There have been a collection of headlines about nuclear energy powered data centers of late – involving major technology companies (Google, Oracle, Microsoft). On the face of it, this makes a lot of sense – a nuclear plant provides a steady flow of energy and datacenters need a constant supply of power.
Nuclear energy has a tendency to inspire extreme reactions. However, given its obvious promise/potential, the lack of forward momentum on nuclear has always intrigued me.
Until I came upon a simple and logical explanation involving learning curves.
All new technology that is deployed en-masse gets the benefits of learning curves. As we deploy it, its costs get cheaper. And, as it becomes more abundant, we also learn how to use it better – increasing its benefits.
Today’s Artificial Intelligence/AI wave is a great example. It is expensive to deploy large language models right now. However, the cost of doing so has come down over ten-fold in the past two years. This will continue happening over the coming years – thanks to learning curves.
However, as Matt Ridley explains in “How Innovation Works”, nuclear energy hasn’t had the benefit of learning curves because of the cost of experimentation.
The cost of error in any nuclear reactor is high. As a result, nuclear is highly regulated. This results in a “waterfall” model of deployment vs. a more learning-filled agile model. For example, this means that teams on the ground are prevented from making any changes from the original spec that was approved by regulators – even if it makes little sense to continue as per plan. The costs of such rigidity add up quickly.
One area of experimentation with nuclear energy of late has been to go “modular.” This means smaller nuclear reactors that can be deployed in various places. This modularity might succeed in reducing the cost of error. That, in turn, might enable nuclear energy production to finally experience learning curves that other energy sources have benefited from.
Either way, the cost of not learning will continue to inhibit the growth of nuclear energy – at a time when solar and wind continue to grow exponentially driven by falling costs from their own learning curves.
This offers an important life lesson too. Learning compounds. While we often focus on the upside, the cost of not learning has a compounding negative effect that can have a debilitating effect over time.
