Nuclear power: to the children of the 80s, it carried the weight of Cold War tension and cautionary tales.
But today, the narrative is shifting. For many, nuclear is increasingly seen as a solution to tomorrow’s energy challenges. And for data centre operators, it’s a reason to get excited, a path to securing a reliable supply of clean energy.
But this isn’t about towering cooling units or radioactive waste; it’s about something smaller, safer and smarter: small modular reactors, or SMRs.
With global energy demands set to exceed 1,000 terawatt hours by 2026 – the equivalent of Japan’s total electricity consumption – these compact, factory-built systems are designed to bring nuclear power closer to where it’s needed most, and may just help elevate the rapidly growing power demands for digital infrastructure.
So what exactly is an SMR?
According to the International Atomic Energy Agency, SMRs are reactors that have a power capacity of up to 300MW(e) per unit, enough to support the energy needs of most data centres.
But despite the name, SMRs aren’t exactly ‘small’ in the everyday sense. While they’re more compact than traditional reactors, often fitting within the footprint of a few football fields, they’re still sizeable installations requiring specialised infrastructure and regulatory oversight.
Think less ‘plug-and-play battery pack’ and more ‘scaled-down power station built with portability and repeatability in mind’.
As they’re still not the kind of asset you would therefore bolt onto the back of a data hall, SMRs are instead more likely to be located nearby – close enough to deliver dedicated power via private infrastructure, but with enough distance to satisfy safety regulations and zoning requirements.
Think of SMRs like a private substation, only built to nuclear standards and capable of offering far more energy.
Daniel Golding, CTO of Appleby Strategy Group and a former Google engineering executive, doesn’t mince his words when it comes to the scale of the data centre energy challenge.
“We need the data centre stream to have more power than is available on the grid today,” he says. “In terms of low- or no-carbon solutions, there are only a couple of options.”
In Golding’s view, current energy options like renewables and natural gas work, but aren’t perfect. Wind and solar remain intermittent, with energy storage tech still lagging behind.
It’s here that SMRs come into play. “The longer-term solution is going to be these small modular reactors,” Golding says. “Because it’s zero-carbon, it’s safe. We need to get the scale to the point where the price comes down.”
In his view, that means building 20 to 30 units before production costs start to resemble anything competitive.
Critically, Golding sees SMRs as a viable base-load energy solution that can flex with the demands of data centre workloads – a characteristic most traditional sources struggle to match.
“They allow us to really follow the load as we need to,” he says, noting that nuclear power – unlike gas – can adjust output relatively quickly, especially when paired with renewables.
But for all the technical promise, he’s quick to point out the main obstacle: “Speed to licensing. That’s it. Speed to licensing, cost per unit. There’s really nothing else.”
Solving nuclear’s old problem for a new era
For Ed McGinnis, former assistant secretary for nuclear energy at the US Department of Energy and now CEO of Curio Energy, the path to successful SMR deployment lies not in technology or economics, but in what happens after the power is produced.
“US SMRs are well-positioned to meet the expected increased electricity demand,” he says, “provided they are deployed as part of a holistic, sustainable nuclear-recycling and closed-fuel-cycle approach.”
In other words, the viability of SMRs isn’t just about kilowatts and capex; it’s about addressing the elephant in the room of nuclear waste. Curio’s answer is NuCycle, a next-generation recycling technology designed to overhaul how nuclear waste is processed.
“It’s a step-change improvement,” McGinnis says. Unlike the traditional PUREX method, which uses nitric acid and extracts pure streams of weapons usable plutonium, NuCycle is a largely dry, modular process based on pyro processing and electrolysis.
Not only does it shrink high-level waste volumes to just 3 to 4% of their original amount, it also removes key security concerns and makes resource recovery economically viable.
“The single biggest barrier to deploying SMRs is the lack of a plan to deal with so-called ‘nuclear waste’,” McGinnis adds.
With support from four US Department of Energy national labs and a network of industry partners, including utilities, fuel-cycle companies and reactor vendors, Curio is positioning itself as a key enabler of a closed-loop nuclear economy.
But the company isn’t just innovating behind the scenes, with McGinnis stressing that community engagement is central to Curio’s model: “We engage very openly, transparently and sincerely. We place a high priority on local communities and long-term stakeholders.”
Looking ahead, McGinnis sees a sustainable, closed-fuel-cycle future as not only technically feasible, but necessary for the digital era.
“The long-term vision for nuclear energy must be predicated on a holistic approach,” he says, especially as AI and high-density workloads accelerate global power demand.
Building the blueprint: Meet the firm bringing SMRs to market
There’s no shortage of companies and startups looking to capitalise on rising interest in SMRs. But one firm that’s further along the path to deployment is ARC Clean Technology, which is already deep into the execution phase.
With roots in a legacy of proven technology and its sights set on large-scale deployment, ARC believes it has the blueprint to take small modular reactors from prototype to grid-connected reality.
“We’re not an R&D company any more – we know what we want to build,” says Lance Clarke, ARC’s VP of commercialisation and strategy. “We’re truly at a commercialisation stage.”
At the centre of ARC’s proposition is the ARC-100, a 100MW sodium-cooled fast reactor based on the Experimental Breeder Reactor-II (EBR-II), which operated successfully for 30 years at Argonne National Laboratory.
Unlike many next-gen designs still seeking validation, ARC’s approach leans into proven technology.
“That’s why we chose it,” says Clarke of the company’s sodium-cooled metallic fuel system. “It’s inherently safe and easier to build, and that makes it more economic and socially acceptable.”
That safety isn’t just theoretical. The ARC-100 is designed to shut itself down passively – using physics, not pumps or operator intervention – a core feature refined through decades of operation in naval and research reactors.
Though ARC has been quietly building towards this moment for years, Clarke says the market is now catching up.
“The sector wasn’t ready for this 20 years ago, but that’s changed: there’s real momentum now, especially with the need for clean base-load power to support high growth sectors like AI and data centres.”
The ARC-100’s design directly addresses those sectors. It’s modular and factory-built, yet powerful enough to support large workloads and flexible enough to pair with intermittent renewable sources.
Regulatory progress is already underway. ARC is nearing a milestone with the Canadian Nuclear Safety Commission and preparing for its first deployment at New Brunswick’s Point Lepreau site. It’s also engaging with US regulators and advancing international partnerships, including one with Korea Hydro & Nuclear Power aimed at accelerating fleet deployment
But even for advanced players like ARC, challenges remain. One is fuel: the global supply of HALEU (high-assay low-enriched Uranium), which is critical for many SMRs, is still heavily concentrated in Russia.
For ARC, however, the bigger near-term hurdle is regulation. “Most of the deployment timelines are driven by the regulatory processes,” Clarke says.
“We estimate our first reactor will be built in three years, with future builds getting down to just two. But we need to start those processes now.”
In the data centre space, ARC sees a perfect alignment of demand and capability. Clarke points out that hyperscale operators aren’t just energy-hungry; they’re also financially well-positioned and more open to emerging technologies.
“We’re seeing a lot of new companies pop up – developers that want to use technologies like ours to sell clean energy to data centres. That’s a clear signal.”
Still, deployment at scale will take time. Critics often cite nuclear’s long lead times, but in some markets, grid connection alone can take up to eight years, making SMRs a timeline-competitive option in the eyes of some operators.
Clarke sees material deployment by the mid-2030s, with mainstream adoption in the 2040s.
“I know that sounds like a long way off,” he says, “but it really isn’t. The key is to act now – start siting, start engaging communities, start aligning with regulators. Because every year you delay, you’re just pushing that horizon further away.”
As for public opposition? Clarke acknowledges it, but believes transparency is the remedy.
“This is statistically the safest energy sector there is. Generation IV reactors like ours only improve that. We need to help people understand the data, not the myths.”
With regulatory traction, global partnerships and commercial conversations in progress, ARC believes the road ahead is clear.
“Make the first few successful,” Clarke concludes, “and the rest will follow.”
RELATED STORIES
Is nuclear the data centre energy saviour?
Data centre power costs, the PPA market and a move to 24/7 power matching
