Google’s search for reliable clean energy for AI has moved into nuclear power. The company has announced an agreement with Kairos Power to purchase electricity from multiple small modular reactors, or SMRs, in a deal aimed at supporting the growing power needs of artificial intelligence.
The agreement is notable because it is described as the first deal of its kind. It is also a bet on a technology that still has a major hurdle ahead: no working SMR has yet been constructed in the US.
What Google and Kairos Power agreed to build
The partnership is designed around a staged rollout. Google and Kairos Power aim to bring Kairos Power’s initial SMR online by 2030, with additional reactor deployments planned through 2035.
If the plan works, the reactors could enable up to 500 MW of carbon-free power to be added to US electricity grids. For Google, that would support a wider push to align its fast-growing AI ambitions with cleaner electricity sources.
Google Senior Director of Energy and Climate Michael Terrell framed the agreement as a response to pressure on the grid. In a press statement, he said, “The grid needs new electricity sources to support AI technologies,” adding, “This agreement helps accelerate a new technology to meet energy needs cleanly and reliably and unlock the full potential of AI for everyone.”
The central idea is straightforward: AI systems require substantial computing infrastructure, and that infrastructure requires power. Google is trying to secure a source that is presented as clean, reliable and capable of supporting large-scale demand.
The technology still has to prove itself
The agreement depends on small modular reactors becoming real operating power sources. That is not guaranteed. Kairos Power has not yet created a working demonstration of the technology.
In July, Kairos broke ground on its “Hermes” non-powered demonstration reactor in Tennessee. That step followed a construction permit from the US Nuclear Regulatory Commission. Kairos expects to bring it online by 2027, which is slightly earlier than the date Google has cited.
That timing matters because Google’s energy plan is not only about buying power. It is also about helping accelerate a still-emerging technology. The deal could encourage development of small modular reactors, but the reactors still have to move from plan and demonstration toward practical operation.
The uncertainty is part of the larger story. Google is not simply signing up for an existing nuclear fleet. It is attaching part of its clean energy strategy to a next-generation nuclear approach that must still show it can work in the US.
Why AI is pushing Big Tech toward nuclear
Google is not the only technology company looking at nuclear energy for massive datacenters. In September, Ars reported on a Microsoft plan that would re-open the Three Mile Island nuclear power plant in Pennsylvania to meet some of its power needs.
The US administration is also moving in the same general direction. In July, it signed a bipartisan ADVANCE act with the aim of jump-starting new nuclear power technology.
The reason is the same across these moves: AI is increasing demand for electricity. Training and running power-hungry AI models has been criticized as wasteful, yet that same demand may also create new pressure to develop power sources that do not rely on fossil fuels.
That creates a complicated tradeoff. On one side, the rush into generative AI raises questions about energy use and whether all of that computing is necessary. On the other side, if demand from AI helps advance nuclear power, the result could support efforts to reduce reliance on fossil fuels and reduce greenhouse gas emissions.
The clean power question is also a control question
The Google-Kairos Power agreement can be read as a step toward next-generation nuclear power as a clean energy source, especially when compared to coal-fired power plants. Alongside solar and wind power, such collaborations may play a role as overall energy demand increases.
But the deal is also drawing skepticism. Some experts worry that the recent Big Tech interest in nuclear power could lead to clean power generation being controlled by the same companies driving the AI boom.
Dr. Sasha Luccioni, Climate and AI Lead at Hugging Face, wrote on X, “One step closer to a world of private nuclear power plants controlled by Big Tech to power the generative AI boom. Instead of rethinking the way we build and deploy these systems in the first place.”
That criticism points to the deeper issue beneath the energy debate. The question is not only whether Google can help bring small modular reactors online. It is also whether the power strategy behind AI should be solved by building more energy supply, by changing how AI systems are built and deployed, or by some combination of both.
For now, the Google and Kairos Power agreement is a sign of where the AI race is heading. The next phase will not be measured only in models and datacenters, but also in grids, reactors and the ownership of the clean energy needed to keep the systems running.