The AI Hype Isn't Dying. It's Growing Up.

Every few weeks now, somebody asks me some version of the same question, usually with a little relief in their voice, which is whether the whole AI thing is finally starting to cool off. My honest answer is that yes, something is absolutely cooling off, but it isn't the part they think it is, and the part that's actually fading is the part that probably should.

What's deflating is the froth, the pure hype, that strange couple of years we just lived through where the mere association with AI was worth real money whether or not anything underneath had actually changed. You could watch it happen right in the numbers. Not long ago, around half of the S&P 500 were name-dropping AI on their earnings calls, which was roughly a fivefold jump in two years, and for a while that name-drop genuinely paid, because the companies that mentioned AI were beating the ones that stayed quiet by a wide margin. But watch what happened to that gap over the following year, because it tells the whole story. The outperformance shrank from something like fourteen points against six, down to eight against four, and then all the way down to about one point against a third of a point, which is to say that simply uttering the word stopped moving your stock. The market got tired of the magic word.

We've seen this exact movie before, and it's worth remembering how it played out. Back in 1999, three finance researchers studied ninety-five companies that did nothing more than add ".com" to their names, and they found that this purely cosmetic change produced cumulative abnormal returns of around seventy-four percent in the days surrounding the announcement, and it didn't matter in the slightest whether the company had any real internet business, because the association alone was enough. We just spent two years living through the AI remake of that study, and it got silly enough that the SEC had to start bringing "AI washing" enforcement cases against firms that were exaggerating what their technology could actually do. When the regulators have to invent a term for your hype, you're somewhere near the top of the cycle.

So if the question is whether that gold-rush energy is fading, the answer is clearly yes, and Gartner made it semi-official when it moved generative AI down into what it calls the "trough of disillusionment," the stretch of the cycle where the first rush of wonder wears off and everyone starts asking harder questions about whether the thing really works at scale. But this is exactly where most of the "AI is dying" takes go wrong, because a hype cycle cooling off is not remotely the same as a technology going away, and the dot-com story is the proof of that. The crash in 2000 didn't kill the internet. It burned off all the companies that were only ever a clever name and a press release, it left standing the handful that were quietly building something real, and within a decade the internet had gone from the thing everyone breathlessly hyped to the thing everyone simply used, all day, without ever thinking about it.

Look underneath the deflating froth right now and you'll find that same pattern already forming, because while the buzzword is steadily losing its power on the stock market, the actual usage is climbing hard in the opposite direction. Nearly eight in ten companies now report using AI somewhere in the business, and the share running it in real production rather than in a forever-pilot keeps rising year over year. The two biggest labs, meanwhile, are posting revenue ramps that would have sounded completely made up a couple of years ago, with one of them reportedly going from around a billion dollars in annual revenue at the end of 2024 to something close to thirty billion by early this year. That's not the signature of a fad quietly collapsing. That's the signature of a tool getting embedded into how work actually gets done.

The most honest picture of the whole transition, oddly enough, comes from a study that made the rounds as bad news. MIT researchers recently reported that roughly ninety-five percent of enterprise AI pilots were delivering no measurable impact on the bottom line, and the headlines happily read it as proof that the entire thing was a bust. But if you get past the number, the far more interesting finding was the reason those pilots failed, because it wasn't the technology that came up short, it was the organizations around it. The failures clustered in companies that expected to sprinkle a little AI on top of a broken process and watch the magic happen, while the small minority that actually did the hard work of restructuring the work around it were creating real, serious value. That's not a story about AI being fake. That's a story about AI quietly turning into the kind of tool that only pays off when you do the boring, unglamorous work of wiring it properly into how you operate.

And that, to me, is the whole shift compressed into a single sentence. We're moving out of the phase where you could drop "AI" into a headline and watch your valuation jump, and into the phase where AI only rewards the people who actually use it well. The froth is leaving, and the froth was never the point in the first place, because the value was always going to end up with the ones quietly integrating this into real work, which is the same thing I keep circling back to whenever I write that fluency with these tools is the entire game, since in a maturing market your edge stops being raw enthusiasm and starts being genuine skill.

So no, I don't think the AI hype is dying, I think it's growing up, and I'm fairly convinced that's the healthiest thing that could possibly happen to it. The loud, speculative, everything-is-a-gold-rush phase really is fading, and it's taking a great many empty promises down with it, but the technology sitting underneath all that noise is quietly becoming part of the furniture, the same way electricity and the internet did before it. Nobody brags about being "on the internet" anymore, they simply are, and in a few years nobody is going to brag about using AI either. They'll just be using it to get real work done, and that quiet, boring, unglamorous moment is precisely when it will matter the most.