Wall Street is absolutely giddy on the AI sugar rush. Market records are breaking weekly, and Nvidia’s valuation has blasted past $4 trillion—a number that still feels slightly unreal. If you’ve been in the markets for a couple of decades, this all feels… familiar. It’s exciting, sure, but it also brings a serious case of dot-com déjà vu.

The big question everyone is grappling with is how to keep riding this bull market without getting completely incinerated when (not if) the correction comes. And for answers, many major investors are dusting off the 1999 playbook.

Don’t Call the Top, Just Rotate

Here’s the thing: trying to time the absolute peak of a bubble is a fool’s errand. You’ll either get out too early and miss the biggest gains, or (more likely) get out too late. The smart money knows this. They aren’t betting against AI; that would be crazy. But they are getting extremely wary of holding the same handful of hyper-inflated stocks that everyone else already owns.

Instead, they are executing a classic rotation.

A look back at the 1998-2000 period shows that the most successful hedge funds didn’t short the internet. They skillfully rode the bubble. They took their massive gains from the first-wave internet stocks and quietly recycled those profits into the next-in-line sectors—like telecoms and infrastructure—that still had room to run. They let the less sophisticated investors hold the bag when the clock finally struck midnight in March 2000.

We are seeing the exact same strategy today. The goal is to move from the obvious winners (the Nvidias and Microsofts) to the “picks and shovels” of the AI gold rush.

Finding the Hardware Store

When everyone is stampeding to find gold, the safest and often most profitable bet is to sell them the picks, shovels, and blue jeans. In 2025, that means looking beyond the chip designers and hyperscalers.

Where is this rotated money flowing?

• The Suppliers: Investors are digging into the supply chain. Think IT consultants who will help companies implement AI, or Japanese robotics firms that benefit from the new tech. One fund manager is reportedly buying into companies that make specialized delivery boxes for chip manufacturers like TSMC. It’s granular, it’s boring, and it’s probably smart.
• The Power Grid: All these new AI data centers are incredibly power-hungry. This has led some managers to an interesting trade: uranium. The logic is that nuclear power may be the only realistic way to fuel this massive new energy demand.
• The Hedges: Other managers, worried about “irrational exuberance,” are moving into software groups and Asian tech stocks that the market hasn’t fully appreciated yet. Some are even using Chinese stocks as a hedge, figuring that if AI advancements in China gain ground, it might take some of the steam out of Wall Street’s rally.

The Risk Remains

This rotation isn’t just offensive; it’s deeply defensive. The skepticism is warranted. We have companies spending trillions of dollars to fight for a market that, in many ways, doesn’t fully exist yet.

There’s a very real danger of overcapacity. This is exactly what happened with the fiber-optic cable boom in the late 90s. Companies laid enough cable to circle the globe a dozen times, anticipating demand that took another decade to materialize, leading to a spectacular crash. The rush to build AI data centers looks awfully similar.

Ultimately, nobody knows when the music will stop. We just know it eventually will. As one manager noted, we’re in 1999 until the bubble pops. The smart play isn’t to guess the date of the crash; it’s to make sure your portfolio isn’t left holding the bag.