Why bubbles are just a part of progress
The tech industry and the market more broadly seem fixated on a single question these days: Is AI a bubble? As usual, the real answer isn’t as simple as a straightforward “yes” or “no.”
There are great similarities with the early 2000s, of course, when somewhere between Netscape and Napster, it felt as if the world was reinventing itself. Everybody talked about The New Economy. The stock market valued companies with no profits but with a homepage as if they were the future. At the peak of that first internet wave, the most valuable company on Earth wasn’t Google or Amazon: it was Cisco. It built the routers that formed the backbone of the internet, and in March 2000 it crowned the global market.
The logic was simple: whoever owned the infrastructure owned the future.
The same, but different
Twenty-five years later we’re pretty much experiencing the same situation. Cisco’s old role as hardware supplier to a global hype is now played by Nvidia. Its GPUs supply the oxygen for the artificial intelligence boom. The curve looks eerily familiar: a vertical rise into the stratosphere, investors elbowing their way to catch the last train, and a CEO proclaiming that demand remains “unprecedentedly strong.”
But there’s a key difference.
The bubble back then was mostly air: websites with flimsy business plans and imaginary clicks that would somehow become money. Today’s bubble is heavier, literally. It sits on millions of chips, data centers, gigawatt-hungry facilities, and mountains of cooling infrastructure. The last revolution pushed bits. This one pushes atoms as well. And that makes it slower, pricier, and far more energy intensive. The dotcom bubble popped on air; this one could burst on energy.
And yet, the pattern repeats. Then, everything became “dotcom.” Now, everything is “AI.” Companies slap “powered by AI” onto their products just as they once glued “.com” to their names. Investors wax lyrical about “inevitable exponential growth.” And, once again, there’s the unwavering belief that if you hesitate today, you’ll be irrelevant tomorrow.
Infrastructure as the backbone
What strikes me most is that, once again, infrastructure feeds the hype. Cisco built routers. Nvidia builds chips. Neither produces content; they build the highways it travels on. Whoever sold routers back then sold growth. Whoever sells GPUs today sells power, literally and geopolitically. Nvidia is even forbidden from shipping its latest Blackwell chips to China. Though, at the time of writing this piece (things change so fast that the situation might be different if you read this) the US was considering allowing some more advanced chips like the H200 to be sold to China again. AI is no longer a pure software story. It’s a story of infrastructure, energy, and strategy. The new arms race runs on silicon.
And of course, just like back then, we hear the mantra: This time it’s different.
It never is, though.
Technology changes; psychology doesn’t. Bubbles are the price we pay for imagination. They overreach, but they also leave behind the infrastructure that enables the next wave. Out of the dotcom rubble eventually emerged Google, Amazon, and the entire social-media revolution. Something lasting will come out of the AI bubble too. We just don’t know what yet.
One player, however, survived both situations: Microsoft. In 2000 it was the wounded dinosaur of the first wave, written off after an antitrust case. Today it is again one of the world’s most valuable companies. Not because it invents every new hype, but because it reinvents itself: from operating systems to cloud, from cloud to AI platforms. While others merely adapted, Microsoft transformed. That is the difference between a survivor and a casualty of a bubble.
Survival of the Phoenix
So yes, I do indeed believe that we are experiencing a bubble. That’s just what happens during every technological revolution.
In fact, every generation deserves its own bubble. They are annoying, risky, and sometimes painful, but they are also part of progress. The bubble always bursts. The only question is what will remain: routers or GPUs, data or power, dreams or action? And perhaps another Microsoft-like Phoenix company, one that survives everything and emerges stronger than before.
