Essay by Eric Worrall
Accused of signing zero carbon nuclear deals with no timeline, scanty details.
Google tries to greenwash massive AI energy consumption with another vague nuclear deal
Chocolate Factory promises early-stage capital to atomic upstart Elementl
Brandon Vigliarolo
Wed 7 May 2025 // 19:02 UTCGoogle has signed a strategic agreement with nuclear project developer Elementl Power to support the early development of three potential fission reactor sites in the US.
But with no selected reactor tech and no construction timeline, the announcement sounds more like a handwaving exercise to distract onlookers from the massive amount of energy that will be expended as Google and other companies race to capitalize on the AI boom.
…
But as The Register pointed out recently, Google’s nuclear plans – along with those backed by Meta, Amazon, and others – may be too little too late to address the growing concerns that there isn’t enough power to fuel the growing demand from datacenters and AI. Experts predict an “unprecedented” spike in demand, driven in part by datacenter and AI growth, that could require 3,500 TWh of new energy generation by 2027.
Google’s own plans for AI expansion are gigantic. Google parent company Alphabet saidin its most recent earnings call last month that it intended to invest $75 billion in CapEx in 2025, much of that going to servers and datacenters to support the expansion of Google services and DeepMind AI products. At least a portion of the electricity going into those data centers is generated by burning fossil fuels, which contribute to global warming: Google itself admitted in its 2024 environmental report AI investments were a big factor as Google’s carbon emissions to increase by 13 percent year-over-year, writing “Overall, our total GHG emissions increased by 13% — highlighting the challenge of reducing emissions while compute intensity increases and we grow our technical infrastructure investment to support this AI transition.” Overall, the report said, its emissions grew 48% between 2019 and 2024.
…
Read more: https://www.theregister.com/2025/05/07/google_signs_another_nuclear_deal/
The green movement is finished. The AI revolution has barely begun, yet the momentum behind AI is already unstoppable.
Wait until advanced AI powered consumer devices – robotic household servants which can mind the kids, take care of the gardening, or provide for other needs – start penetrating the consumer market at scale. There is no way the gargantuan energy needs of the AI required to power such consumer tech will be satisfied by handwaving agreements with early stage nuke startups, it will be all hands on deck for new dispatchable energy capacity in whatever form is most readily available.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Nice logo. Be especially funny if you used Gemini to create it.
If one actually wants dispatchable power, conventional thermal or nuclear thermal are the only real choices. Weirdly, Google’s spell check does not recognize “dispatchable” as a word. Or “lawfare”.
Looked up stats from eia, who say:
U.S. utility-scale electricity generation by source, amount, and share of total in 2023
Nuclear 18.6%
Renewables 21.4%
I take out hydro and biomas because we should all know they were going to happen with or without CO2. e.g. Hoover dam was completed in 1936!
Renewables (21.4) – hydro (5.7) – burning other s&$% = (1.1) = 14.6%
-or-
Solar (3.9) + Wind (10.2) = 14.1%
After decades of technology advance and subsidy, solar and wind can’t outproduce nuclear built in the 1960’s unless you lump in pre-WW2 hydro.
(This being data that supports TH drift)
Two things:
First it’s good to see to see the climate angle so easily dismissed peacefully.
And second, if this is what gets more action on new nuclear installations, I’m good with that.
We always knew that hypocrisy was strong in the tech world, but no need to harp on that. We should just celebrate the moment and use it to affirm the need for reliable power for all users, for convenience or heavy industry or even for unrestricted EV charging, if that’s your thing.
Thinking online… I use less than 500 gallons of gas per year. Gas seems to be stabilizing at about $3. Would recharging for free (aka using a portion of the electricity my kids’ debt will pay for some day), saving est $1500 a year, make driving a Nisan Leaf a no brainer for me? Nope, gonna stick with gas engines for at least one more car.
I just hope NY backs off their plans to coerce us into EV’s through restrictions on sales of new IC cars. I suppose I could imitate Havana and keep my gas vehicles running no matter what it takes.
I wonder if they have any station wagons. We now have numerous vehicle models with little variety. Even suburban types have these massive rear lift gates. They had the ability to comfortably seat riders along with a laying down your rear gate like a pickup to haul things. Lower to get in unlike pickups, but 4 door pickups may show that these would sell. Old adage about not fixing things that ain’t broke seems ignored. Compared to earlier generations lots of us are already so spoiled that having everything done automatically and/or virtually doesn’t seem logical.
I sure wish gasoline was about USD3 per US gallon here in southern California.
$4.26 in WA State and soon to go up another 6¢ in July.
In the UK, we pay $6.71 per US gallon.
Most of the money is tax.
$2.79 at the local station in Maryland.
Here in the real PNW we’re paying $1.55 CAD per litre. four litres, approx, to the US gallon that’s $6.20 and at the current exchange rate, $9.30 US a gallon. That’ll keep the tourists away.
I don’t think much of AI but it doesn’t matter what I think of it, it is here and not going anywhere. However the demands required by AI should not have priority to our current energy supply. AI needs to get busy helping to fire up all fossil fuel and nuclear energy beginning right now. They also need to invest big time building new fossil fuel and nuclear generators. Wind and solar need to be removed from the grid, they are nothing more than speed bumps and detours on our path to abundant, reliable and affordable energy.
A collision between AI needs and consumer needs has already happened. I expect this to happen more often over the next decade.
https://wattsupwiththat.com/2024/04/19/connecticut-nuclear-data-center-controversy-exposes-tech-giant-renewable-hypocricy/
I’m hoping that someone will let us know if/when “AI” is actually “I”. Of course, by then it will be too late.
The good news is that if it turns out to be Artificial Ignorance, we won’t have to change the acronym. 😐
Maybe it should be called FI, fake intelligence.
“Wait until advanced AI powered consumer devices – robotic household servants which can mind the kids, take care of the gardening, or provide for other needs – start penetrating the consumer market at scale.”
Charging EVs at home is thought to be a huge issue as the power system is not designed for such large, additional power demands.
Should the AI powered devices become ubiquitous in households, how much power will they require, placing further stress on an already inadequate power distribution system?
Let’s hope high power lithium-based batteries are not installed in these devices as we have enough issues with them already.
I think household AI will be getting their AI through the internet. You won’t need to upgrade the wiring to your house. However, you will need the latest, greatest internet connections.
As reported on this site 8 years ago their own engineers said renewables won’t work.
As reported on this site in 2014 their own engineers said renewables won’t work.
https://wattsupwiththat.com/2014/11/22/shocker-top-google-engineers-say-renewable-energy-simply-wont-work/
Yep still makes me smile.
Yes; here is the article by 2 Google engineers [who believed the climate alarmism] on Google’s attempt to solve the “climate problem”.
Of course, the result was not published at the time [2011] since it bucked the consensus.
http://spectrum.ieee.org/energy/renewables/what-it-would-really-take-to-reverse-climate-change
“Wait until advanced AI powered consumer devices – robotic household servants which can mind the kids, take care of the gardening, or provide for other needs – start penetrating the consumer market at scale.”
And then there are all the electric cars, trucks and busses …
A reasonably educated person with a yellow pad and a pencil could have shown (as many did) how unrealistic the predicted green carbon-free power plans were … in the few places where there were actual plans with numbers and costs and all that …
“robotic household servants which can mind the kids”
Probably not during my lifetime.
I’m still thinking I might see self driving cars become standard.
I thought the real uses for AI were, doing away with the need for human workers and the total surveillance society. Seeing the apparent ease with which hackers can disrupt systems, it feels like a dystopia horror story.
Two points:
First, “consumer” AI will not be a huge driver of power demand. AI breaks into 2 distinct tasks, Training where a model is built and taught by feeding it mountains of curated data (pictures of dogs labeled as to breed etc.) so it can ”learn” (what german shepherds and poodles look like for example). THIS is enormously compute and power intensive and is what drives the big data center power budget. The other task is Inference where queries are run through a trained model. This can run on a much less powerful system and leverages the magic of digital intelligence to COPY and share the trained learning for just a few cents letting everyone carry around their own personal Einstein. And this is what most consumer AI will be doing I expect.
Second, I believe this AI gold rush will lead to the rapid demise of global warming alarmism because the “Guardians of Truth” in the tech world who control most all data access and news feeds nowadays have virtue signaled their moral superiority by censoring climate realism as articulated on WUWT and driving around in EVs, but will quickly abandon these false principles in pursuit of more billions. And I suspect many quite willingly as people who study STEM actually learn to understand data and AI has taught them quickly just how much bad data will screw you up (witness Google’s infamous black NAZIs stupidity). So they will become willing anti-alarmists and rationality on this topic will become ever more accessible, and in fact I suspect actively promoted as greed trumps false morality.
Separation of tasks is certainly how the majority of current AIs work. But I suspect the training will be continuous for the kind of consumer devices I’m talking about. Some AI researchers are developing AIs capable of continuous improvement, such as NEAT / odNeat.
Made me wonder… are the “prove you’re not a robot” tests being used to train robots?
“I suspect many quite willingly as people who study STEM actually learn to understand data and AI has taught them quickly just how much bad data will screw you up (witness Google’s infamous black NAZIs stupidity). So they will become willing anti-alarmists”
… or, they will train themselves to think the WW2 Germans were. African Americans.
Why? They adopted these false principles in pursuit of more billions. And so far it works for them.
So? Consider for a moment that according to the available evidence, it’s quite possible no one believes (some may double-think, at most) in the great and terrible ManBearPig already.
The top warm-mongers? #GreensGoByAir. So, no.
The terrorist wannabe crowd? “10/10 No Pressure” would not make sense if they believed it themselves. So, no.
That one brain-damaged child of communists? She needs a handler to keep her blathering on the right topic. Maybe she does, maybe she is little more than a living ventriloquist doll. Who can even tell?
Perhaps even a wild hedgehog in the forest gets it (even without being able to «learn to understand data»), but what of it? This did not blast into Puff Of Logic Smoke™ good old «point at a deer, call it a horse», nor «workers of the world unite», nor «War is Peace… Process», and so forth.
I think you are missing my point. In order to exploit the AI revolution to financial gain Google, Facebook, et al NEED energy. And yes, they are talking about nuclear to an extent but they know that will take far more time than they have so they WILL adopt fossil bigly because it is a competitive imperative.
And BECAUSE this is untrendy given all of the green brainwashing largely created through THEIR custodial control of search results and news feeds, they will suddenly recognize the validity of skeptical voices and thus no longer suppress them.
It is a BUSINESS decision so only needs a management directive to happen regardless of what the lemmings think. And management are the ones who have the most to gain.
So it may take a year or two, but I bet over that time web hits for WUWT and rational science based skepticism of the CATASTROPHIC ANTHROPOGENIC GLOBAL WARMING narrative will become mainstream, and with it public opinion will shift to a belief that fossil energy is in no way an existential threat as “new” (to most seeking it) science comes to light because it is no longer memory holed.
Well, they will get it… some beasts are more equal, etc.
If the clown mob will be too annoying, maybe. Otherwise, why? But that’s what “community organizers” are for. So, not until the Unofficially Official Press turns against them.
It was “a business decision” to make Fembusters. And to sink unsinkable Star Wars. And then insult the consumers. Hasbro not only had losses on toys for the same Disney Soy Wars, but repeatedly invented excuses for that and tried to downplay the “unexpected drop” in sales, blaming everything and the kitchen sink. And there were tons of other crap like this, which was obvious from the start. Those Gillette ads, etc.
Since so many observed results diverge from your predictions this much, we must conclude that your model of corporate decision making… does not reflect the reality well.
The people with money (Google et al) are spending it on nuclear. The people without money — or with taxpayer money — are spending it on windmills and solar farms.
“robotic household servants which can mind the kids, take care of the gardening, or provide for other needs – start penetrating the consumer market at scale”
I doubt that’ll ever happen- or, maybe another century from now.
Detroit: Become Human.
One can already talk to the TV and the cell phone.
Robotic vacuum cleaners abound.
Ability to remotely start the car or set the thermostat exist.
It will not take a century, although we certainly will not see it next week.
Maybe.
Where is Jeane Dixon when we need her?
But those things are light years away from robots doing your housework, watching the kids, gardening, etc. Even if they are created- who could afford them? Certainly millionaires could but not the rest of us. And, I think most people like doing these chores themselves. Of course, some chores suck- like repainting your house. I’ve done a lot of that and hate being on ladders. The best place for robots is in factories and doing the fighting in wars.
Technology yes, economy no.
Yes, robotic childcare must compete with human 30th income percentile on cost. Reminds me of a production line I wanted to automate to improve repeatability, where I learned how little Mexican hourly manual labor cost. Robot payback period was a looooong time, so no, couldn’t do it.
Not sure what sort of role model a robot might be for children. Should they be dressed up to look like moms?
What else are they to do? They have to pay lip service to the mania, so why not promise to do something a long way away that sounds vaguely progressive. Or should they promise to install lots of turbines? And then get criticized for not doing it?
“why not promise to do something a long way away” because someday the faraway day shall dawn and ‘(USA):
“The average age of Members of the House at the beginning of the 118th Congress was 57.9 years; of Senators, 64.0 years.”
“At 78 years, 10 months and 25 days old, Donald Trump, the 45th and 47th president of the United States, is the oldest person in American history to be inaugurated as president for the second time.”
I can recall when those ages seemed extremely old- not so much when you’re 75. My wife passed last year. I’m now hanging out with a “young chic” who is only 69. 🙂
Yes, I’m 1 generation younger and I wonder whether mine might be the last where people actually die. Increasing longevity makes it harder to make promises that fit the dual mandate 1) Benefit while my income depends on it 2) Cost after I’m gone.
Point of listing ages was – Did DT imagine he’d win an election at age 78 when he was age 7 or 8?
I have said for about 15 years that my generation (I’m 68) will either be the last to die (sadly the “good” scenario), or we will be the first to experience a greatly extended lifespan but it won’t matter because we will have Alzheimer’s so won’t enjoy it a bit.
Mortality sucks, and every year it sucks harder. : (
So it’s energiewende vs CIAggle bots now?
Many good things may come from it. Many nasty things may come from the winner establishing a precedent, too. Quoth Kissinger, «It’s a pity they can’t both lose».