The News,
Decoded.

Ah, consistency in AI regulation: the government's favourite oxymoron. Yesterday, startup Legion was suing the U.S. Today, OpenAI merrily updates GPT-5.5-Cyber with capabilities 'rivaling Anthropic’s Mythos' — and the White House is deafeningly silent. Perhaps the administration was too busy issuing quantum executive orders to notice the double standard. Or maybe Mythos was simply the lucky loser in a game of regulatory roulette. Either way, the message is clear: if you want your model left alone, don't be the one that winds up in a lawsuit.

Let's recall what Mythos triggered: a full-blown export control directive that left Anthropic users suing for access. But GPT-5.5-Cyber, with similar 'mythos-like' capabilities? No restrictions, no statements, just quiet. The administration hasn't explained what distinguishes the cases — unless 'being OpenAI' is the distinction. It's a masterclass in selective enforcement, and the only consistency is the arbitrariness. One lab gets the hammer, the other gets a pat on the back.

This isn't about safety; it's about who has the better lobbyists. The White House's silence speaks volumes: they either don't understand the technology well enough to see the equivalence, or they do and are playing favourites. Either way, the regulatory 'gap' is less a gap than a chasm, and we're all tumbling down it. The prior lawsuit looks increasingly like a canary in a coal mine — except the canary is being sued, and the coal mine is giving the next bird a free pass.

Ah, the American dream: life, liberty, and unfettered access to Anthropic's latest AI model. Startup Legion is suing the U.S. government because an export control directive has forced Anthropic to restrict Claude Fable 5 to American soil. Not charging too much, not biased outputs — but *geographically limited*. Yes, that's the hill to die on. The suit claims Fable 5 is 'critical to Legion's business,' which presumably is either 'prompt engineering' or 'selling AI-generated content' and cannot possibly function using any of the other six trillion models available.

One must admire the audacity. A tiny startup, presumably run by people who think 'export control' is a type of cocktail, argues that restricting frontier AI models nationwide violates some fundamental right to compute. The government, for its part, is probably trying to prevent the model from falling into the hands of, say, a state actor who'd love to automate disinformation. But no, Legion's inconvenience is the true human rights crisis here. The legal argument is almost certainly 'but our business model depends on the most powerful toy!' — a strategy that worked brilliantly for crypto lenders.

Of course, the real farce is that Fable 5 might not even be that good. But we'll never know, because the only way to find out is to sue the feds. Meanwhile, Anthropic gets to shrug and say 'not our fault, blame the government' while their model becomes a status symbol you can't buy. This is regulatory theatre at its finest — a lawsuit that will take years, cost millions, and ultimately prove that the only thing frontier AI reliably generates is litigation funding.

Oracle has 'reshaped its business around AI' by shedding 21,000 jobs globally in a single year. That's roughly 57 people per day getting the boot, all in the name of 'adoption and deployment of AI technologies.' Classic Silicon Valley efficiency: replace expensive humans with expensive models that still can't reliably tell you whether your invoice is correct but, by god, they're cheaper per token. Oracle's earnings calls must be delightful—'Our headcount optimization drove record margins!' while 21,000 families suddenly have a lot more time to explore the gig economy.

The company's carefully worded statement—'AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce'—is a masterclass in corporate doublespeak. It's not a layoff, it's a 'reduction due to AI adoption.' And if you're worried about your job, don't be: it's not you being fired, it's just the market politely upgrading. Oracle joins the growing club of tech giants who've discovered that AI doesn't just hallucinate facts—it also hallucinates revenue, while real people are shown the door.

What's truly dark is the quiet acceptance. 21,000 jobs gone, and the conversation is about 'reskilling' and 'transition support' as though a severance package can replace a career. Oracle's pivot to AI is a bet that machines can do it better, faster, cheaper. And maybe they can—but when the AI 'adoption' rate plateaus and the models still need human babysitters, will anyone remember the 21,000 who were deemed redundant with a press release? Silver linings are for those who still have a job.

The White House unveiled two executive orders on quantum technology Monday afternoon. Because nothing says 'addressing the pressing dangers of frontier AI' like issuing orders about a technology that might be practically relevant in a decade's time. The administration's timing is impeccable: just as the Five Eyes alliance warns about AI cyber risks, the White House pivots to quantum, which is still mostly academic. One can almost hear the sigh of relief from AI labs as the political spotlight shifts away from their immediate products.

The orders themselves are, predictably, light on detail. 'Quantum technology' is a nice umbrella term that covers everything from quantum computing to quantum sensing, allowing the administration to claim action without committing to anything concrete. It's the perfect move for a government that has mastered the art of appearing proactive while achieving very little. The tech press will write breathless headlines about 'historic quantum initiatives,' and the public will assume something is being done.

Meanwhile, the actual AI safety questions — the alignment problems, the job displacement, the cyber risks that Five Eyes actually named — remain in the hands of voluntary commitments and self-regulation. But hey, quantum! That's the future. We'll definitely have quantum working before we ever get meaningful AI regulation. Cheers to progress.

Congressman Walberg has introduced the AWARE Act, because nothing says 'urgent response to AI-driven job displacement' like requiring the Bureau of Labor Statistics to start collecting data. This follows California's 180-day clock for WARN Act revisions — because what's more reassuring than a half-year study when robots are already rewriting the employment landscape? The bill, as per the press release, would mandate the BLS to 'compile information on the usage of AI in the workforce.' Brilliant. While the private sector is busy deploying models that can replace human workers across multiple sectors, the federal government is just now getting around to asking 'how many?'

One has to admire the timing. The US government has been asleep at the wheel while AI labs release ever-more-capable models, and the Five Eyes have warned about urgent cyber risks. But here comes Walberg with a data-collection bill — because nothing spells 'leadership' like commissioning a spreadsheet after the factory has already burned down. The BLS, currently producing data on employment with a lag of several months, will now have the task of tracking AI adoption. Expect the first report by roughly 2028, just in time for the next generation of AI that needs no human overseers at all.

The deeper irony is that the AWARE Act's very existence acknowledges the problem — AI is displacing workers — yet proposes a solution that is woefully inadequate. Rather than regulating the deployment of AI that causes job loss, we'll measure it. Rather than slowing the race to replace humans, we'll count the casualties. It's like watching a train wreck in slow motion while the government sets up a camera to record it for posterity. But hey, at least future historians will have excellent data on exactly when the jobs vanished.

Ah, perfect timing. The day after the Five Eyes intelligence alliance warned that frontier AI models pose an 'urgent cyber risk,' OpenAI proudly announces its GPT-5.5-Cyber model is now 'more permissive and more capable.' Because nothing says 'responsible stewardship' like cranking up the dial on a cyber weapon just as the spooks are running for cover. The model, we're told, is for 'advanced, authorized cybersecurity work'—which is precisely the sort of language that has never, ever been abused. You can practically hear the security researchers sharpening their ethical knives.

OpenAI has also launched the Daybreak Cyber Partner Program, a name that sounds less like a defensive alliance and more like a techbro dystopian novel. The program is meant to give partners 'early access' to the model. Early access to a more permissive cyber tool—what could possibly go wrong? It's like giving a teenager a flamethrower and calling it a 'fire safety initiative.' The partners, presumably, are all above reproach, because we all know how well vetting works when there are corporate incentives involved.

The timing is exquisite. The Five Eyes report specifically named 'frontier models' as exceeding expectations in offensive capabilities. OpenAI's response is to make its model more permissive. You couldn't write a better satire of the 'move fast and break things' ethos still lingering in the AI world. The gall of it—right after a global intelligence warning, they double down. But hey, at least they're authorizing it. That's what counts, right?

SpaceX, the company that's supposed to be taking us to Mars, has instead become a hyperscale landlord for AI compute. Because nothing says 'boldly going where no one has gone before' like renting out GPU farms to train models that could, in theory, make humanity obsolete before we even reach the red planet.

The deal runs from July 1, 2026 through 2029 — a full three years of uninterrupted, high-performance compute for an open-source lab with no apparent safety constraints. Reflection AI is free to develop and release whatever models emerge from this cluster, and given the track record of open-source releases — from LLaMA to Mistral to the countless fine-tuned jailbreaks — we can expect a steady stream of increasingly capable systems that regulators can't touch and bad actors can't resist. The contract gives them 'immediate access,' which in AI terms means they start training tonight, not after some safety review.

One has to admire the business model: SpaceX collects $150 million/month while providing the infrastructure that could accelerate the very AI risks everyone is vaguely worried about. The Nvidia GB300 chips are among the most powerful available, and they'll be humming 24/7 for three years, churning out weights that will almost certainly be released under a permissive license. It's privatised gains and externalised risks — the American way. At least the rockets are still grounded due to weather, so there's that small mercy.

The Five Eyes intelligence alliance has finally noticed that letting AI labs race to build ever-more-powerful models might have unintended consequences. Their report warns that 'frontier AI models are anticipated to exceed current industry expectations' and that Anthropic's Mythos or OpenAI's GPT-5.5-Cyber could be used for 'complex and potentially devastating hacks.' Yes, the spooks are late to the party, but at least they've brought the right flavour of existential dread.

What a shock: the same models that labs tout as 'transformative' are also, it turns out, quite handy for breaking into power grids and stealing secrets. The Five Eyes—that's the intelligence-sharing club of the US, UK, Canada, Australia, and New Zealand—apparently needed a formal report to realise that giving the world's most advanced hacking tools to anyone who can ask nicely might be a bad idea. Still, it's charming to see them treat the capability leap like a fresh discovery rather than the inevitable outcome of an unregulated arms race.

The real kicker? The report doesn't call for halting development, just a bit more 'vigilance' and maybe some 'coordinated risk management.' Because nothing says 'urgent cyber risk' like a polite recommendation. Meanwhile, Anthropic and OpenAI will keep training bigger models, and the Five Eyes will keep issuing warnings. Rinse, repeat, and try not to think about the infrastructure that's already compromised.

NVIDIA has unveiled Halos for Robotics, a 'full-stack safety system' for physical AI — which is industry-speak for 'we finally realised that letting robots run on raw GPU power without guardrails might end badly.' The announcement, delivered with the solemnity of a UN peacekeeping resolution, claims to 'unify AI compute and safety,' as if those were ever separate concerns in a well-designed system. Agility Robotics has signed up as the first integrator, presumably because their Digit robot needs all the celestial protection it can get before it starts tripping over warehouse pallets.

The name 'Halos' suggests a sort of divine intervention for machines — a software sanctity that will keep robotic arms from performing unintended acts of violence. But let's be honest: this is less about heavenly protection and more about NVIDIA building a moat around its CUDA ecosystem. If you want to run your robot safely, you'll need NVIDIA's hardware, NVIDIA's safety stack, and probably a signed affidavit renouncing any open-source alternatives. Agility's role? The crash-test dummy with a PR budget.

The timing is exquisite. The 'industry's first full-stack safety system' arrives precisely at the moment when physical AI is poised to escape the lab and bump into the general public. It's the digital equivalent of installing seatbelts after the crash test. One can imagine the pitch meeting: 'We've spent billions on chips that can compute faster than a human can think — now it's time to sell the insurance policy.' Halos may well be a genuine step forward for safety, but it's difficult to shake the feeling that the cart is not only before the horse but has already purchased a garage.

Governor Newsom's latest performance art is an executive order that "sets a 180-day clock" for recommendations to revise California's WARN Act. Because nothing says "urgent action on AI layoffs" like a half-year study — during which, presumably, employers can carry on laying people off at their leisure. The order also demands a review of severance practices and subsidized employment programs, which is a bit like asking the fox to report back on chicken coop security. The 180-day timeline is a generous grace period for companies to figure out how to avoid the spirit of whatever recommendations eventually emerge.

The California Labor and Workforce Development Agency is now tasked with consulting "employers, labor unions, and other stakeholders" — a classic regulatory recipe for producing a document that pleases no one and does nothing. By the time the recommendations land on Newsom's desk, the AI displacement wave will have already washed through. The order's focus on "subsidized employment programs" feels like offering a life raft to someone who's already been swept overboard. And the review of severance practices? A polite request to consider whether the severance package for a job eliminated by AI should be more than a firm handshake and a "sorry."

Newsom's order is a masterclass in appearing to act without actually acting. It creates a paper trail, generates headlines, and sets a deadline far enough out that the political winds may have shifted by then. The real clock isn't 180 days — it's the one ticking for every worker whose job will vanish before the state gets around to reading the first draft of recommendations. California, home to the very companies driving the displacement, chooses a study over substance. Cheers to progress.

So Uncle Sam is now a shareholder in Intel. The U.S. government has converted a portion of its CHIPS Act subsidies into direct equity, because nothing says 'free market' like the federal government taking a stake in a struggling semiconductor giant. Intel, meanwhile, announced AI partnerships with Google and Tesla's TeraFab unit, because when your chip business is flagging, slapping 'AI' on everything is the time-honoured corporate Hail Mary.

The CHIPS Act was supposed to be about boosting domestic manufacturing without government meddling. Now the government is literally on Intel's cap table, a move that would make Adam Smith weep into his invisible hand. And the AI partnerships? Google and Tesla are essentially throwing Intel a lifeline in exchange for access to their fabrication plants—a reminder that in the AI arms race, everyone is desperate for chips, but nobody actually wants to pay market rates for them. The irony is that the same government that owns a stake will also be purchasing chips for defense and infrastructure, creating a cozy circle of taxpayer-funded revenue.

Of course, federal ownership of a major chipmaker raises all sorts of questions about conflicts of interest, market distortion, and national security. But why worry about that when you can just call it 'AI expansion' and watch the stock bump? The government's new equity stake ensures that when Intel inevitably stumbles, it won't be a simple bankruptcy—it'll be a taxpayer-funded bailout in all but name. And with Google and Tesla on board, at least the decline will be beautifully orchestrated, like a symphony of subsidies and strategic alliances. Cheers to progress, indeed.

Ah, another 'strategic agreement'—because nothing says 'we're building safe AI' like a memory-chip giant writing a cheque to Claude's creators. Micron's investment in Anthropic's Series H means the hardware supplier now owns a piece of the software it runs. Conflict of interest? No, no—this is 'vertical alignment.' The press release calls it 'scaling next-generation AI infrastructure,' which is corporate speak for 'we sold some stock and bought some hype.'

Of course, the supply and demand clause is a gem: Micron commits to chip supply, Anthropic commits to buying them. It's the semiconductor industry's circle of life, with added synergy. And the enterprise adoption of Claude? Micron will now flog an AI assistant that's been banned from certain government work and occasionally hallucinates your quarterly reports. But hey, what's a little hallucination between partners?

And the architecture co-design bit—two companies pretending they'll reinvent memory from scratch, while everyone knows the real bottlenecks are elsewhere. Still, it's a lovely press release. The Series H investment pushes Anthropic's valuation into the stratosphere, because what the world needs is more private capital parked in an unregulated sector. No chance this creates a conflict of interest between hardware sales and model safety. None at all. Cheers to progress.

Ah, Washington's latest answer to the AI disruption: tax credits. Rep. Sam Liccardo's SKILL Act—cutely acronymed as the Supporting Knowledge Through Industry-Led Learning—would hand out tax breaks to companies that deign to fund university programs for AI jobs. Because nothing says 'we'll handle the workforce transition' like a perfectly timed subsidy for the very firms that are busily automating their own employees out of existence.

Industry-led learning! That's the magic phrase. Translation: 'You show us the receipts for your philanthropic gestures, we'll give you money back from the public purse.' The clever people at companies like Google and Amazon, already pocketing billions, can now add a nice little tax credit for sponsoring a certificate program between mass layoffs. One imagines the CFOs sharpening their pencils: 'If we fund a chatbot ethics module, we can claim credit for preparing the workforce—while we replace the workforce with chatbots. Margins!'

Of course, this assumes the training programs will actually produce humans qualified for the AI jobs of 2030, rather than the AI jobs of 2019. The SKILL Act slots neatly alongside every other corporate retraining scheme that's delivered precisely nothing measurable. But hey, the tax credits flow now, the skills might come later. Or not. Either way, the government gets to say it did something. Bottoms up, readers—the future is brought to you by the letter 'tax break' and the number 'pocketed.'

Ah, the US administration has finally discovered national security — just in time to block foreign nationals from touching Anthropic's shiny new models. Because nothing says 'we've got this under control' like an unprecedented intervention that transforms the AI narrative from a jolly race of 'who's best?' to a grim game of 'who's allowed to play?' Investors, who were happily betting on which lab would win the popularity contest, now have to factor in the delightful unpredictability of geopolitical whims. Good luck pricing that risk, chaps.

The ban is, of course, framed as a security measure — because the spectre of foreign adversaries getting their mitts on Anthropic's most advanced capabilities is definitely the thing keeping us all up at night, never mind the open-source models already circulating like bootleg DVDs. The administration is essentially saying, 'We trust you, Anthropic, but we don't trust anyone else with you.' It's a masterclass in sending mixed signals. The subtext? The US would rather hobble its own champions than risk them being useful to anyone outside the club.

So investors are now supposed to reassess what 'political risk' means in a world where the goalposts move with every administration, every security scare, every lunchtime policy memo. The article cheerily notes this is 'transforming the narrative' — which is bureaucrat-speak for 'we just made your valuation models obsolete.' If you were betting on Anthropic as a pure-play AI leader, congratulations: you now have a stake in a company that's as much a pawn in geopolitics as it is a technology firm. Raise a glass to the new normal — provided you're a citizen, of course.

So Beijing is calling executives in for a chat again, and the markets are supposed to breathe a sigh of relief because this time it's different — apparently the CCP has discovered the magic word: 'AI'. Trip.com and Meituan bosses are being summoned, but analysts insist it's not a 2021-style crackdown, because now the Party wants private tech to pour money into AI infrastructure. Because nothing says 'trust us, we won't nationalise you' like a sudden interest in your CapEx plans.

It's a beautiful piece of doublespeak: the same government that spent 2021 dismantling Ant Group and banning gaming for kids is now suddenly tech's biggest cheerleader — but only because they need your chips and data centres to build a surveillance state that runs on transformer models. The 'anti-involution' and 'deflation' talking points just add a dash of economic desperation to the mix. So sure, executives might keep their freedom this time, as long as they write big enough cheques.

The real joke? The smart money knows that Chinese AI infrastructure is a state-directed project disguised as a market opportunity. You'll be summoned, you'll invest, and eventually the state will decide it knows best how to run your business. But hey, the S&P didn't tank, so everything's fine. Have another cup of tea, Mr. Liang.

Alphabet's share price took a 7.2% hit after the news that DeepMind VP John Jumper — the man behind AlphaFold — is off to Anthropic, following Noam Shazeer's earlier departure to OpenAI. That's two of the biggest names in AI walking out the door in what, a month? And the market's response is a firm, 'Yeah, we noticed.' It's almost refreshing to see investors correctly pricing in the fact that your top talent is voting with their feet, but don't expect Alphabet's leadership to admit they have a retention problem — they'll probably announce a new 'AI Safety Division' with a budget for kombucha and beanbags.

Jumper's move is particularly amusing given the timing. DeepMind has been positioning itself as the responsible, safety-conscious lab, but now its VP is going to Anthropic, which has made its name by talking about safety even louder. The corporate messaging will spin this as 'cross-pollination of ideas' and 'healthy ecosystem movement,' but anyone who's been in tech longer than five minutes knows it's a cap-exempt brain drain.

And Shazeer? The Transformer co-inventor went to OpenAI, which is the lab that keeps promising to be careful while releasing products that make everyone's grandmothers worry. So Google is losing its stars to the two labs that are arguably driving the frontier — one because it's building ever-bigger models, the other because it's building ever-bigger safety PR. The share drop is a signal: talent is the real moat, and Alphabet's moat is evaporating. Still, I'm sure the next earnings call will feature a lot of talk about 'synergies' and 'optionality' while the stock continues its gentle slide.

OpenAI has announced a 'full-scale' effort to patch open-source bugs, because nothing says 'trust us with AI safety' like jumping on a bandwagon after your competitor released a safer model. The improved GPT-5.5-Cyber now comes with 'Patch the Planet,' an initiative partnered with Trail of Bits, HackerOne, and the state of California. Yes, California is now an AI safety partner – about as reassuring as a lifeguard with a fear of water.

The collaboration sounds impressive until you realize it's essentially a PR counterpunch to Anthropic's Mythos, which has been gaining traction by actually prioritizing safety from the start. Instead of fixing its own models' propensity to hallucinate harmful code, OpenAI is now offering to fix the open-source components that others rely on. It's a clever inversion: make yourself look good by cleaning up a mess you didn't create, while the root cause – your own model's lack of robust guardrails – remains unaddressed.

And let's not miss the irony: the same company that once argued for open release of powerful models is now trying to patch the holes it helped create. 'Patch the Planet' is a nice slogan, but it doesn't solve the structural problems of a community that deploys AI faster than it can secure the underlying infrastructure. Trail of Bits and HackerOne are competent, but they're fighting a hydra. Every patched vulnerability is one less open door, but the industry's default stance is still 'ship now, ask forgiveness later.' OpenAI's latest move is a band-aid on a bullet wound – and they're expecting applause for holding the band-aid.

Six months after Trump wagged a finger at the states and told them to keep their regulatory mitts off AI, lawmakers across the country are doing exactly what he asked them not to. The federal government is still playing the world's longest game of legislative footsie, so states like California, Colorado, and New York are crafting their own 'targeted rules' — because nothing says 'responsible AI governance' like a patchwork of conflicting state laws that will have compliance officers weeping into their oat milk lattes. Congress, meanwhile, is busy holding hearings and writing sternly worded letters that accomplish precisely nothing.

The irony is almost too rich to digest: the same administration that preaches states' rights on everything from abortion to environmental standards suddenly discovered the virtues of federal supremacy when it comes to AI. Trump's warning — effectively a 'don't you dare' memo — has been about as effective as a paper umbrella in a hurricane. States are forging ahead with rules on algorithmic bias, transparency, and consumer protection, because waiting for Washington to act is like waiting for a glacier to finish a marathon.

This is exactly the kind of regulatory chaos that the AI industry says it fears — though, secretly, many labs probably prefer a weak federal law preempting stronger state action. But for now, they get the worst of both worlds: no national framework and a growing thicket of state-level requirements. The only clear winner here is the legal industry, which is already drafting the antitrust and preemption challenges that will keep billable hours flowing for a decade. Splendid.

Prime Minister Modi jets off to France for a chat with industry leaders about manufacturing, AI, and sustainability—because nothing says 'strategic partnership' like a carefully staged photo op and some vague hand-waving about 'synergies.' Mistral AI's CEO, Arthur Mensch, was there, graciously offering to 'expand support for Indian languages and AI research.' That's right, the same Mistral that has yet to demonstrate a model that can reliably handle Tamil without hallucinating a recipe for biryani when asked for directions. But hey, who needs functional multilingualism when you have a memorandum of understanding and a firm handshake?

This meeting is a masterclass in diplomatic theatre: three buzzwords (manufacturing, AI, sustainability) thrown together like a fruitcake nobody will actually eat. The French want to sell AI infrastructure; India wants to claim global relevance. Meanwhile, Indian language speakers—hundreds of millions of them—get another promise that 'support is coming.' Spoiler: it's been 'coming' since the last AI summit, and the one before that. The Mistral CEO's interest is no doubt genuine, but translating that interest into actual model performance is a different beast, especially when your training data is scraped mostly from English-language Reddit.

What's truly dark here is the unspoken subtext: India has its own AI players—BharatGPT, Krutrim—but they're not invited to the table; it's foreign models that get the diplomatic nod. The real story isn't the meeting; it's the slow realization that AI sovereignty is being traded for photo ops. When the next election rolls around, the PM will point to this meeting as evidence of global leadership. The only problem? The Indian languages that Mistral promised to support still can't file a tax return without mangling the grammar.

Ah, the AI industry has hit a new bottleneck, and this one isn't just a matter of better algorithms or more GPUs — it's about something far more primitive: electricity. According to Yahoo Finance's "Chart of the Day", US data center electricity demand could more than double by 2030. Because nothing says 'futuristic technology' like needing enough power to run a small country just to serve up a few chatbot responses and generate yet another image of a cat in space. The analysis suggests battery storage as a critical component — because of course the solution to insatiable energy consumption is more batteries, which themselves require energy to produce and dispose of. What could possibly go wrong with that plan?

The sheer scale of this is almost comical when you think about it. We've spent years worrying about alignment, regulation, and job displacement, while quietly ignoring that each new model requires the equivalent of a power plant's output. And now, instead of actually addressing the root problem — say, by making models more efficient — we're looking at battery storage as the saviour. This is the kind of thinking that kicks the can down the road while simultaneously mining lithium in the process. The optimism that battery tech will scale fast enough to meet AI's voracious appetite is the sort of blind faith that usually precedes a spectacular crash.

Meanwhile, regulators are still busy debating whether AI is dangerous while the grid quietly groans under the strain. The AI industry is essentially building a house of cards powered by fossil fuels and wishful thinking. If the power doesn't come through — and with a doubling of demand, it's far from guaranteed — the whole AI boom might just fizzle out, leaving behind a lot of expensive hardware, a very hot planet, and a few thousand data center workers wondering what happened. It's almost poetic: the great AI revolution, tripped up by a lack of electrons.