Your artificial intelligence (AI) portfolio probably looks a lot different than it did six months ago. Here’s why that’s okay:

If you’ve held a key position in artificial intelligence (AI) in the past year or two, chances are you’ve seen many of the same names. Nvidia, Advanced Micro Devices, Microsoftsome hyperscalers, and perhaps even some software-as-a-service (SaaS) businesses that have “AI” built into their investor decks somewhere. At the time, it felt like if the CEO or someone at an earnings call whispered, “Introducing AI,” the stock price would skyrocket 15% overnight.

If you were paying attention today, your list of trending AI stocks would look different. As a result, some AI positions have been significantly reduced. Some of the things you don’t own are up.

The move away from AI began quietly. In early 2026, investors began asking the question the market had been avoiding for two years: If AI is going to reshape every industry (i.e., “Is AI going to take my job?”), why are the companies that are being restructured trading at the same multiples as the companies that are reshaping them? In other words, these large private and public A.I. Why are some companies spending tons of cash on computing for fundamentally unprofitable reasons, but the actual customer demand and revenue doesn’t justify the cost?

The market is reevaluating the price of AI holdings

morgan stanleyThe Global Investment Committee has put together a useful framework. The market is moving away from AI “builders,” i.e., infrastructure providers and chip companies, to AI “adopters,” which are companies that use AI to actually improve productivity and profit margins, as shown in their income statements.

The flip side of this is that companies most at risk of disruption will be repriced. That’s what happened to software. Even if it went too far, selling the software wasn’t unreasonable. It was the market that was separating the companies with the pricing power to survive AI from those who would lose it.

When Anthropic released an agent tool that could automate enterprise workflows, there were legitimate questions in the market. “Why pay SaaS fees per seat when AI can do the job for you?” The resulting sell-everything panic punished good companies along with bad ones, but the underlying question is valid.

On the other hand, semiconductor sales were steady. for example, Russell 1000 Semiconductor The index diverged significantly from the Russell 1000’s software sector. Physical AI infrastructure continued to build. Data center cooling companies reported record backlogs. The fiber connectivity company has launched a new density-optimized product line for hyperscale environments. The portion of the AI ​​stack paid in real dollars, based on real contracts, continued to grow.

What does a healthy AI portfolio look like now?

Portfolios built for the next phase of AI trading look more like layered infrastructure positions than focused technology investments.

Regardless of which AI platform wins, think in terms of who gets the reward. Cooling infrastructure includes OpenAI, Anthropic, or alphabet Win the model race. Data centers require cooling equipment anyway.

A good example here is: Vertive(New York Stock Exchange: VRT). The company directly benefits from AI’s power and cooling demands, supplying the thermal infrastructure that all data centers need, regardless of which model wins. The other is Equinix(NASDAQ:EQIX)operates the physical backbone of the Internet and leases data center capacity and interconnect services that scale with your AI workloads.

With fiber connectivity, it doesn’t matter whether the winning AI runs on Nvidia or AMD GPUs. Fiber is required in both cases. Enterprise AI tools deployed at scale under long-term contracts provide revenue visibility without price changes due to quarterly changes in sentiment. Amphenol(NYSE:APH) powers AI clusters with high-speed connectors and interconnect systems that are becoming increasingly important as computing density increases.

Here’s my subjective view of an AI portfolio that looks different than it did six months ago. The changes themselves indicate that the trade is maturing, rather than disappearing. Early stage bull markets in major technology changes tend to push everything up because optimism is high, the future feels limitless, and it’s filled with venture capital money.

But as speculation fades and the difference between real business and hype becomes clearer, the next stage is far less forgiving. Companies that survive this kind of rotation, and that are worth owning after a year, will be those with sustained demand and a clearly defined role in the AI ​​ecosystem that can persist even when sentiment cools.

Mika Zimmerman has no position in any stocks mentioned. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Amphenol, Equinix, Microsoft, Nvidia, and Vertiv. The Motley Fool has a disclosure policy.