Tag Archives: small modular reactors

nuclear-powered data centers on Earth, solar-powered data centers in space

This might be a bit of a rambling post. I’m thinking about data centers and energy. First of all, we have a sudden boom in energy demand caused by AI, and the fossil fuel industry is not able to meet it. The amount of energy fossil fuels can supply us may have peaked in a physical sense, in an economic sense, and an environmental sense (i.e., finally seeping into the thick skulls of public perception that the long-term environmental damage is more than we can accept). And that supply is not close to the demand. I think this would be true even without the supply disruption caused by the insane, illegal, unprovoked U.S. war of aggression in Iran. But that is just one more nail in the fossil fuel coffin. I think the long-term political grip of the fossil fuel industry on our global civilization may be coming to a close. We don’t have to stop using fossil fuels for this to happen, in my view, we may need only to continue using about the same amount of fossil fuels we are now as demand continues to grow and be met by other types of supply.

Sure, we can turn to renewables to meet a portion of this new demand. Cost-efficiency of solar, wind power, and batteries is increasing exponentially, which is another nail in the coffin of fossil fuels. For that know-it-all uncle at the Thanksgiving table (a fictional character, at least in my extended family where all my uncles are quite nice) who talks about renewables “not providing a baseload”, cheap abundant batteries seem to be an answer. But solar and wind both put demands on our physical environment. I was wondering why we can’t put data centers in high rises. Part of that answer seems to be that cooling is very difficult in a high rise – when you see those pictures of the huge sprawling data centers in the suburbs, the roof of that massive building is devoted largely to cooling technology. I also wondered why those roofs aren’t covered in solar panels, and from a little research (Gemini, not independently fact checked) this need for surface area for cooling also seems to explain that. Developers first redevelop underused industrial and commercial properties until they reach the limits of that (which seems somewhat okay, other than creating dead zones in the urban landscape where jobs and stores used to be. But you don’t care, because you are entertained by your mixed-reality glasses while your autonomous electric vehicle transports you through this industrial hellscape between your home, stores and job, or whatever physical places you still need/want to go. You better get your ass on a treadmill at least part of the day though. Then again, drugs that at least partially offset the risks of not exercising also seem to be a growth industry.). And then, they start buying prime farmland and clearcutting forests. In fact, it seems that if a data center wants solar energy, what they will often do, once they have developed the property for the data center itself, is buy up an even larger parcel of prime farmland somewhere nearby for a large solar farm. There is a certain irony to fossil fuels in that by burning plants that grew 100 million years ago, we drastically reduced the amount of land needed to grow food for literal horse-power. Land become an almost negligible factor of production, and now it is becoming more important again as we push up against an energy supply ceiling. So even though the economics of renewables this may work out, there are physical, environmental, and political limits to how far it is all likely to go, and it may not be enough to meet demand. (When I say the economics work out, I assume the companies developing the land, and the politicians giving it to them in hopes of receiving a stream of tax revenue in return, do not pay the environmental costs in dollars. Which is the American/Suburban way.)

Now to nuclear power. The small modular reactor technology seems like a potential solution to the data center power demand problem. And not just the data center problem, but with abundant cost-efficient electricity we can finally electrify transportation, grow large amounts of food under lights, and desalinate large amounts of saltwater, potentially raising a number of physical capacity limits holding back our current fossil-fuel-supply-limited civilization. Is the supply of nuclear fuel for these new reactors going to be a problem? Apparently the Trump administration has taken a step that has been taboo since the inception of nuclear power, which is to give plutonium from decommissioned nuclear weapons to the private companies developing this technology (note the post I am linking to here, while an interesting read, goes in some directions I do not necessarily condone). This makes a certain amount of sense to me. It also raises a number of long-term proliferation, terrorism, and accident risks that need to be carefully managed, and the reckless dummies in the Trump administration are certainly not the organization to manage what they are unleashing.

If land itself turns out to be limiting, there is the data centers in space possibility. Like a lot of people maybe, I have been picturing those giant suburban data centers in space, and it seems somewhat daunting to get them up there. But here is a post (again, these are kind of interesting but junky opinion posts, not professional journalism. Go ahead and say it – not unlike the one you are reading right now!) saying we should think in terms of a smaller rack of computers being loaded on to a large number of satellites, all solar powered and communicating with each other by lasers. In other words, Starlink, brought to us by Elon “Hugo Drax” Musk. Recall that in Moonraker the movie, Hugo Drax terrorizes humanity from space, while in Moonraker the book he does what evil villains always do, and just hijacks some nuclear weapons and holds humanity for ransom. After Covid-/Biden-/Trump-flation, I would suggest at least ONE TRILLION DOLLARS. Or perhaps bitcoins.

$100 BILLION DOLLARS!!! … wait, don’t you think that sounds just a bit outdated?

Where are the Small Modular Reactors?

Small modular nuclear reactors might be a key part of the solution to the climate crisis, or they might not. They seem to be near the end of the R&D stage (although doubtless they can and will continue to be improved) and at the very beginning of pilot testing/proof of concept. They are not yet economically competitive with other forms of power generation (including traditional large nuclear reactors), which you could say about pretty much any new technology. There is a chicken and egg problem where you have to implement it and scale it up for the unit price to come down, and it is hard to get the private sector (and public sector, if they are short-term financial return focused) to take the chance on implementation of something that might end up not working out. Of course, the idea is to invest in a portfolio of things that have some chance of working out, such that there is a high chance at least one of them will pay off. Anyway, some facts and figures in this Physicsworld article (isn’t that name slightly redundant?):

  • Nuclear power generated 17.5% of the world’s electricity in 1996, vs. only 9% today (2026 if you are a future historian reading this).
  • I somewhat naively thought US firms might be leaders in this technological knowledge (if not in implementation of anything at scale, where I would never be that naive). But it turns out that there are two of these reactors currently operating in the world, and they are in Russia. Two are being built in China.

Sure, there are mining, supply chain and waste problems, but in my view you have to balance them against the unfolding global ecological catastrophe caused by burning fossil fuels for two centuries and counting.