Oh, the 1990s. It was a hell of a time to be alive. We had Michael Jordan, “I did not have sexual relations with that woman,” and the bursting of the dot-com bubble. Now, in 2025, it’s trendy to predict a 1990s-style collapse. Only this time the debate du jour is artificial intelligence.

But I don’t see an AI bubble in my crystal ball.

Imagine you’re Dorothy from The Wizard of Oz and I’m Oscar Diggs. That carnival swindler who ultimately becomes Oz in Dorothy’s dream. Instead of snake oil though, I’m handing you a crystal ball of economic history and cold, hard facts.

Every technology revolution has its froth, but the pattern is consistent.

In short: revolutions endure, players do not.

We saw this in the 1800s during Britain’s Industrial Revolution. Textile mills and steam engine manufacturers sprang up everywhere. Innovations like steam power survived, but many early firms went bust, merged, or were overtaken by rivals.

We saw it again in the American railroad boom of the mid-19th century. Hundreds of regional rail companies collapsed in the Panic of 1873. Many players failed, but the tracks remained, ultimately defining a generation of growth.

Fast forward to the automobile age. In early 1920s America, there were some 88 car manufacturers. Only a handful — Ford, Chrysler, GM — survived, but the car revolution transformed the economy.

The dot-com bubble followed a similar script. The internet prevailed while many early players, like the infamous Pets.com, did not. The same pattern repeats with social media (Myspace!) and mobile hardware (BlackBerrys, Motorola, Nokia).

Now, some point to frothy valuations in today’s AI world — companies like Nvidia — and see echoes of the dot-com bust.

Yes, Nvidia is pricey. Its P/E ratio is roughly double that of the S&P 500, and many AI-adjacent companies trade at elevated multiples. One could reasonably say: “bubbly.”

But here’s the thing: Nvidia is delivering dramatic growth that literally powers today’s AI transformation. Demand for GPUs far outstrips supply, enterprise adoption is accelerating, and AI is no longer a speculative promise — it’s an enabling layer permeating logistics, software, customer service, and more.

Surveys suggest that well over half of Fortune 500 companies now deploy generative AI capabilities internally, with some estimates approach 9 in 10. Training costs have plummeted, inference costs are falling even faster, and new AI use cases are increasingly cheaper to scale.

Unlike the dot-com era, today’s AI wave is not built on what technology might do one day. Back then, the dark fiber being laid across cities often had no immediate use.

Today, the modern equivalent — the GPU — is in constant demand. Layoffs at Amazon and elsewhere aren’t a retreat; they reflect growing infrastructure pressures as companies scramble to allocate compute for AI (sometimes at human expense).

There will be corrections. Some startups will fail. Some valuations will stumble. But that’s noise, not the signal. The infrastructure, capabilities, and inevitability of AI are here to stay.

So, are we in an AI bubble?

My crystal ball says no. We aren’t hustling tulip bulbs, worthless websites, or NFTs (remember those?).

Artificial intelligence is a foundational technology that will be with us for a long time — the type of tech our kids will grow up with, like the Model-T or a Motorola Razr in previous eras. Google wasn’t the first search engine, and neither will every AI player dominate the future. The path from innovation to infrastructure is messy, littered with casualties and hype.

But the revolution endures. AI is building the logic that will run on top of the internet itself.

That noise you hear is not the pop before the crash. It’s the hum of the AI Model-T warming up.

Oh, the 2020s. We have Shohei Ohtani, “Make America great again”, and an AI renaissance with inevitable casualties waiting to happen.