The data center sector is walking a tightrope between ever-increasing power demands and its commitment to decarbonizing. Building out and securing low-carbon power sources is becoming increasingly imperative for the sector, fueling a renewed interest in nuclear energy, especially small modular reactors (SMRs).

For proponents, SMRs offer the perfect solution to data center needs, namely, consistent low-carbon baseload power. They also provide high integration potential due to their modular nature and ability to be deployed quickly in various locations, independent of any external power sources or grid connections.

However, significant concerns remain over whether the technology can be successfully scaled for use. Critics would point to the fact that SMRs remain firmly in the demonstration phase, and with countless proposals out there, it remains to be seen which of the SMR developers - if any - will produce a commercially viable product.

In 2024, several data center operators took the plunge and partnered with SMR vendors, begging the question: Are we on the brink of a nuclear revolution in the data center sector?

What are SMRs?

The International Atomic Energy Agency defines SMRs as small power reactors with lower outputs ranging from less than (up to) 10MW, known as microreactors, to a standardized capacity of 300MW.

SMRs are designed to be portable and can be shop-fabricated and transported as modules, allowing for on-site installation. Their smaller footprints and flexible deployment make them suitable for regional or industrial clusters. SMRs are designed to operate for long periods of time before refueling, with some lasting up to 30 years.

There are various types of novel reactor concepts. The initial generation I and II reactors were developed by the US military in the 1950s. Of current designs, Gen III pressurized water reactors are the most common, and operate as miniaturized traditional nuclear plants.

Recently there has been a proliferation of Gen IV concepts, which hold promise of much higher efficiency through alternative cooling methods, including gas-cooled, liquid metal-cooled, and molten salt designs. However, these concepts have little to no real-world industrial experience.

Hyperscalers on board!

The commercialization of SMRs will depend on overcoming significant hurdles, including financial challenges. Construction timelines proposed for most SMRs span the mid-2030s, with considerable delivery uncertainty and cash flow risks deterring traditional debt financing. As a result, SMRs have historically relied on government funding, with the US, UK, and Canada all launching funding rounds in recent years to support domestic SMR development.

February 2024 saw the first privately funded SMR agreement between Westinghouse and Community Nuclear Power to deploy four SMRs in North Teesside, UK. This milestone was followed closely by a series of commitments from the hyperscalers, with Google, AWS, and Oracle all inking long-term agreements with SMR providers to power their operations.

For Ivan Pavlovic, executive director, energy transition at investment bank Natixis, these agreements could support the sector similarly to how renewables were backed through state subsidies.

"Just as renewable energy benefited from feed-in tariffs and green certificates, SMRs may rely on private contracts with large off-takers to support early development, mimicking the conditions of renewable financing," Pavlovic says.

The hyperscaler agreements all represent long-term commitments to the sector into the 2040s. Google, for example, has penned a 20-year master plant development agreement with Kairos Power, a molten salt-cooled Gen IV SMR.

The long-term commitment was crucial for Kairos, said Mike Laufer, the firm's CEO, as it met the two “main challenges” for financing nuclear projects - "the long time scales involved and the need for financial backing to cover the development period, even under an aggressive timeline like 2030–2035 for our initial deployment."

The commitment of these companies marks a significant step toward the commercialization of SMRs. Acting as early adopters, and in AWS's case, investing directly into the SMR firm in the form of X-Energy, developers are provided with the financial stability and long-term agreements necessary to substantially de-risk SMR projects, which could pave the way for broader market acceptance. The agreements also serve to temper concerns about rushed deployment, which can inflate costs and cause delays.

A mutually beneficial arrangement

For Pavlovic, data centers are likely to be the “best possible off-takers” for nuclear energy. “They need low-carbon, 24/7 electricity, and strong financial balance sheets to support long-term contracts,” he says.

SMRs have exceptionally high capacity factors, which is a measure of how often a power plant operates at maximum power, and how consistently it produces energy over time. This is shown in both Nuscale’s and Rolls Royce’s SMRs, which have both registered a capacity factor of 95 percent or more, increasing their attractiveness to data centers. In addition, with most projects ranging between 60MW and 300MW, they offer large amounts of clean, consistent power, free of issues of intermittency and curtailment, as seen in solar and wind.

"SMRs can be placed almost anywhere and offer an exceptional capacity factor, a key measure of energy consistency, which surpasses even that of gas or coal," states James Walker, CEO of microreactor company Nano Nuclear. These attributes explain why the tech giants have begun to view nuclear as the preferred solution for reliable, low-carbon energy.

The relationship is likely to be mutually beneficial, with data center developers being one of the few industries with the capital and forward-thinking to take a risk on a yet-to-be-proven technology. Kairos Power's Laufer emphasizes that partnerships with hyperscalers not only provide financial security, but also facilitate iterative learning and cost reductions. "The partnership with Google provides a strong alignment for both parties, enabling cost reductions and learning through the deployment of multiple reactors of the same kind,” he says.

This allows companies such as Kairos to build "something that is either exactly or very close" to their solution as part of a demonstration project, as seen in its Hermes Demonstration Reactor. Laufer adds that the more measured approach ensures a "true learner effect" that reduces costs before entering the capital-intensive construction phase. This method contrasts with traditional nuclear, which often bypasses smaller-scale demonstrations, leading to cost overruns and delays. In turn, this has prevented many companies from backing SMRs.

Rolls Royce SMR has also adopted this measured approach. According to Harry Keeling, the company’s head of development of new markets, Rolls Royce's "approach gives customers certainty that when we commit to timelines, we will be able to deliver, and this creates trust among investors."

Subsequently, Keeling has contended that “in the next ten years, we are likely to see a consolidation around a few leading SMR technologies akin to Boeing and Airbus in aviation.”

This, in turn, could support the industry in achieving the fleet-level economies of volume necessary to serve the data center market. In doing so, SMR developers hope to meet one of their biggest challenges: creating cost certainty.

Cost certainty through modularity

The path to cost certainty involves careful, measured steps. Data center commitments have provided developers the flexibility to avoid hastened deployment. However, concerns remain over whether SMRs will be financially viable for widespread use.

Harnessing SMR modularity will play a central role in creating the cost certainty required. The modularity of SMRs compares favorably to renewable energy, especially solar, as they can be built in-house off-site and shipped as modules. Moving away from the single monolith model of traditional nuclear to the many small modules model will mean that economies of scale can be achieved during the component manufacturing process, reducing costs and easy scalability.

Keeling argues that modularity makes SMRs much more attractive to the financial community, as it "de-risk projects." Therefore, unlike traditional nuclear, where 50 percent of the energy cost comes from debt, "modularity allows a reduction of on-site construction risks, streamlines operations, significantly shortens project timelines, making nuclear power accessible to a wider range of customers."

Rolls Royce SMR has embraced this approach, constructing its entire power plant using standardized modular pieces. "Every plant uses the same 1,000 modular pieces, ensuring standardization and volume economies," he says.

Modularity not only reduces costs but also aligns well with data centers’ scalability needs. Clayton Scott, chief commercial officer at NuScale, another SMR developer, highlighted the appeal: "This modular approach provides data center operators with greater options when selecting the right size power plant to meet capacity and economic considerations."

In addition, as they are not location-dependent like many other clean energy systems, modular reactor deployment flexibility is much higher, making them more suited for scaling in Edge and remote installations that may lack good grid connections or clean energy access.

This has led some companies to focus explicitly on the data center market. Start-up Deep Atomic is doing that, seeking to offer a 60MW "behind-the-grid island power solution for data centers," according to a company representative. The small size will support greater deployment flexibility, with potential hybridization with energy storage systems and renewable to support the data center operator's drive to net zero.

In addition, by streamlining factory-built components and minimizing on-site risks, SMRs can control costs and timelines more effectively than traditional nuclear projects. Keeling emphasizes that from financing to licensing to operation, “standardization reduces uncertainty at every stage, making SMRs far more commercially viable than traditional nuclear projects."

For data centers, this provides a compelling case for adoption, as SMRs can deliver consistent, low-carbon energy tailored to their needs.

Challenges in financing and regulation

Despite growing private sector interest, SMRs face significant hurdles in achieving cost certainty and delivery reliability.

A report by Germany's Federal Office for the Safety of Nuclear Waste Management highlights the steep construction costs associated with SMRs. BASE found that achieving the same global output as today's large-scale nuclear plants would necessitate scaling SMR deployment by three to 1,000. This translates to constructing approximately 3,000 SMRs globally to make their production economically viable.

Critics have also argued that several SMRs have sold the market on inflated unit economics while grossly underestimating the time and capital it will take to commercialize their products.

In a recent report, Kerrisdale Capital claimed that SMR developer Oklo, which has signed several supply agreements within the data center sector including Equinix and Prometheus Hyperscale, is beyond optimistic in its timelines.

The firm is working towards submitting a license application in 2025, hoping for a first reactor deployment by late 2027. However, according to a former NRC Commissioner, its near-term projections are steeped in "hubris," as the company lacks the long-term supply of enriched uranium required for its reactor technology.

Given that Oklo has already signed several agreements with data center firms, the accusation that it is unlikely to meet its timelines weighs heavily, as it may impact the overall confidence that SMRs can be successfully commercialized.

Both Nano Nuclear and NuScale have also stoked controversy, with both firms facing charges from short seller Hunterbrook Media that their timelines are unrealistic and their products may not be able to live up to lofty claims. NuScale canceled a project in 2023, citing a lack of demand, which provoked further concern throughout the sector.

In addition, with most SMRs still in the concept exploration phase and more than 80 designs under development globally, there is still fundamental and inherent uncertainty about what designs can acquire regulatory approval and subsequently scale their solution to a commercial market.

However, SMR developers such as Rolls-Royce argue that the trick is to center product development around the regulation process.

"We've spent significant time with regulators in all our target countries. The feedback we receive is consistent: our reactor is 'boring,' and in nuclear, boring is the highest compliment," notes Keeling.

Developers also point to crossover between regulatory frameworks, which could expedite market licensing in new jurisdictions. This is more apparent with Gen III reactors, as they are already well understood. Gen IV reactors, on the other hand, are still highly experimental, and that may subsequently impact their timelines.

"While US regulatory approvals provide a strong foundation, each country has unique regulatory environments that require careful navigation for international market penetration," said Laufer.

Consequently, while challenges remain - particularly around cost certainty and regulatory hurdles - the SMR sector is positioned to benefit from its alignment with data centers' energy needs and modular design.

Partnerships between the data center sector and SMR developers could catalyze a new era of reliable, scalable, and low-carbon nuclear energy, addressing the data center sector's power demands and supporting the sustainable development of SMR technologies.