Google gifted the world a path to cheaper nuclear energy when it signed a multiyear agreement with Kairos Energy to supply electricity.

Kairos designs salt-based fourth-generation small modular reactors (SMRs) and Google is acquiring seven. Each reactor will generate about 70 MW (about one-twentieth scale of a traditional nuclear reactor).

The key point to take away is that Google’s bulk order will put Kairos on an appealing learning curve; every subsequent reactor should become cheaper than the one before, as production, deployment and operational methods improve, breaking a cycle of ever more expensive nuclear power. SMRs could have a learning rate of between 10% and 20% for each doubling of cumulative production.

Let’s call it 15%, which is about three doublings, so the eighth reactor could rock in about 30% cheaper than the first. It’s remarkable that it took a boom in generative AI to reverse thirty years of bad US government policy around nuclear energy.

Even the environmental organisation Sierra Club has stopped fifty years of opposition to nuclear power. Solar and batteries will be the dominant provision for electricity in the next two to three decades in the distributed age, but nuclear will act as a bridging fuel from the 2030s until then.

Google’s move will start a chain of events for the development of modular nuclear technology at a scale that benefits us all.