The AI Bubble - We've Seen This Movie Before

I've been laid off. It has happened before. My last layoff was back in 2001, and came long with another famous bubble burst, the "dot-com" bubble burst. I was less than 5 years out of school with my Electrical Engineering degree and a great paying job at a great company. Most of the people I worked with were already millionaires on paper - their company stock and stock options were gold. Then it wasn't, and none of us were millionaires.

Senior leadership hadn't seen this coming. They just built a brand new, huge, and expensive headquarters building that they were moving a bunch of us into from various older facilities. I think I worked in it for all of a month before my layoff. In retrospect it was easy to see what had happened - one of our biggest competitors was cooking the books (claiming sales numbers without properly accounting for the liability that their equipment was being sold on credit, not sold for cash) and suddenly there was a glut of "used" equipment on the market that hadn't come out of the shrink wrap, and was selling for pennies on the dollar.

In the years following, we've seen most of the players in the dot-com bust consolidated or disappeared. Amazon is now synonymous with online shopping, not just buying the occasional book. Brands like Nortel, Lucent, and Tellabs are, well, footnotes - parts of other companies if they even exist at all. The "wild, wild west" that the commercial Internet was has become a two-tiered system of the "haves and have-nots," depending on how good they were at things like SEO and delivering views back to their advertisers. We communicate with our friends not via phone calls and emails but with social media posts - often that are visible to the entire world - that exist on one of about 6 sites.

What happened? Well, I'd say there were three notable actions that occurred to create and then "pop" the dot-com bubble:

1. A mad rush to transform how business is done. Every brick-and-mortar was under siege from the online world. driven by the beliefs (sometimes quite dubious) that somehow doing business online was so much faster, cheaper, and better that the last one to do so would close up shop within 6 months - at least, that's what it felt like. While it was true that this change was coming, it seemed to be driven by hyperbole, hopium, and FOMO more-so than actual business acumen and plans.
2. A circular economy in one major industry vertical. In order to be an online powerhouse, the side-streets of the Internet all had to be upgraded to superhighways all at once. To do this, telecom hardware providers had to pump out huge volumes of hardware, and every aspiring dot-com company had to buy all this expensive hardware (not to mention all the backbone and distribution players), building a vicious cycle of codependence between these (essentially) two parties to support astronomic growth.
3. Indiscriminate funding of any company that said the magic words "e-commerce" or "subscription." It almost didn't matter where you were in the country, you couldn't throw a stone without hitting a dot-com startup employing anywhere between 10 and 200 fresh-faced true believers, most of whom probably couldn't have explained the business plan that was going to eventually lead to profitable income - let alone their founders. But somehow they got funded anyway.

So when any one of these three factors got disrupted and unbalanced, a whole house of cards came tumbling down. It didn't happen quite all at once of course. One of the first things that happened seemed innocuous enough. Lucent (in particular) started selling more of their equipment on credit plans that they themselves backed, and then was using what are now unlawful accounting practices to count those sales as finalized and realized profits when in fact they were still liabilities. This led to the market being flooded with telecom equipment it couldn't afford, and as more of the companies racing to define this new era started folding, lots of barely used - or even still in the original packaging - telecom equipment was suddenly sitting around with nothing to do. The telecom hardware industry crashed as people could buy this unused hardware for pennies on the dollar. So now we had dot-com companies starting to drop away, and the telecom hardware industry was hit extremely hard at the same time, finding that what was for decades a cornerstone of American industry going through the leanest times in memory.

I'll admit, I'm no historian, I'm not an economic expert, and I'm no investment guru, so those observations are certainly open to interpretation, nuanced investigation, and debate. That said, I think they hold true at a macro level, and I think anybody who was "around" that industry at the time can look back and confirm a similar outlook on what happened.

What's perhaps even more interesting is how all the things that we were so eager for during the dot-com era have essentially come true: most businesses have transformed into online entities, e-commerce dominates, and anything that "can" be a subscription essentially is. It just took a good 10-20 years longer than people had hoped in 2001.

All this has happened before, and all of it will happen again. - Cylon Dogmatic Belief

Do you see parallels yet to what's happening with GenAI? I certainly do.

1. A mad rush to transform how business is done. Every company seems to be convinced that they need GenAI to replace people, and that somehow people want to interact with GenAI as they interact with their companies, and that their financial viability depends on making this transformation happen before the end of 2026.
2. A circular economy in one major industry vertical. In order to build AI machines, specific AI focused chipsets are needed, and AI forward companies are buying them up like nobody's business. Had a look at NVIDIA's stock price recently? Were you aware that 40% of their Q2, 2025 revenue comes from only two customers - customers they refuse to identify in their financial reports? Did you know the quarter before it was 4 unnamed customers that were responsible for 54% of that quarter's revenue? Five years ago NVIDIA's top customers were workstation and video card manufacturers with no single customer responsible for anywhere near 20% of revenue, now suddenly it is a hand-full of (undoubtedly) AI companies. What happens when one or two of these companies falter? How much NVIDIA hardware will suddenly be available on the 2nd hand market, depressing sales of new hardware, not to mention the loss of anywhere from 10 to 20% of NVIDIA's expected revenue per now defunct customer? How will other chipset manufacturers be hit by that same glut of 2nd hand hardware?
3. Indiscriminate funding of any company that said the magic words "LLM" or "Agentic." I'd be happy to have you pick any 10 random "AI" companies that have popped up in the past 24 months and ask you to describe their business plan. I promise that less than 50% of them will make any sense to you or me. But somehow they got funding.

But there's an interesting difference this time around. Companies seem to be "pre-firing" employees, based on the savings in work they "expect" to get from AI. (I wonder if an AI based financial model convinced them of this?) It's also interesting to me to see who's running the big companies that are positioning themselves at the front of this new bubble: many are the companies (and leaders of them) who managed to come out on top during the dot-com bust. I do hope their survivorship bias isn't getting in the way of sound decisions.

Am I being cynical? I hope so, it would be wonderful if I'm just worrying for nothing. Unfortunately I don't believe I am. But even as I say that, I remind myself that all the things businesses were focused on doing during the dot-com cycle ended up happening over a longer timeline. I find myself believing that a lot of the "promise" of the GenAI cycle can come true over the longer timeline as well. What remains to be seen is if we'll have the patience to do it right the first time, or if we'll burst our own bubble and have to do it right after relearning the dot-com lessons all over again.

I hope you don't get splashed by a popping bubble.