utilities’ growth potential. The key element is how much does this new growth from data centers, new power generation, and clean energy impact returns going forward? Will returns go up to 10% possibly because of the yields right now and additional earnings growth? We think the sector going into 2025 is about 10% overvalued. Earnings multiples are higher than the 20-year average. We think the market is pricing in a little bit too much growth. That said, many utilities just raised their capital investment plans during the third quarter earnings season in November. We do think there will be higher growth than in the past, but we just think investors should get ready for an 8% type of total return, not the 30% that investors saw in 2024.

Hampton: What caused the volatility in the sector?

Miller: Fundamentally, a lot of it was interest rates. The market was anticipating higher interest rates in 2023. That hurt high-yield type of investments. Utilities typically are thought of as bond alternatives, and certainly many investors buy utilities for their yield. That wasn’t a surprise that utilities would underperform while the market was expecting higher rates.

That mentality flipped as inflation came down. The market in 2024 started anticipating the Federal Reserve pausing or cutting interest rates. That helps high-yield sectors like utilities. I think 2024 was really the year when the market latched on to utilities as an AI play, and that certainly boosted them relative to the typical impact from interest rates on a normalized basis.

Hampton: What’s the prediction for 2025?

Miller: Interest rates will still be a factor. We’ll have to continue watching what the Fed decides and the trajectory of interest rates. But I do think we’re going to see a lot of shakeout related to utilities’ growth plans. Every utility is talking about growing at 6%, 7%—some are even talking about 8% earnings growth over the next three to five years. I think it will be interesting in 2025 to see which utilities start growing at that rate and can support their capital investment plans that would drive that type of earnings growth. I think there could be some utilities that can’t reach some of those high growth rates and start pulling back expectations. That would definitely hit valuations for some.

Hampton: Any names Morningstar considers undervalued?

Miller: There aren’t any screaming buys right now after the sector’s big run in 2024. We do like Duke Energy DUK. It’s a large-cap company based in North Carolina, serving Florida and several other states. We expect earnings to grow at the high end of management’s 5% to 7% range. That’s one of the sector’s top growth rates now. It trades at a discounted valuation to other utilities with similar growth potential. The capital investment plan that they’ve put out, $73 billion over the next five years, is one of the largest in the sector. A lot of it is going to invest in clean energy, which has a lot of support from the state governments where they serve customers. The discussion about the potential pullback in federal clean energy mandates and incentives—we don’t think that’s going to impact Duke’s clean energy investment program very much, if at all.

We also like Evergy EVRG. It’s a mid-cap utility based in Missouri and Kansas. This is one of the few utilities trading below fair value, about a 5% discount to our fair value estimate. One of the reasons is that management continues to target a lower growth rate in their outlook than most other utilities. But we think that they can get to the high end of management’s 4% to 6% earnings growth range over the next three years, in line with other utilities. The market has really stuck on that lower growth relative to most other utilities, and thus the stock trades at a discount to its peers. Evergy is one of the few utilities where you get a dividend yield over 4%. They just raised their capital investment program over the next five years by 30% in November, and a lot of that is to help serve some large customers that are coming in, notably Google, Meta META, and Panasonic PCRFF, to build data centers and EV battery manufacturing.

This story first ran in Q1 2025 issue of Morningstar Magazine.