First Microsoft, then Google. Who’ll turn next to nuclear power plants to ensure their long-term electricity requirements? I’m guessing Elon is well on the way by now.
Many a politician is now thinking, “If these guys are doing it, did we miss out?”
Well, yes and no. These big tech companies are doing it, but for the wrong reasons, and with the wrong technologies.
A Belgian journalist wrote today: “Als AI een fata morgana blijkt te zijn die alsnog een energiedoorbraak heeft geforceerd, zou dat een mooie troostprijs zijn”
What he means, I guess, is this. If AI proves to be a hoax, the benefit could be that it may pave the way to new modern nuclear power plants, and thus save the climate. How’s that for a visionary statement: failing AI goals saves the climate. Or more direct: we somehow hope to save the climate, not by appropriate action, but by failing another dumb project.
Ok, back to real life…
Here’s a more extensive list of wrong thinking flying around today:
- AI in the form of LLM or the like, is not intelligent, nor will it save the world. There are plenty of useful AI tools, but they are small and they have been around for a long time, proving useful, and not requiring a significant part of the world’s electricity production. LLM are not among them, they are just statistical models of all the nonsense that has been published in the cloud sofar. Nobody will benefit from an hallucinating Google. Turning to nuclear power to feed huge AI models is not a priority.
- Even more so for crypto currencies.
- As tech companies predict they will need massive amounts of reliable electricity, while trying to keep their green washing promises, they start loosing faith in sun and wind. Installed power does not deliver during Dunckelflauwe.
Moreover, “free” sun and wind turns out to be a costly mistake: CREG estimates 7 billion EUR for a switch in the North Sea iso the provisioned 2.2 billion, and a switch does not produce electricity. So they finally decide to invest in nuclear technology. And because they only have eyes on their own dedicated needs, they select SMR technologies. However, from a societal point of view, SMRs make no sense at all. Society needs massive nuclear energy conversion in the very near future, so it needs to invest in large reactors, not small reactors. The argument that an SMR requires less investment is not true if you know for a fact that we will need to build large numbers to cover large needs. - LWR SMRs require HALEU fuel with up to 20% U-235 enrichment. That’s just physics: small core needs dense fuel. HALEU is not readily around, as the large legacy LW reactors are all using LEU fuel below 5%. The easiest way to get it is to dilute legacy HEU. However, the SWU for that that stuff has been payed in the past, so this fuel is better put to use in high flux MTRs and reactors for producing medical isotopes. Today, UZ Leuven gets delivered only half of the required Tc99m, due to production problems in both BR2 and HFR. Better not have cancer these days, while Microsoft is training hallucinations.
- Belgian captains of industry woke up today, though the CREG broke the news in May this year, and Brigid pointed out the flaws in the ELIA investment plan months ago. Now the question is raised whether the North Sea switch should be put on hold or even cancelled, because of budget overrun. If every project with a budget overrun were to be canceled, metro line, parking lot, government software, you name it, we wouldn’t have any project left. So no, not because of budget overrun. But because of the insight that sun and wind are not going to power our modern society. And it’s certainly not true that sun and wind are free. However, as long as nuclear engineers in Cabinets are payed to confirm government members that sun and wind are both free and sufficient to power a modern society, we’re not getting anywhere.