DeepSeek Gives the Lie to the Fossil Fuel Industry’s Latest Stalling Tactic

Cince we’ve all been weathering the head-spinning assault on the Constitution by the new administration (and, at Third Act and elsewhere, trying to do something about it), I thought it might make sense to provide you with one interesting piece of good news.

It concerns this DeepSeek Chinese AI program that you’ve doubtless been reading about in recent days. I’m the last person to turn to for an analysis of its virtues (I remain fully dependent on my highly-developed Natural Cluelessness), but I am very clear that it complicates the main current task of the fossil fuel industry: glomming onto AI as the latest excuse for building out a bunch of gas-fired power plants.

That narrative—which has been building for a year or so—holds that we will need so much electricity for the data centers to keep this technology running that we’ll have to give up on dealing with climate change for now. It reached its zenith last week when the new administration announced something called Stargate, a $500 billion plan that, as U.S. President Donald Trump put it, would be “the largest AI infrastructure project in history.” This was the moment when he declared an “energy emergency” so that we could build more power plants (but not, of course, the solar or battery parks that Silicon Valley experts have testified would be the most efficient way to power these megacenters).

The increasingly gloomy idea that there was no possible way we could every deal with climate because AI would soak up every new electron that sun and wind could ever provide, may not be quite as true as it seemed to some a week ago.

I would venture to say, given Trump’s predilections, that he neither understands nor cares much about the AI part of all of this, but he completely groks it as a way to pay back Big Oil for the $445 million they invested in the last election. (Political donations come in millions with an M, and the paybacks come in billions with a B—of our money). As Bloomberg reported, the whole DeepSeek incident shows how dependent on this AI story the fossil fuel industry is as an excuse for expansion (just as a couple of years ago it was dependent on the Ukraine war story):

In one brutal blow, DeepSeek has revealed just how many energy-related businesses in the U.S. have been banking on an artificial intelligence boom—and the surge in power demand it was supposed to bring.

For the past year, their growth expectations and share prices were boosted by the belief that AI would require an unprecedented wave of data center construction, with some centers needing as much electricity as entire cities. Utilities and power plant operators benefited, too, but the effect went far wider than such obvious industries, touching an astonishing array of companies.

That became clear the moment China’s DeepSeek unveiled a chatbot that could rival the best American AI programs while using just a fraction of the electricity, perhaps as little as 10%. DeepSeek’s announcement hammered the shares of uranium producers and natural gas pipeline operators alike. Companies that supply power plant equipment and data center cooling systems suffered as well in Monday’s big selloff.

I don’t think we know enough yet to know if that claim—”rival the best American AI programs while using just a fraction of the electricity, perhaps as little as 10%”—is actually true. There are voices in the U.S. today beginning to claim that DeepSeek plundered American code to make its breakthroughs (which is truly funny, since American AI merrily plundered everything everyone has ever written, to make its breakthroughs). And there are others saying that DeepSeek, by making AI more affordable, will actually increase the amount that it is used.

But it does seem as if something new is afoot—the search for efficiency, instead of just massive brute force—in constructing artificial intelligence. As the investment gurus Dylan Lewis and Tim Beyers at The Motley Fool put it:

One of the main things that has popped up a lot in the reporting on this is that the compute necessary for what is running on DeepSeek is a fraction of the compute for some of the other systems. Watching the way that the market is processing this, we are seeing Big Tech companies take a hit. We are seeing some of the chip companies take a hit. We're also seeing energy companies take a hit because there is this feeling that maybe as we get a little bit more technologically advanced as other players start coming into the space, some of the energy demands for this technology won't be as big as people have maybe originally thought.

and

There is no way we are going to be building out the amount of energy infrastructure required to service all this at the level we are talking about in the timeframe we were talking about. Then what happens? You have a constraint. Do you keep doing what you're doing and overwhelm the energy infrastructure, knowing full well you can't build it out at the level that you want to, in the timeframe you want to, or do you do what the industry always does, which is find areas of efficiency to scale in a better, more economical way? That's what always happens.

I’m not beginning to tell you how all this comes out. All I’m saying is, the increasingly gloomy idea that there was no possible way we could every deal with climate because AI would soak up every new electron that sun and wind could ever provide, may not be quite as true as it seemed to some a week ago. (It would be awfully nice if this kind of move toward computing efficiency catches on—here’s another story from this week, about new software fixes that seem capable of reducing power demand at these data centers by 30%.)

You could, I think, even draw a crude analogy between DeepSeek and solar power, in that it seems to be producing the same thing that OpenAI and Meta are producing for a fraction of the cost, the same way that photovoltaics produce power more cheaply than Exxon. And since it’s open source, it undercuts them in another way too: Anyone can get their hands on this and work with it. (“Anyone” meaning anyone who knows what they’re doing—not me, obviously). The advantage of hoarding chips, which has been Big Tech’s strategy, may turn out to be kind of like the advantage of hoarding “reserves” of hydrocarbons—less solid than might have been expected. To complete this imperfect analogy, AI, like the solar cell, may have been invented in the U.S., but it’s China who may figure out how to make the most of it.

It was only two (very long) weeks ago that former President Joe Biden, in his farewell address, warned us against the “tech-industrial complex.” Some youngsters, working around the constraints imposed by the U.S., seem to have struck a blow in that direction. It’s obviously far too much to hope that the U.S. and China might cooperate to develop this new technology in some rational way—the best we can hope for, I think, is that they won’t actually destroy the planet en route to whatever nirvana these new intelligences have in mind for us.