Data centers are betting billions on small nuclear reactors, but experts say the technology may not arrive in time and costs remain far above wind and solar

Data centers are betting billions on small nuclear reactors, but experts warn the technology may not arrive in time

Data centers are facing a genuine energy crisis: surging demand from AI workloads, growing public backlash against gas turbines, and pressure to find power sources that are both reliable and low-carbon. Small modular nuclear reactors have emerged as the industry’s preferred answer — compact enough to fit near a facility, firm enough to run around the clock, and clean enough to satisfy climate commitments.

But whether that answer actually holds up is a different question.

Then there’s NuScale itself — Adam Stein of the Breakthrough Institute points out the company has been around for 15 years without landing a solid customer.

A power-hungry industry looking for a cleaner fix

The pressure on data centers isn’t abstract. Elon Musk’s xAI facilities in Mississippi and Tennessee were running 27 unpermitted gas turbines when a coalition of environmental groups filed a Clean Air Act lawsuit against them. In Texas, Governor Greg Abbott paused new data center approvals in early August over energy and water concerns. The message from regulators and communities is getting harder to ignore.

SMRs check several boxes that gas turbines don’t. They’re compact — multiple reactors can fit within a couple of buildings — leaving room for the data center itself to expand. They run continuously, which matters for facilities that can’t tolerate interruptions. The preferred deployment model is co-location: reactors sited near the facility, still connected to the grid, but reducing the need for new transmission infrastructure. Their emissions profile also aligns with the climate commitments many tech companies have made publicly.

What small modular reactors actually are — and where they stand

The U.S. Energy Information Administration defines SMRs as reactors generating between 70 and 350 megawatts — well below the roughly 1,000 MW output of conventional large reactors. According to the think tank Third Way, 22 SMR designs are currently under development in the United States, though only two — both from NuScale Power Corporation — have received approval from the Nuclear Regulatory Commission.

As of mid-2026, not a single commercial SMR is fully operational in the country. That gap between design approval and actual deployment is where the story gets complicated. Regulatory clearance is a milestone, not a finish line — construction, supply chains, and grid integration all follow, and none of them move quickly.

The cost problem that won’t go away

The clearest real-world cost signal comes from Canada. Ontario’s Darlington Project is projecting construction of four BWRX-300 SMRs at $140 per megawatt-hour for the first reactor, with subsequent units coming in around $80 per megawatt-hour. Wind and solar’s levelized cost of electricity, by comparison, ranges from under $40 to $86 per megawatt-hour, according to Lazard data.

NuScale’s 2023 collapse of its Utah project — 12 modules, 600 MW — shows what happens when costs keep climbing. Edwin Lyman, director of nuclear power safety at the Union of Concerned Scientists, is direct: “They’ve been promising for years now that they have deals right around the corner, and nothing happens. It’s not clear they’ve done anything to improve the economic viability of their reactors.”

Why nuclear startups can’t move at Silicon Valley speed

Tech culture celebrates fast iteration. Nuclear development structurally can’t offer it. As Patrick White of the Clean Air Task Force put it: “You can’t just bang out a nuclear reactor in your garage and put it out on the street after six weeks.” Design, testing, and approval cycles operate on timelines that have no equivalent in software.

Supply chain constraints compound the problem. Victor Ibarra Jr. of the Clean Air Task Force notes that bottlenecks across the manufacturing supply chain are creating real obstacles to commercialization. Then there’s NuScale itself — Adam Stein of the Breakthrough Institute points out the company has been around for 15 years without landing a solid customer. That’s not a startup stumble. It’s a structural signal.

Safety concerns that regulators haven’t yet resolved

Speed and safety don’t mix well in nuclear development. Lyman argues the industry’s push to fast-track deployment is genuinely dangerous — the NRC has not yet validated that proposed SMR designs meet the safety threshold needed to grant developers the benefit of the doubt. Justifications for reduced safety features rest on hypothetical, unproven reactor configurations, not demonstrated performance.

Co-location with data centers adds another layer of concern. Lyman calls it an unrealistic model, citing the potential consequences of a radiological release near a densely operated facility. Financial pressure to deploy quickly may push developers toward cutting costs on safety and security — consequences that, unlike a buggy software release, can’t simply be patched.

State interest is real — but the timeline isn’t

Political momentum is building. New Jersey Governor Mikie Sherrill signed legislation in July to invest in advanced nuclear energy, and NuScale announced a partnership with the Tennessee Valley Authority for a 6 GW SMR deployment program, though specifics remain scarce.

Interest isn’t capacity, though. Jon Gordon of Advanced Energy United says SMRs are unlikely to meaningfully contribute to near-term grid capacity in regions like PJM Interconnection. Brad Tietz of the Data Center Coalition put it plainly: “It’s a huge opportunity, I think, in the future. We’re just not there yet.” The energy crisis facing data centers is immediate. The solution being promoted most loudly is years — possibly a decade — away from meaningful scale.

Daniel Garcia

Daniel García is an Editor-in-Chief with strong expertise in structural work and engineering principles. He combines this technical foundation with deep knowledge of energy, spatial design, and emerging technologies, bringing a forward-thinking and analytical approach to editorial leadership.