In Continental Europe (especially the Low Countries and CEE) the fundamentals for grid demand are very strong (including continued electrification policies) and current wholesale electricity prices easily support new build nuclear even at LCOEs like the FOAK plant at DNNP without needing RAB. Plus energy sovereignty, already on the agenda and even more so since Hormuz. No AI boom needed.
What’s held Europe back is their suicidal anti-nuclear policies of the last forty years, now quickly disappearing, as well as the bureaucracy, permitting, and NIMBYism that also holds back Canada. That’s quickly disappearing too as industry begins to genuinely pack up and leave due to prices and politicians start to panic.
AI bubble bursting will indeed hurt US deployments. (Especially agree on the Oklo and Aalo types tied most closely to Big Tech.) There is however a world beyond the US, and I think it’s waking up from its 30-year Pax Americana slumber.
Good article. Only one comment - please use the current name of the nuclear facility in Middletown, PA. It is officially the Crane Clean Energy Center, not TMI.
Good exposition of the risks of excessive expenditure on LLM vs other advancing SW intelligence and a contraction in LLM investment setting back fissionelectric development and deployment.
the data center and LLM boom repeat the pattern of imbalanced allocation to software and related domains with confident future demand while other domains are cost burdened, stagnant and mired in a 3 decade and counting industrial obsolescence - which keeps the heartland depressed and younger generations in the precariat. Probably investment capital should be allowed to do its competitive thing with LLM and fission, we do need some investment in LLM and fission technology. Public subsidies should not be provided they are better spent on infrastructure for starved domains. Agencies can help infrastructure development for Compute by removing malthusian regulations. Agency partnerships could engage the infrastructure components of Compute to steer towards reusable energy (to accommodate potential future changes in domains needing energy locally), and intensive footprint and balanced conservation. there is also the hazard of EA and other liberal paradigms learned into machines which agencies as yet seem wholly incapable of addressing.
fissionelectric ventures should design their development to get as much push from Compute as possible at the same time being able to Decouple from Compute if it goes down.
Very North American view.
In Continental Europe (especially the Low Countries and CEE) the fundamentals for grid demand are very strong (including continued electrification policies) and current wholesale electricity prices easily support new build nuclear even at LCOEs like the FOAK plant at DNNP without needing RAB. Plus energy sovereignty, already on the agenda and even more so since Hormuz. No AI boom needed.
What’s held Europe back is their suicidal anti-nuclear policies of the last forty years, now quickly disappearing, as well as the bureaucracy, permitting, and NIMBYism that also holds back Canada. That’s quickly disappearing too as industry begins to genuinely pack up and leave due to prices and politicians start to panic.
AI bubble bursting will indeed hurt US deployments. (Especially agree on the Oklo and Aalo types tied most closely to Big Tech.) There is however a world beyond the US, and I think it’s waking up from its 30-year Pax Americana slumber.
very good piece. This is related and thought I would share.
https://needsofthemany98.substack.com/p/the-two-bubbles-america-is-building?r=gwg0&utm_campaign=post&utm_medium=web
Good article. Only one comment - please use the current name of the nuclear facility in Middletown, PA. It is officially the Crane Clean Energy Center, not TMI.
Fair point but for the general reader will have a far lower chance of knowing what is being discussed.
Good exposition of the risks of excessive expenditure on LLM vs other advancing SW intelligence and a contraction in LLM investment setting back fissionelectric development and deployment.
the data center and LLM boom repeat the pattern of imbalanced allocation to software and related domains with confident future demand while other domains are cost burdened, stagnant and mired in a 3 decade and counting industrial obsolescence - which keeps the heartland depressed and younger generations in the precariat. Probably investment capital should be allowed to do its competitive thing with LLM and fission, we do need some investment in LLM and fission technology. Public subsidies should not be provided they are better spent on infrastructure for starved domains. Agencies can help infrastructure development for Compute by removing malthusian regulations. Agency partnerships could engage the infrastructure components of Compute to steer towards reusable energy (to accommodate potential future changes in domains needing energy locally), and intensive footprint and balanced conservation. there is also the hazard of EA and other liberal paradigms learned into machines which agencies as yet seem wholly incapable of addressing.
fissionelectric ventures should design their development to get as much push from Compute as possible at the same time being able to Decouple from Compute if it goes down.