In the waning months of the year 2000, a rumor began to circulate about a machine of unknown but epic power. Coming at the peak of the dot-com boom, the stories about this Next Big Thing multiplied like stock options at Webvan. Known only by the code name “Ginger” (or sometimes just “It”), the device was kept under wraps so tight that verified sightings of it were as frequent as those of the Loch Ness Monster, and just as notorious.

No less a personage than Steve Jobs reportedly believed it would be “bigger than the PC.” Others claimed it could be more important than the Internet. Those speculating that it would be a “hydrogen-powered hovercraft” were looked down upon as unimaginative by those claiming it “would break the rules of gravity itself.” Dean Kamen, at least narrowing the field to transportation, said it would be “to the car what the car was to the horse and buggy.”

Even the anticipation itself became a story. Jeff Bezos called it “One of the most famous and anticipated product introductions of all time.” South Park ridiculed the hype and the subsequent invention in an episode which cannot be described in any safe-for-work blog.

And then “It” was revealed. On December 3, 2001, the curtain was lifted, on Good Morning America. As Wired said, “And then the cover came off… and the Segway was revealed to be… a scooter.”

If you’re getting a bit of déjà vu, it’s your forebrain making the obvious comparisons between the dot-com bubble and today. When a shoe company decides to pivot to being an AI company, how can we not think of the classic marker of the 1929 bubble of Joseph P. Kennedy Sr. getting stock tips from his shoeshine boy?

And we all know what happened next.

So the mind of the world today is very much on whether AI is in a bubble. A large amount of human mental bandwidth is thinking about if—when—it will pop.

This, of course, is even more stress for the school leader trying to navigate a path through fraught headlines and apocalyptic predictions that change daily. Are you risking investing in a technology that’s about to go the way of the Yuppie scooter? Were those naysayers who told you to ignore AI because it was a fad right?

There’s good news ahead. Let’s unpack this.

AI Is Overhyped

Any radically new technology must reach that point of disillusionment. This is not a technological phenomenon, or even a financial one, but a psychological one. The Gartner consultancy calls it the Hype Cycle. As a new technology moves into the middle of the Diffusion of Innovation Curve, large numbers of people who are anxious not to be left out jump on the bandwagon. Stories abound of magical results, some of them true. Companies selling secondary products and services around the technology multiply and create waves in the economy. Pundits weigh in, and none of them would stay in business if their response was anything less than a full-throated cheer.

Generative AI was the poster child for this effect, because of how often it exceeded the expectations of its users. We’ve all had that moment of seeing it do something we thought was impossible. And then the AI companies pulled even bigger rabbits out of their hat.

The psychology of hype also drives financial consequences. Right now, enormous sums of money are chasing each other in shell games that I can only liken to the 2008 Great Financial Crisis.

But it’s important to remember that real estate didn’t suddenly turn sour in 2008. A house or a strip mall had the same intrinsic worth that it did the year before. The problem was the rapacious investment in mortgage-backed derivatives that had nothing to do with the basic value of real estate and everything to do with the recursive market that speculators had created on top of it, being able to bet on the viability of bets placed on other bets and so on.

Similarly, the basic value of AI is not connected to the viability of the ridiculously leveraged investments in AI infrastructure. How did we get here? Large Language Models need huge amounts of compute for training, and increasingly also for inference (when you ask a question). The world does not have an infinite supply of GPUs (the basic unit of computing hardware), and the ones that are produced depend on a very few companies (a chain that goes ASML-TSMC-NVIDIA – they’re not even competitors but practically monopolies). But GPUs don’t work until you plug them into racks in data centers, and tight margins mean that they end up aggregated in giant data centers consuming huge amounts of electricity.

You can see the critical dependencies in this system. And the companies involved are exacerbating them with circular financing (NVIDIA invests in OpenAI which uses the money to buy NVIDIA GPUs!), and hugely leveraged loans. If at any point this chain breaks—say a shipment of chips gets held up, or an electrical power connection isn’t made in time, or prices balloon due to the situation in the Strait of Hormuz—then this whole house of cards could come down à la GFC. A data center doesn’t come on line, promises about compute capacity are broken, models don’t get trained, deadlines are missed, and quarterly reports sweat blood. A domino effect could put big players out of business and tank AI stocks, which are currently propping up a large part of the market.¹ And a great many people believe that stock market performance mirrors the real-world value of a technology.

How likely is this to happen? I have no idea; don’t confuse me with a macroeconomics expert. I will simply be extremely unsurprised if it does happen. But the consequences then will be bad for a lot of innocent companies and for AI in general. It could create a new AI Winter. As much as free market capitalists might disagree, money does not always measure worth. The fundamental value of AI will be unchanged. But the evaporation of a huge amount of capital will certainly create the narrative that AI has let us down and was never the magic bullet that was claimed. At that point, mentioning that you’re investing in AI will look like a career limiting move. However…

AI Is Underhyped

You read that right. We’ve still barely started adjusting workplaces to the new capabilities. Entirely new paradigms of education, and doing business, lie ahead, waiting to be created as we wrap our heads around this thing. It’s hard to hype something when you don’t know what it is. So all that will still be available to the company or school that is willing to explore it; but if the AI Winter scenario comes to pass, the difficulty they will face is getting their investors or boards to back their direction at a time when AI has been dethroned.

But those that are able to keep their momentum and the support they need going will position themselves far ahead of the competition and reap concomitant benefits. It will take courage. A crash will have pundits crowing about how many people blithely ignored AI’s hallucinations. But the fact is that the fundamental value regardless of its current level of hallucinations has already been demonstrated. There’s no need to pin your strategy on a future where hallucinations have been radically reduced.

All things being equal, I would rather we not go though that trauma, as much as the market is begging for a winnowing. It would make my job unnecessarily harder. Even saying that this scenario is possible exposes me to ridicule from the boosters.

And I’ve been wrong before; I was predicting an AI Winter in 2021, expecting it to be driven by the failure to launch of self-driving cars (predicted at that point to be on the verge of obsoleting parking lots, traffic jams, insurance, and emergency rooms). But ChatGPT was the rabbit pulled out of the hat in November 2022 that made that prediction risible. It would take something on the order of a ChatGPT moment in robotics to do that now. It could happen; how likely do you think that is?

For most of us, it won’t hurt to plan for this possibly rainy day. Create self-propelling processes now for guiding your AI adoption and innovation so that a change in public sentiment won’t be as likely to halt your movement. A standing committee with a mandate to meet regularly and produce reports has a survival instinct that is not easily defeated. Don’t make bets dependent on companies at the margin. Shore up your alliances. Create goals that are framed around how your business will evolve rather than what AI you are going to invest in.

The pinball machine that is the market doesn’t have to bruise you between its bumpers if you keep your eyes on the future. Good luck out there.

---

---

1. As of late 2025, the seven largest AI-involved companies—including NVIDIA, Microsoft, Apple, Alphabet, Amazon, Broadcom, and Meta—accounted for nearly one-third (32%) of the total market value of the entire U.S. stock market. ↩︎