The AI rally shifts course: why JPMorgan says to look beyond Nvidia

Let's be honest: for the past year and a half, the AI investment strategy looked embarrassingly simple. You bought chipmakers (you know which ones), added a few cloud giants, and the returns chart shot into the stratosphere. But nothing lasts forever in markets, and Mira Pandit, global market strategist at JPMorgan Asset Management, has brought a sobering message: the easy ride is over, and the real work has begun.

Pandit's core point is that AI trading is going through a so-called "broadening out". That does not mean the bubble has burst. It means the bubble is changing shape. The money that used to be vacuumed up exclusively by hardware makers is now starting to flow into other vessels.

What has changed?

The market has had its fill of promises about a bright future and is now looking for those who will service that future physically. Investors have grasped a simple fact: for GPT-5 (or 6, or 7) to work at all, you need more than H100 graphics processors. You need gigawatts of power, advanced cooling systems, copper cabling and real estate for data centers. That is why we are seeing a cyclical turn toward utilities, energy and industrials. This is no longer the glamorous tech sector but gritty infrastructure, without which AI remains just code on a server.

The second important point JPMorgan highlights is geography. For a long time the AI rally was a purely American party. Now the focus is shifting to international markets. Europe and Asia have companies that are embedded in supply chains or develop applied software, yet trade at a fraction of the price of overheated American counterparts. Smart money is hunting for undervalued assets where others are not yet looking.

Why does this matter right now?

We are at an inflection point. Expectations for Big Tech are stretched to the limit, and any slip in a quarterly report is punished with an instant drop in share prices. Meanwhile, "boring" sectors that benefit from AI adoption (for example, companies using neural networks to optimize logistics or production) show growth potential. This is a classic capital rotation: investors take profits on the race leaders and move them into laggards that now have a growth driver.

For the industry, this is a sign of maturity. Hype is turning into business processes. If we used to pay for the "shovels" (chips), we are now starting to pay those who dig with the shovels, and those who sell water to the gold prospectors.

The bottom line: the "winner takes all" bet no longer works. The AI portfolio of the future is not a list of five tech giants but a complex mix of energy, industry and international software. Are you ready for that kind of diversification?