AAPL

TSLA

AMZN

META

AMD

NVDA

PEP

COST

ADBE

GOOG

AMGN

HON

INTC

INTU

NFLX

ADP

SBUX

MRNA

Markets

EQIX

Your Artificial Intelligence (AI) Portfolio Probably Looks Very Different Than It Did 6 Months Ago. Here's Why That's OK.

March 29, 2026 — 12:50 pm EDT

Written by

Micah Zimmerman for

The Motley Fool->

Key Points

- The AI trade is rotating from hype-driven “builders” to profit-driven “adopters,” as investors demand real revenue and margins instead of just narrative.

- An AI portfolio built for the next phase of the trade looks less like a concentrated tech bet and more like a layered infrastructure position.

- These 10 stocks could mint the next wave of millionaires ›

If you have had a heavy artificial intelligence (AI) position over the last year or two, it probably included many of the same names: Nvidia, Advanced Micro Devices, Microsoft, a few hyperscalers, and maybe some software-as-a-service (SaaS) plays that had "AI" somewhere in the investor deck. Back then, if a CEO or someone on anearnings callwhispered "AI implementation," it felt like the stock would rocket 15% overnight.

Today, if you've been paying attention, the list of trendy AI stocks looks different. As a result, some AI positions are down significantly. Some of the things you didn't own are up.

Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »

The rotation away from AI began quietly. In early 2026, investors began asking a question the market had avoided for two years: if AI is going to reshape every industry (i.e., will AI steal my job?), why are companies that are being reshaped trading at the same multiples as those doing the reshaping? In other words, why are some of these massive private and public AI companies fundamentally unprofitable, burning massive amounts of cash on compute, while real customer demand and revenue don't justify the costs?

Image source: Getty Images.

The market is repricing its AI holdings

Morgan Stanley's Global Investment Committee put together a useful framework: The market is shifting from AI "builders," the infrastructure providers and chip companies, toward AI "adopters," which are companies using AI to actually lift productivity and margins, as shown in their income statements.

The flip side of that is the repricing of companies most at risk of disruption. That's what happened to software. The software sell-off wasn't irrational, even if it was overdone. It was the market trying to separate companies whose pricing power survives AI from those that lose it.

When Anthropic released agentic tools that could automate enterprise workflows, the market asked a reasonable question: Why pay per-seat SaaS fees if AI can do the job? The resulting sell-everything panic punished good companies alongside bad ones, but the underlying question is legitimate.

Meanwhile, semiconductors held up. For example, the Russell 1000 Semiconductor Index diverged sharply from the Russell 1000’s software sector. Physical AI infrastructure kept building. Data center cooling companies reported record backlogs. Fiber connectivity companies launched new density-optimized product lines for hyperscale environments. The parts of the AI stack getting paid in real dollars, on real contracts, kept growing.

What a healthy AI portfolio looks like now

A portfolio built for the next phase of the AI trade looks less like a concentrated tech bet and more like a layered infrastructure position.

Think about it in terms of who is getting paid, regardless of which AI platform wins. Cooling infrastructure doesn't care whether OpenAI, Anthropic, or Alphabet wins the model race. Data centers need chillers either way.

A good example here is Vertiv (NYSE: VRT). This company is a direct beneficiary of AI's power and cooling demands, supplying the thermal infrastructure every data center needs, regardless of which models win. Another is Equinix (NASDAQ: EQIX), which operates the physical backbone of the internet, leasing data center capacity and interconnection services that scale alongside AI workloads.

Fiber connectivity doesn't care whether the winning AI runs on Nvidia or AMD GPUs -- it needs fiber either way. Enterprise AI tooling deployed at scale on long-term contracts has revenue visibility that doesn't reprice with every quarterly sentiment shift. Amphenol (NYSE: APH) powers AI clusters with high-speed connectors and interconnect systems that are becoming increasingly critical as compute density increases.

Here's a subjective take about an AI portfolio that looks different than it did six months ago: the change itself is a