AI’s Insatiable Appetite for Electricity Could Trigger a Recycling of a Long-Neglected Energy Source—And Why That’s a Bitterly Offensive Idea
The headline from CNBC on 17 May 2026, “AI’s insatiable appetite for electricity could revive a forsaken energy source,” isn’t just a catchy bit of tech insurance. It’s a grim admission: humanity’s brightest invention might be dragging us back into the dark ages of energy policy. AI isn’t just eating data anymore—it’s gobbling electricity with the voracity of a locust swarm during a national emergency. And now, in a twist only a Silicon Valley socialist could love, it might be pulling us back to an energy source so tiresome and controversial that even coal was embarrassed not to use it.
CNBC’s report specifically highlights how AI’s rotational demand for power—driven by constant training of mega-models and real-time inference hell—has pushed grids to their absolute limits. Countries like the US, Germany, and Australia are already sweating bullets over crippling energy shortages, with officials whispering about chartering Arctic icebreakers to power data centers. But here’s the kicker: some experts are proposing a return to a crucified energy source—nuclear power. Yes, the same one that scared everyone since Chernobyl and Three Mile Island. One “forgotten” technology that likes to keep its distance from tech utopians.
### The Great AI Energy Larceny
AI isn’t just a software beast; it’s a power vampire with a PhD in thermodynamics. Training a single large language model can guzzle as much electricity as 100,000 average households for a year. And that’s before you factor in the actual applications—generative AI tools used for everything from writing email summaries to creating deepfake politicians.
The source material from CNBC doesn’t name specific companies or models, but it’s fair to assume it’s the usual suspects: OpenAI’s GPT series, Anthropic’s Claude, and Meta’s Llama ecosystem. These aren’t just clever algorithms; they’re energy-hungry monsters requiring exascale computing. One analyst likened it to “feeding a jet engine to a toaster.” Worse, the problem isn’t just about capacity. It’s about sustainability. Nvidia GPUs might be the brain of AI, but they’re fed by grids powered by coal in parts of Asia and gas in Europe. The result? A kind of neocolonial exploitation where digital giants force energy poor nations to subsidise their luxury tech.
### Why’s Everyone Panicking Now?
The timing is textbook corporate recklessness. On 14 May 2026, The Guardian ran a piece about “AI Bonnie and Clyde” conducting digital arson via autonomous tech—a separate crisis, but one that highlights how quickly AI is spiraling out of control. Meanwhile, governments are now waking up to the fact that AI’s growth isn’t just a tech problem; it’s an energy crisis. The UK’s national grid operator recently leaked a memo stating that if AI development continues unchecked, we’ll need a 30% increase in baseload power by 2030. That’s not a challenge; that’s a death sentence for current energy policies.
But here’s where CNBC’s angle gets awkward. Instead of screaming about emissions or fossil fuels, the story leans into a “solution” that’ll make even Boris Johnson blush: revive nuclear. The logic is that fission, once shunned for being too dangerous and too slow to build, could now be revived via modular designs or advanced reactors. Proponents argue it’s clean, reliable, and scalable. Critics? They’d rather see a particularly creative AI-generated meme than a new nuclear plant.
### The Backlash Is Coming from Every Direction
This isn’t just environmentalists howling at the moon. Energy companies are livid. A spokesperson for Drax Group, a UK power giant, stated: “We’re not built to run saunas for Silicon Valley’s latest fad. If AI wants to win, it should be footing its own bills.” Meanwhile, climate activists are calling nuclear a “terrible compromise.” Greta Thunberg, via a leaked EU statement, raged: “We’ve fought for renewables for a decade. Now we’re told to pray to a radioactive deity? Madness.”
Politicians, meanwhile, are dancing on eggs. US Energy Secretary Jennifer Granholm hinted that tariffs on Chinese AI hardware might be on the table if they continue “abusing” American grids. But China, of course, is thriving. Their AI clusters are reportedly powered by state-backed nuclear plants, merging state control with tech ambition. One Chinese tech executive joked in a private interview: “We’ll just tell our grid to ‘finish the job’ when it’s cold. That’s how we do it here.”
### What’s at Stake—and What’s Next?
The consequences here are distilled down to a single word: chaos. If AI continues its exponential growth without a viable energy framework, we could face blackouts, skyrocketing energy bills, or a return to debates about fracking that should’ve died in 2010. For ordinary people, this means either paying more for electricity or facing disruptions in AI services. Imagine a world where your ChatGPT costs as much as a holiday.
The next phase is predictably bureaucratic. Governments will either crack down on energy-hungry tech or roll out pie-in-the-sky subsidies for nuclear. The US might push for new reactor developments, while China could double down on state-funded projects. Meanwhile, environmental groups will likely sabotage any nuclear proposal at every turn. It’s a three-way war: tech companies, energy firms, and activists, all fumbling for a solution while the planet burns.
### This Is My F***ing Take, and You Can Take It or Shove It
Look, AI isn’t evil. It’s not here to break the planet. But let’s not pretend it’s some eco-friendly saviour. Every time we worship at the altar of innovation, we have to ask: At what cost? The electricity crisis isn’t just a tech problem; it’s a moral one. We’re allowing a select few corporations to drain the lifeblood of our energy systems so they can make a chatbot that mispronounces “bollocks.”
Nuclear isn’t the holy grail. Renewables aren’t a panacea. The real answer is probably a brutal mix of all three—and a willingness to stop treating AI as a temple for Sisyphus to push stones uphill. If we’re going to fund this digital revolution, let’s make sure it doesn’t come at the expense of our grandchildren’s planet. Or at least, let’s make them pay for it with crypto.
Frankly, if AI needs that much power, maybe we should ask it to build its own grid. With lasers. And less whining.