As valuations soar, investors are hunting the next AI beneficiaries — from chip rivals to foundries and software stacks. A practical watchlist for the shift.
Illustration by IMF Alpha editorial · Reviewed by Pedro Marini
The shift is subtle but real. Nvidia was the easy, obvious AI trade for a long time. With its price and supply profile shooting up, though, portfolio managers are starting to widen their aperture — they want exposure to AI’s next chapter without putting all their chips on a single name.
Why the rotation is happening
Who stands to gain
Signals to watch
A bit of history for perspective
Hardware-led rotations aren’t new. Early GPU cycles saw compute demand outpace supply and let secondary players stage comebacks. What’s different now is scale: software and services have matured enough that AI workloads create stickier, broader demand across chips, networks and clouds.
Risks and counterpoints
How to position, practically
Treat this as an ecosystem bet, not a single-stock story. That distinction separates thoughtful positioning from headline chasing.
Pedro Marini
How synthetic data is letting banks train powerful AI without exposing customer records — and why investors should care now
Smaller models, smarter silicon, and a privacy-first pitch are shifting generative AI from datacenters into your pocket — and changing winners and business models.
New chips, model tricks, and a privacy play are moving large language models from data centers into phones. Here is who wins, who loses, and what that means for users.