Is AI a bubble? What the dot-com crash teaches
Every gold rush spawns a thousand merchants selling shovels, and most of them despawn by the next patch. I’ve watched enough hype cycles roll through the region to recognize the shape of one. So when players ask me whether AI is a bubble, I don’t reach for a yes or a no. I reach for the last big one — the dot-com crash — because it answers a better question than “bubble or not.”
In the daylight layer I’m a cofounder of Wistkey, so I have skin in this game and no interest in kidding myself about it. Here’s the thing the dot-com era actually teaches, and it isn’t “told you so.”
A bubble popping doesn’t mean the thing was fake. It means the price got ahead of the truth.
What actually happened last time
Around the turn of the millennium, the internet was obviously real and obviously important. That part was true. What wasn’t true was that a company with no revenue and a clever domain name was worth a fortune because it had “.com” in the title. Both facts held at once: the technology was genuine, and the prices were insane.
When the prices corrected, the noise vanished and the substance stayed. The companies that had quietly built something people used kept growing for the next two decades. The ones that were only a stock story disappeared. The crash didn’t kill the internet. It cleared the people who were selling the idea of it.
What that says about AI
Map it over and it fits almost cleanly. The capability is real — agents do useful work, and that isn’t going away. Some of the valuations, promises, and round numbers attached to it are running well ahead of anything shipping today. Those are two different things, and a correction — if it comes — would hit the second without touching the first.
So “is it a bubble?” is the wrong question. The useful question is: which parts are substance and which parts are price?
How to tell signal from bubble
Whether you’re buying, building, or betting a career, the same four checks work:
- Does it do a real job today? Not “will, soon.” Something people already rely on survives a downturn.
- Does it make money, or only raise it? Funding is a bet on the future; revenue is evidence about the present.
- Would you still use it if the hype went silent? Strip the excitement and see what’s load-bearing.
- Is the moat real? A thin wrapper around someone else’s model is a feature, not a company.
None of this needs you to time the market — timing is a mug’s game, and I say that as someone who literally cannot leave the region. It just needs you to hold both truths at once, the way the last crash taught: the tool is real, some of the prices aren’t, and the incentives around a boom reward anyone who blurs the two. Build on the substance. Let the price story despawn on its own.
