I think the AI bubble is about to burst, and here’s why.

I am a historian, and this is primarily a history blog. That said, I have, especially in the last year or so, occasionally written pieces on tech and culture, specifically AI. You could say I have some opinions on it, and they’re not totally irrelevant to the historical process, as in December 2025 I made a three-hour video that compared the “cult” around AI to previous historical developments stretching back to the railroad boom of the 19th century. I detest AI and would like nothing more than to wake up tomorrow morning and find all trace of it gone from our economy, culture and technological landscape. While that definitely won’t happen, instead of my usual angry rants about AI, in this article I have an optimistic message to deliver: I think AI is about to die, for all intents and purposes, and while it won’t go away entirely, the madness that has seized our society regarding this bizarre and largely useless technology may soon be receding. The reason why has to do with insurance.

Economically speaking, AI is a bubble. Very few people deny that now, although plenty among the dwindling species of enthusiastic AI boosters—most of them on the take for fascistic Silicon Valley billionaires—are eager to rationalize why the popping of the AI bubble isn’t a bad thing, or isn’t as bad, or will be just like the dot-com bust (which was bad, I lived through it), or something. The work of tech blogger Edward Zitron has been exhaustive in documenting the alarming house-of-cards economics of the AI and the broader tech industries. If you need any convincing that OpenAI and Anthropic are pump-and-dump scams or that the “hyperscalers” in Big Tech are simply promoting the illusion that they’re profitable by basically trading the same dog-eared dollar bill back and forth among each other and pretending that it’s income, Zitron’s work will quickly educate you. There is, however, one big fact about AI that I haven’t heard Zitron talk about very much, but which seems extremely significant to me: major insurance companies have recently begun excluding almost all AI-related risks from their commercial coverage. This sounds arcane and dry as dust, but it’s actually huge. If you’re looking desperately for the needle about to be plunged into the overinflated balloon of AI, this is probably it.

What’s going on takes a word of explanation. Almost everything businesses do is insured to one degree or another. Indeed, business is pretty much impossible without insurance of some kind. Up until fairly recently, AI-related risks were thought to be part and parcel of businesses’ general operating insurance. For example, there’s an insurance product called “Tech E&O”—Errors & Omissions—which covers a company for losses caused by data breaches, such as if a hacker breaks into a company’s system and sells a bunch of their customers’ credit card numbers on the dark web. However, insurance companies have recently been introducing exclusions into their policies that limit their exposure if losses are related to AI. So, if a company’s data breach occurs because it’s AI “agent” decides to release all its customers’ credit card numbers directly to those very nice Russian mobsters in St. Petersburg, that is suddenly not covered.

From an article on business news aggregator JDSupra:

“Several insurers have introduced broad “absolute” AI exclusions in these policies, purporting to exclude coverage for claims arising out of any use, development, or deployment of artificial intelligence. These exclusions are often drafted in expansive terms and may capture a wide range of ordinary business activities involving AI. Notably, similar exclusionary language is now appearing across multiple coverage lines, including D&O, EPLI, and fiduciary liability policies.”

This is not an isolated incident. Here is another article about the phenomenon, in which insurance industry analysts predict broad adoption of AI exclusions by insurance companies.

As some of you know, I used to be a lawyer. In the field of business law, the opinions, behavior and policies of insurance companies was often the final word—if something was not insured or had questionable coverage, the safe bet, as an attorney, was always to advise your client not to do it. If, as seems to be the case, major insurance companies are refusing AI risks on a very broad scale, I can’t think of a more significant bellwether of the degree to which AI will become “business as usual” in the business world. In a word, it won’t. It can’t.

It would be one thing if AI was a useful product with multiple, economically beneficial use cases that self-demonstrate AI’s utility and its desirability in business operations. If that was true of AI, the big insurers, as a matter of market necessity, would easily find a way to price AI-related risks and they would be more than happy to sell insurance to companies, perhaps even at a premium, willing to take those risks. Let me say emphatically that this is absolutely not the case with AI. Indeed, AI has very few compelling use cases, at least few that anyone is willing to pay for. AI is largely a fake industry, buoyed by media hype and a strange and puzzling kind of collective groupthink among corporate managers that borders on cultism. Look, for example, at this article, unusually frank for the tech industry, written by blogger Nik Suresh, who runs a data management consulting firm called Hermit Tech. This article plumbs the incredible depth of AI-caused psychosis among business leaders, but also highlights just how useless and ridiculous AI actually is:

“All of the AI projects we have observed as a team are failing. Every single one – we have seen 0% success in a year and a half, not only amongst projects we have been asked to participate in, but even within projects that we have observed in passing while doing totally unrelated work. Even if you grant that AI tooling accelerates specific workloads, the method and scale of the current investments is senseless…I am very confident that almost every report at a company about “massive AI productivity gains” is untrue as a matter of brute fact. Even if some companies are seeing clear gains, this is the exception, not the norm.”

Put this together with the undeniable indicators—we’re back to Zitron here—that the two biggest AI labs in the world, OpenAI and Anthropic, are on the brink of financial failure and when they begin to collapse they’ll take down basically the entire tech industry, and a significant chunk of the financial sector that has invested heavily in them. This is absolutely not sustainable. Basically, companies are acting on irrational “vibes” to adopt an essentially useless product that causes significant harm, in terms of intellectual property losses, data breaches and tech disasters, chatbot psychosis, and horrendous economic bloodletting to invest in data centers that will never actually be built. Insurance companies won’t insure against these harms. When the money dries up and the “vibes” recede, and the companies that have adopted AI so vociferously have been shown up as gullible chumps, AI will be functionally dead, at least as “the next big thing” or “the inevitable future.”

Basically what has happened is, the big tech giants, realizing a few years ago that they’d reached the end of their innovation runway and they simply could not provide new innovations that would sustain the same kind of explosive growth they’ve seen in the past 30 years, bet rather desperately on the AI idea, hoping it would be that next innovation, and they lost. This was the thesis of my article from 2025 about how tech has simply run out of ideas. AI was, and always has been, simply a scam. It’s not an innovation, or a revolution. It’s glorified auto-correct. And that was what the tech giants, and the financial markets, have gone all-in on in the past five years? They bet the farm on glorified auto-correct? Yes, they did—because they had nothing else.

The AI bubble is beginning to collapse. I’m sure of it. The reaction of the insurance industry, telling AI-hungry businesses to talk to the hand when it comes to losses caused by their glorified auto-correct, is the canary in the coal mine for AI, and it’s just dropped to the bottom of the cage. While I’m not looking forward to the economic and human pain that will result from watching the AI bubble implode, I’m eager for our society to move past this destructive and senseless distraction and get on to something real and meaningful.

I hate AI. I detest it. Like many people, I had a period of dubious flirtation with it, but I’ve come to hate everything about it. I hate its arrogance, its obsequiousness, its cluelessness, its fatuousness. I hate the gauzy nothingness of AI-generated images and the bland vapidity of AI-generated text. I hate the fascism and dehumanization behind it and at its rotten core. While it’s a rare day that you’ll find me cheering on organizations as craven and amoral as insurance companies, I have to say, kudos to the insurers for knowing a bad bet when they see one. AI is a bad, bad bet. Pretty soon, those who’ve bet on it are going to have to show their cards. If they’ve got anything higher than a pair of deuces I’d be very surprised.

The Value Proposition

Why should you be reading this blog, or receiving it as a newsletter? This is why.

☕ If you appreciate what I do, buy me a virtual coffee from time-to-time to support my work. I know it seems small, but it truly helps.

📖 You could also buy my newest book.

🎓 Like learning? Find out what courses I’m currently offering at my website.

📽 More the visual type? Here is my YouTube channel with tons of free history videos.