How the rise of data centres is leading to a nuclear renaissance

“Small is beautiful”, proclaimed the classic 1973 book by British economist Ernst Schumacher, challenging the century’s mania for gigantic projects. He opposed nuclear power, among the largest industrial endeavours of his time, multibillion-dollar assemblages of concrete, steel, wires and uranium. But the emerging nuclear power renaissance, driven by data centres, combines the small and the large.

On Wednesday, e-commerce and web services giant Amazon said it would be an anchor investor in a $500 million fund-raising by X-energy, a new developer of small modular nuclear reactors (SMRs). Amazon also said it would support SMR projects in its home state, Washington, as well as in the data centre hotspot of Virginia.

Amazon and X-energy intend to have more than 5 gigawatts of SMRs operational by 2039. This is approximately equal to the 5.6 gigawatts of conventional large nuclear reactors at the UAE’s Barakah plant.

Last week, tech rival Google also ordered six to seven SMRs from Kairos Power, while last month, Oracle said it would use three SMRs to power a more than 1 gigawatt centre and meet “crazy” needs for power. Microsoft had announced it would buy electricity from the infamous Three Mile Island nuclear power plant, site of a 1979 accident, if its owner restarts it.

These are welcome steps forward for nuclear power. Outside a few countries such as China and the UAE, nuclear capacity has been going backwards for years, as ageing reactors were closed down and not replaced, new plants took decades to build and run heavily over-budget, and countries such as Germany phased-out operational sites.

The improvement in renewable energy such as wind and solar, sharp falls in costs for battery storage and, in the US, a glut of cheap shale gas, made nuclear power economically uncompetitive. Environmentalists, often adhering to 1970s-era orthodoxies and fears of nuclear accidents such as Chernobyl in 1986, campaigned heavily against new nuclear investment, and overregulation and legal challenges drove up construction times and costs.

But three factors may create a more radiant future.

First, climate. At the Cop28 talks in Dubai last year, a group of more than 20 countries, including the UAE, US and UK, affirmed a goal of tripling nuclear capacity by 2050 as a source of reliable low-carbon electricity.

Second, energy security. The Russian invasion of Ukraine and the cut-off of much of Europe’s gas made the continent, and other isolated energy markets such as Japan, South Korea and Taiwan, aware again of the value of power generation which is not affected by weather and whose fuel can be stockpiled for years. But Western countries and allies want to steer clear of Chinese or Russian reactors and fuel, so they need to rebuild decades of atrophied domestic capability.

Third, electricity demand is rising again in developed countries, after decades when it barely grew. Demand for electric heating, air-conditioning and battery cars is one component.

The explosive rise of data centres is another, driven by the artificial intelligence boom. Even if the overall electrical needs of data centres are not huge, they are very significant in specific areas, far-exceeding local capacity in areas such as Virginia. Meeting this need with renewables is difficult as the best solar and wind sites are distant and constructing new transmission cables across state borders is a regulatory thornbush.

But to answer these needs, new nuclear plants need to be much quicker and cheaper to build. The International Energy Agency estimates that nuclear electricity in China costs 6.5 US cents per kilowatt-hour, which is cheaper than gas, and reasonably competitive with large-scale solar or wind power. China builds numerous plants in sequence and has managed to standardise them and train up its workforce.

But the cost in the US is 10.5 cents, and in Europe, 14 cents. New reactors are rarely built, face endless public and legal challenges, excessive and often capricious regulation, and lack of expertise from developers whose last serious construction programmes were in the 1970s or 1980s.

SMRs promise the needed improvement. A conventional nuclear reactor may typically be about 1,000-1,400 megawatts. SMR designs, by contrast, range from a few megawatts, designed for remote communities, isolated industries or military sites, or ships, to Rolls-Royce’s 470MW unit, really a medium-sized reactor. X-energy’s system features 80-megawatt reactors which can be bundled into a “four-pack”.

SMRs cover a wide range of designs, from variants of traditional models, to radical new concepts. They are often intended to be inherently safer than conventional reactors, not requiring external cooling, the problem that hit Japan’s Fukushima plant in 2011 when its backup diesel generators were swamped by a tsunami.

Their biggest selling point is that they should be quicker and, eventually, cheaper to build than large conventional reactors. Many of their components will be fabricated in assembly-line fashion, gaining manufacturing experience and lowering costs by standardisation. On-site construction and changes of plan, the bane of many new nuclear sites, will be minimised. Financial exposure will be less, lowering the risk and cost of capital.

Unlocking the promise of SMRs needs deep-pocketed, long-term and risk-tolerant investors. After several false turns and doors barred firmly by engineering, finance or regulation, the industry may finally have found its key in cash-rich Amazon, Google and Microsoft.

Some Middle East countries are also awake to the promise of SMRs, as their net-zero carbon and data centre ambitions grow. In December, the Emirates Nuclear Energy Corporation signed co-operation agreements with X-energy and three other SMR developers. Saudi Arabia’s King Fahd University of Petroleum and Minerals is working on its own SMR design, and the kingdom is co-operating with South Korea on its Smart reactor.

SMRs are still a tough sell in the Gulf, even with its booming electricity needs, given the abundance of land for cheap solar power backed up with batteries. Yet the tech giants’ commitments demonstrate confidence as well as urgency to meet vast energy projections. Artificial intelligence may be the parent to beautiful small reactors.

Robin M Mills is chief executive of Qamar Energy, and author of 'The Myth of the Oil Crisis'