Big Tech’s race for nuclear energy

Unlock Editor’s Digest for free

Roula Khalaf, editor-in-chief of the FT, selects her favorite stories in this weekly newsletter.

Big Tech is going nuclear. Last week, Amazon struck a deal with Washington state utilities to support the development of four next-generation “small modular reactors,” with a similar deal in Virginia, and took a stake in X-energy , a developer of SMR. Google has agreed to buy electricity from SMRs to be built by a start-up, Kairos Power. And last month, Microsoft agreed to a 20-year power purchase agreement that will involve Constellation Energy reopening a unit at the Three Mile Island plant in Pennsylvania that was closed in 2019 (not the one closed in 1979 after partial fusion).

The rise of the technology industry towards nuclear power partly reflects the take-off of energy-hungry artificial intelligence; an AI query consumes up to 10 times more energy than a standard Google search. Goldman Sachs estimated data center electricity demand will increase 160% by 2030. In the United States, data needs, in addition to the electrification of transportation and a manufacturing recovery triggered by efforts of “reshoring”, are expected to at least double the growth in electricity demand over the next decade compared to the previous decade.

In Europe, Goldman estimates that electricity demand could increase by 40% between 2023 and 2033. The International Energy Agency said last week that after the age of coal and the age of oil, the world is entering in a new phase of transition. the age of electricity.

Tech companies know that to build data centers in countries like the United States, they will have to manage a lot of their own energy. Their net-zero commitments require sources to be green, and they have already invested heavily in wind and solar. Expanding their portfolios to nuclear power is understandable, but it’s a gamble.

Nuclear power in principle has a strong claim to being part of the climate solution. It’s low carbon, provides plenty of power for decades, and doesn’t fade when wind or sunlight does. The problem is that building large power plants is extremely expensive and time consuming.

SMRs – reactors of up to 300 megawatts, compared to 1,000 MW for large nuclear power plants – claim to offer a cheaper and faster alternative. Largely prefabricated in cookie-cutter designs, their small size means in theory they can be installed close to where electricity is needed and at sites such as old coal-fired power stations that are already plugged in to the network.

But they could face costs just as high as larger units to get their designs approved by regulators, in an industry where safety is paramount. They could divert critical investments away from proven solar, wind and battery power systems. SMRs also remain to be proven. Based on the three SMR type projects in operation and a fourth under construction, the Institute of Energy Economics and Financial Analysis calls them “still too expensive, too slow and too risky”.

Bringing big tech’s financial influence and innovation could help drive the development of SMRs and accelerate the shift from largely government-led and funded nuclear development to private funding and initiative (watch this what Elon Musk did to the space economy). But finding ways to reopen or extend the life of existing nuclear power plants could prove more feasible; In addition to Three Mile Island, a Michigan plant is being brought back into service.

Either way, increasing demand for AI-driven data before 2030 means big tech will likely need to invest even more in wind and solar. Amid competition for resources, regulators will need to ensure that deep-pocketed technology companies do not block much of the new energy supply. One option could be to insist that clean energy projects for data centers be large enough to also supply the grid or other customers. It is also possible to use AI to improve energy efficiency in factories, offices and on networks. In the new era of electricity, AI must not be just a mouth hungry for energy to feed, but a central part of the green solution.