A Journey in Nuclear Power Generation

I recently decided to return to my roots and in the process have realized the past is indeed prologue. Let me explain.

Early in my career I took a job with a division of the Washington Public Power Supply System, (better known as “WOOPS”) that was in the process of building several huge light water nuclear power plants in Washington State. WOOPS squandered billions and defaulted on public bonds leaving pension funds and others holding the bag. Those plants are now empty hulks of moss-covered concrete. The failure of WOOPS coupled with Three Mile Island and Chernobyl served to put the last nail in the coffin of nuclear power generation. I didn’t see it returning – ever. That was the mid 80’s.

But forty years later, AI is eating the world. It has also eaten all our energy reserves.

Unfortunately, the US electric grid and generation capacity - designed for the industrial age - is not capable of powering AI’s future. Solar, wind, battery storage and microgrids have helped but are not capable of absorbing the huge needs alone. Something needs to change. Data center companies are nearly at a standstill waiting for electricity and the incoming administration wants to address this need with carbon spewing coal, gas and fossil fuel plants which might not be a terrible idea for a very short time, but we must create better and cleaner forms of energy generation quickly. It would appear, short of obtaining the miracle of fusion (companies are working on this too) that nuclear is in the picture again, but this time “nuclear” is a different ball game.

Large light water reactors are still being built, but not in the US. They still cost billions to build, produce huge quantities of radioactive waste and take decades to perfect. These old designs are likely not the answer to our power needs. Nothing has changed in this area since I worked for WOOPS.

Recently, however, several enterprising companies have redesigned nukes making them much smaller (some the size of a shipping container) and use radically different designs from their light water ancestors. These reactors, called SMR’s, or Small Modular Reactors, are an enticing power source for hungry data centers. Ideally, there would be one SMR for each huge data center thereby conserving public power and renewable sources for other uses and either forestalling or eliminating the need for more large reactors. And while many of these SMR designs still produce nuclear waste some produce so little (and in different forms) that they are almost completely free of waste.

There are many companies and countries designing and building SMRs. (follow this link for a list and explanation of various technologies). I particularly like fast breeder reactors cooled by liquid sodium such as Oklo’s design. This design keys off a sixty-year-old proven design that the US failed to deploy because it did not produce plutonium that the country needed for the arms race. It can make enough fuel to keep itself going for years, creates tiny amounts of waste and provides valuable medical isotopes to boot. And if that wasn’t enough, it shuts itself off without any engineering help; no pumps to fail, no meltdowns!

So, in this case, the past is prologue; nukes are back, and I’ve come back to (nearly) where I began. I’m hoping the intervening forty years has created not only a safer method of nuclear energy generation, but that medicine and other worthy endeavors will benefit from AI powered by clean nuclear energy.