Google agreed to buy power from SMRs to be built by a start-up, Kairos Power.
Goldman Sachs reckons power demand from data centres will grow 160 per cent by 2030.
In Europe, Goldman estimates power demand could grow by 40 per cent from 2023 to 2033.
Expanding their portfolios to nuclear energy is understandable, but something of a gamble.
SMRs — reactors up to 300 megawatts, compared with 1,000MW for large nuclear plants — claim to offer a cheaper, faster alternative.
Big Tech is losing its mind. In the last week, Amazon acquired stock in SMR developer X-energy and reached an agreement with utilities in Washington state to support the construction of four next-generation “small modular reactors.” A similar agreement was reached in Virginia. Google consented to purchase electricity from SMRs developing by Kairos Power, a start-up. Additionally, Microsoft last month agreed to a 20-year power purchasing agreement that calls for Constellation Energy to reopen a unit at the Three Mile Island plant in Pennsylvania that was shut down in 2019 (not the one that closed in 1979 due to a partial meltdown).
The rise of power-hungry artificial intelligence, which can use up to ten times as much energy as a typical Google search, is partly responsible for the tech industry’s rush for nuclear. Data center power demand is expected to increase by 160% by 2030, according to Goldman Sachs. In the US, it is anticipated that data requirements, transportation electrification, and a manufacturing renaissance spurred by “reshoring” initiatives will at least double the growth in electricity demand over the next ten years.
Power demand in Europe is expected to increase by 40% between 2023 and 2033, according to Goldman. Last week, the International Energy Agency announced that the world was moving into the age of electricity, following the eras of coal and oil.
Tech companies are aware that they will need to organize a large portion of their own power in order to successfully build data centers in nations like the US. They have already made significant investments in solar and wind power, which is required by their net zero pledges. Although it makes sense, diversifying into nuclear energy is a bit of a risk.
According to theory, nuclear has a good chance of contributing to the climate solution. It emits little carbon dioxide, provides a lot of power for decades, and remains stable in the face of wind and sunlight. The issue is that building large plants is extremely expensive and time-consuming.
SMRs—reactors with a maximum power of 300 megawatts, as opposed to 1,000MW for large nuclear plants—professe to provide a quicker and less expensive option. Theoretically, because of their small size and large prefabricated, cookie-cutter designs, they can be placed on locations like abandoned coal plants that are already connected to the grid and close to areas where power is needed.
However, in an industry where safety is crucial, they might have to pay just as much as larger units to have their designs accepted by authorities. They might take vital funds away from tested wind, solar, and battery power systems. SMRs haven’t been proven either. According to the Institute for Energy Economics and Financial Analysis, SMR-style projects are “still too expensive, too slow, and too risky” based on the three that are currently underway and a fourth that is being constructed.
Bringing in Big Tech’s financial might and creative flair could help SMRs grow and hasten the transition from nuclear development that is primarily led and funded by the government to private funding and initiative (just take a look at what Elon Musk has done to the economics of space). In addition to Three Mile Island, a plant in Michigan that is being recommissioned, it may be more practical to find methods to reopen or prolong the life of already-existing nuclear stations.
In any case, Big Tech will most likely need to make even more investments in wind and solar given the spike in demand for AI-driven data even before 2030. Regulators must make sure that wealthy tech firms do not dominate the new energy supply in the midst of the competition for resources. Demanding that clean energy projects for data centers be sufficiently large to supply the grid or other clients could be one way to go. Additionally, AI has the potential to increase energy efficiency across grids and in offices and factories. AI must play a key role in the green solution in the new electrical era rather than being just another energy-hungry mouth to feed.
