Is the AI Bubble Popping? What History Says About Stocks That Boom
Unpack what the AI bubble debate means for investors, from Gary Gensler's recession warning to Yardeni's bubble meter, and see which corners of the market history has rewarded after a pop.

Key Takeaways
- Gary Gensler says it flatly: the AI boom is already a bubble, and a downturn lands before the next presidential election.
- Yardeni's meter splits 'bubble' from 'about to burst', and right now it flashes the first but not the second.
- History, not hype, is the guide: bubbles that lack triggers rarely end in a bear market.
- Neoclouds, the debt-heavy renters of AI computing power, are where Gensler expects the first breakage.
- Busts leave bargains behind. The dot-com era left cheap fiber optics for whoever stayed patient.
- No source names a single stock that booms after a pop, so beware anyone who does.

Imagine a party where everyone insists the music is about to stop, and nobody leaves. That's the AI trade right now. Former SEC chair Gary Gensler told CNN it's already a bubble.
Yardeni Research reads the same dials and lands somewhere calmer: a bubble, yes, but not one about to burst. Alphabet's Sundar Pichai adds a sobering note that if it does burst, no company is immune.
History offers a script. Busts punish the over-borrowed and leave cheap infrastructure for the patient. So which corners of the market have rewarded patience before? Let's get started.
Spotting the AI Bubble
Gary Gensler doesn't tiptoe. "It's already a bubble," the former SEC chair told CNN, and he expects the pop before the next presidential election. Picture a gold rush where everyone sells shovels and nobody has dug up the gold yet.
His evidence is the spending. Roughly $800 billion a year is pouring into AI infrastructure, and Big Tech is spending five times more than it did three years ago. Money is arriving faster than profit, a mismatch that must balance eventually.
Yet the room isn't unanimous. Ed Yardeni calls this a self-correcting bubble that doesn't have to burst, and he argues that widespread bubble talk keeps the downside smaller. Skeptics and optimists agree on one thing: the froth is real.
Reading the Bubble Meter
What if a gauge could separate 'bubble' from 'about to burst'? Yardeni Research built one, the AI Bubble Meter, with two dials. One asks whether bubble conditions exist. The other asks whether the forces that ended past booms have shown up.
The Conditions dial reads 83, flashing bubble. The Triggers dial sits at 34, far calmer. One trigger is emerging: the two-year Treasury yield is up 115 basis points from its December low, though that is still short of the climbs seen before earlier reversals.
In the firm's own telling, the preconditions are in place but the forces that ended past booms are only partly present. Watch the Triggers dial. If it climbs toward High, the risks climb with it.
Tracing Bubbles Through History
Every transformative technology gets its fever. Gensler points to railroads, electricity and the internet, each followed by deep downturns after builders funded infrastructure ahead of revenue. AI is the newest chapter in an old, familiar story.
Yardeni Research mined market history back to 1965. When bubble conditions ran hot but triggers stayed cool, a bear market followed within two years only 9% of the time. When both ran hot, that share jumped to 42%.
These are historical associations from the firm's model, not forecasts. Still, they hand investors a useful habit: judge a bubble by its triggers, not by its mere existence.
Finding Winners After the Pop
Here's the uncomfortable truth behind the headline promise: the sources name no single stock that booms after a pop. What they do show is where value pools. Bubbles leave behind cheap, high-value infrastructure, the way the dot-com era left fiber optics.
The shakeout is a different story. Gensler expects breakage among neoclouds, the debt-heavy firms that lease out supercomputing power, while the major companies survive. Arthur Hayes expects capital to eventually exit overvalued AI stocks and banks for gold and Bitcoin.
Timing matters, though. If triggers stay muted, Yardeni's history shows a median inflation-adjusted gain of 18.3% over two years, so sitting out the party carries its own cost. Pain first, bargains second is a pattern, not a promise.
Weighing the AI Bubble Risks
Think of neoclouds as the first dominos in a long row. Elon Musk warned that the existing neoclouds face "guppy-level death and consolidation," and that some will make it through, though not many.
The hidden plumbing worries Gensler most: circular financing and off-balance-sheet deals that blur how much risk sits where. Arthur Hayes adds a GPU mismatch, with five-year loans funding hardware that goes obsolete in two.
Even so, Gensler pegs the damage as significant as the dot-com bust but short of the 2008 crisis. A bruise, not a break. On Wall Street, the optimists note that AI stocks have lost altitude without imploding.
Long Story Short
The AI bubble is real by most readings, but 'real' and 'bursting' aren't synonyms. Yardeni's meter flashes bubble conditions with the triggers only partly in place. Gensler, meanwhile, sees a repricing and a softer economy before November 2028. So watch the Triggers dial, Treasury yields and the hyperscalers' third-quarter earnings for signs of idle capacity. Keep an eye on debt-heavy neoclouds, where Gensler expects the first breakage. Treat every forecast, including the ones here, as a view rather than a verdict. Bubbles end in wreckage, then in bargains: fiber optics after the dot-com era, cheap infrastructure for whoever waited. Stay diversified, stay skeptical, and keep some dry powder for the morning after. History tends to pay the patient, just not on the schedule they hoped for.