Oklo (OKLO): Consuming Nuclear Waste To Power AI

Why AI Is Driving a New Nuclear Power Cycle

As the booming demand for energy by AI data centers is completely changing the forecast for energy consumption in the coming decade, more power generation is needed fast.

Ideally, it should come from carbon-neutral renewable sources like solar and wind. In practice, utility-scale batteries are only getting started and are not yet enough to ensure that intermittent renewables can be relied upon for continuous operations of data centers.

This is why the tech industry has been turning toward nuclear energy instead. The early moves have been of restarting recently closed conventional nuclear power plants, like the Three Mile Island nuclear reactor, which is being restarted in partnership with Microsoft.

But as tens or even hundreds of GW worth of data centers are being built, new nuclear reactors are needed. Unfortunately, conventional nuclear designs are slow to build, burdened by complex permitting, and still carry public stigma from past incidents such as Fukushima and Chernobyl.

This is why a new generation of nuclear power plants, Small Modular Reactors (SMRs), is the new trend of the nuclear industry. They are expected to be quicker to build, cheaper once built in series, and more flexible in their deployment.

Many SMR designs are replicating, on a smaller scale, the pressurized nuclear plants that the industry is familiar with. But some are moving a step beyond into the 4th generation of nuclear power plants, with one company having captured a lot of investors’ attention: Oklo.

The Ongoing Nuclear Renaissance

A Strategic Concern

Depending on the adoption rate and data center build-out speed, data centers could see their energy requirements multiply by 2x-6x by 2030.

This demand for energy will be difficult to satisfy in the West, where power grids have long been neglected and power generation mostly stagnant. Meanwhile, conventional nuclear power has only planned to rise in emerging countries for the late 2020s.

Source: The Economist

So while AI model companies might have a head start in the West, constraints on power generation might ultimately give an advantage to China. This is why SMRs are now being embraced both by policymakers and AI companies to bridge the gap.

For example, Google signed with Kairos for up to 500 MW of SMR capacity starting in 2030, while X-energy plans to deploy 12 Xe-100 reactors in Washington State to service Amazon.

Source: GE Vernova

Not All SMRs Are Equal