Dhaval Joshi has a warning for investors: artificial intelligence isn't one giant bubble waiting to pop. It's a rolling sequence of smaller bubbles, each inflating and deflating in turn. That pattern might keep the overall market from crashing, but it brings its own set of problems.
The Rolling Bubble Thesis
Joshi's argument flips the usual doomsday narrative. Instead of a single, synchronized collapse across all AI-related assets, he sees a chain reaction. Money piles into one hot sector — say, AI chips — then that gets overheated and pulls back, while the same capital rotates into another area, like cloud infrastructure or AI software. The cycle repeats.
This isn't a one-time event. It's a process that could stretch on for years, with each mini-bubble masking the risk of the whole system. The market doesn't crash because there's always a new bubble to catch the falling money.
Why a Crash Might Be Avoided
If Joshi is right, the rolling nature is a kind of shock absorber. When one sector deflates, the broader indexes barely notice because another sector is still climbing. That could prevent the kind of panic that hits when everything falls at once.
But avoiding a crash isn't the same as being safe. The calm on the surface hides a lot of churning underneath. Investors who jump into the latest hot AI trend might find themselves holding the bag when that particular bubble deflates.
The Cost of Misallocated Capital
That's where the real danger lies, according to Joshi. Rolling bubbles encourage capital to chase whatever is shiny right now, rather than what has solid long-term fundamentals. Money flows into overhyped projects, overvalued startups, and speculative bets, while more mundane but durable investments get starved.
This misallocation doesn't show up on a balance sheet overnight. It builds slowly, distorting prices and redirecting resources away from productive uses. By the time the misallocation becomes obvious, a lot of damage has already been done.
Uncertain Returns in a Shifting Economy
The longer-term picture is just as murky. Joshi points to economic shifts that could disrupt the rolling pattern. If interest rates move, or growth slows, or some other macro force intervenes, the sequence could stall. And when it does, the returns investors expected from AI might not materialize.
That uncertainty cuts both ways. Some AI companies will thrive, but many won't. The rolling bubbles mean the winners and losers get sorted out slowly, not in one dramatic reckoning. For investors, that makes it harder to know when to hold on and when to fold.
The challenge now is figuring out which parts of the AI boom are real and which are just the next bubble in line. Joshi's warning suggests the answer won't come all at once.
