Data Centers Go Nuclear: Why AI Giants are Investing in SMRs

 

 

The rapidly increasing power demand of data centers, driven by the AI boom, has become a stubborn constraint on the speed of their expansion. Small modular reactors (SMRs), which are smaller, partially factory-built nuclear fission reactors, are frequently proposed as the ideal solution for powering data center facilities. Offering clean, continuous, and reliable power wherever needed, it would seem that SMRs hold the key to solving AI energy demands - but how realistic is the prospect of this new nuclear renaissance?

 

IDTechEx analysts have closely followed trends in both data center infrastructure and emerging energy technologies, and the newest "Nuclear Small Modular Reactors (SMRs) Market 2026-2046: Technologies, Players, Benchmarking, Forecasts" report highlights the progress of the most promising SMR projects, with a data-driven benchmarking scheme for 10 different reactor types and 20-year forecasts for the SMR market split by technology and region. SMRs have the potential to become an integral part of the global electricity production mix for data centers and the grid at large, with forecasts that SMRs could reach over 1900 TWh of yearly electricity production by 2046.

 

 

 

 

 

The core value proposition of SMRs revolves around the fact that by making nuclear power projects smaller, they become far less likely to encounter serious cost or construction time overruns as is common in the traditional nuclear industry. Their smaller power output (usually below 300MWe) and footprint also means they can be delivered in a more flexible range of locations, delivering electricity much closer to the data center where power is needed and coming online more quickly than larger energy infrastructure.

 

For the SMR market, economies of scale are transferred from the size of the individual plant (as in a conventional large nuclear project), to manufacturing higher volumes of individual SMRs. Series production of SMRs has the potential to offer highly competitive LCOEs, especially when considering that nuclear power is available continuously unlike intermittent renewables - a fundamental requirement for data centers.

 

The reality however is that the first-of-a-kind (FOAK) of an SMR design will likely be expensive. Most SMR developers agree that overcoming this inertia to commit to production of at least 6-10 units of the same reactor design would unlock cost reductions of up to 40% compared to the FOAK. The evidence for this is in similar learning curves demonstrated by manufacture of small nuclear reactors for submarines.

 

 

 

Data center hyperscalers like Microsoft, Amazon, Google and Meta are some of the few private organizations with deep enough pockets to fund not only single data centers but entire fleet projects. In 2026, these deals are already taking shape, with each hyperscaler making commitments with different nuclear power developers.

 

SMR designs can be categorized by their different reactor technologies, with three of the most promising concepts being High-Temperature Gas-cooled Reactors (HTGRs), Liquid Meta Fast Reactors (LMFRs), and Molten Salt Reactors (MSRs). It is surely no coincidence then that Google, Amazon, and Meta have each made major commitments with SMR startups each developing a different type of reactor technology: Google with MSR startup Kairos Power, Amazon with HTGR developer X-Energy, and Meta with two different LMFR startups - Oklo and TerraPower.

 

HTGRs, LMFRs, and MSRs are referred to as fourth generation (Gen IV) nuclear reactor designs. In thefull SMR Market report, IDTechEx compares these reactor types in a quantative benchmarking scheme. The benchmark draws on metrics like industry interest, safety, and efficiency to compare these Gen IV types to each other and also to more established Gen III+ nuclear reactor technologies like pressurized water reactors (PWRs) and boiling water reactors (BWRs).

 

 

 

 

 

Before getting too excited about the prospect of nuclear-powered data centers, it's important to take a look at the timelines involved. Even the most prolific SMR startups like Kairos Power, X-Energy, and TerraPower anticipate that it will take until around 2030 for their first fully functional SMR to start supplying electricity. Even if delays are avoided, note that these goals represent the first-of-a-kind (FOAK) deployment. A larger rollout to get to gigawatts of power will take at least a few more years.

 

In comparison, the data center industry anticipates that significant electricity shortages could hinder development as early as 2028. To put it simply, SMRs will not be the solution for imminent data center power shortages in the near-term. However, this does not mean that they could not become a critical piece of infrastructure in the long term.

 

While gas plants will be the primary source to cover the energy shortfall in the next few years, recent global events have underscored how volatile oil and gas markets can be. Nuclear SMRs offer a flexible, scalable solution for low-carbon data center power with fuel that can be readily stockpiled.

 

 

Data center hyperscalers have become organizations with capital rivalling small nations. IDTechEx's Nuclear SMRs report investigates how governments and data center giants are investing in small modular reactors to future-proof their electricity demand, in a market that could reach US$53.8 billion in 2036 and almost US$300 billion by 2046.

 

 

For more information on this report, including downloadable sample pages, please visit www.IDTechEx.com/SMRs. For the full portfolio of energy and decarbonization research available from IDTechEx, please visit www.IDTechEx.com/Research/Energy, and for further research relating to data centers visit www.IDTechEx.com/Research/DataCenters.