For over a decade after the financial crisis, the investment playbook was simple: buy growth, buy tech, and pretty much ignore everything else.

Capital-light business models were king. Software companies could scale infinitely without building factories or warehouses. Margins expanded beautifully. With interest rates stuck near zero and money flowing freely, the longest-duration assets—especially mega-cap tech stocks—delivered spectacular returns year after year.

Meanwhile, businesses tied to physical stuff like factories, power plants and construction equipment were left in the dust, frequently dismissed as "value traps" destined to disappoint.

But Goldman Sachs equity analyst Peter Oppenheimer thinks that story might be changing in a big way. In a new research note, he raises an intriguing question: Could those supposedly washed-up "old economy" stocks actually be the next phase of the AI trade?

AI Infrastructure Is Very Much a Physical Thing

Here's the thing about artificial intelligence that's easy to overlook: it doesn't just live in the cloud. Training large language models and running massive AI systems requires enormous amounts of computing power, which means data centers. Lots of them. Really big ones.

Tech giants like Amazon.com Inc. (AMZN), Meta Platforms Inc. (META) and Alphabet Inc. (GOOGL)—the so-called hyperscalers—are pouring staggering amounts of capital into building this infrastructure. And that spending doesn't stop at chips and servers.

It spills over into utilities, power grids, industrial equipment and construction. All the unglamorous stuff that makes the digital world actually work.

Many of these sectors have been starved for investment since 2008. Overcapacity and disappointing returns crushed capital spending in traditional industries, which became dominated by low-multiple value stocks that nobody wanted to touch.

That drought is ending now.

"Investment in AI infrastructure coupled with the renewed boost in defense spending are re-igniting returns on investment in many of the physical assets that had long lagged behind, just as investors fear the slowing of returns from record high levels in the technology space," Oppenheimer said on Thursday.

For the first time since the internet went mainstream about 25 years ago, future tech growth depends heavily on physical infrastructure: data centers, semiconductor fabrication plants, power grids and energy supply. AI isn't just software. It's energy-intensive, capital-intensive and infrastructure-heavy.

Is Big Tech's Winning Streak Losing Steam?

At the same time this infrastructure buildout accelerates, investors are starting to ask uncomfortable questions about whether Big Tech can actually justify these enormous AI spending sprees.

"The surge in AI capex by the hyperscalers has prompted investors to question their ability to collectively generate adequate returns on investment, prompting the pace of performance to slow, and the dispersion of returns within the technology sector to widen," Oppenheimer noted.

The numbers tell the story. The Magnificent 7 delivered a jaw-dropping 75% return in 2023, the year after ChatGPT launched and kicked off the AI frenzy. Those gains slowed to about 50% in 2024. So far in 2025, returns have slipped below 25%.

The era of uniform mega-cap dominance appears to be fading.

From Value Traps to Value Creators

Since early 2025, market leadership has shifted noticeably. Gold, emerging markets, Japan's Topix, industrial metals and value stocks have all outperformed. Over the past three months, the iShares Russell 1000 Value ETF (IWD) has outperformed the iShares Russell 1000 Growth ETF (IWF) by 18 percentage points.

That represents a sharp break from the prior decade's pattern.

Oppenheimer argues the opportunity set is shifting, as "value parts of the market that have long been neglected as 'value traps'" could evolve into "value creators" by boosting cash flows and returning more capital to shareholders through dividends and buybacks.

Goldman maintains that equities are likely to remain the best-performing asset class overall. However, the drivers of returns are broadening beyond a narrow slice of mega-cap tech names.

What This Shift Means for Your Portfolio

The AI revolution isn't just about semiconductors and software anymore. It's about power plants, transmission lines and industrial capacity. The boring stuff that keeps the lights on.

For investors who've spent years chasing the next hot AI software company, the next leg of this trade may look surprisingly different. It might run through power grids, industrial metals and factories rather than just elegant code and user interfaces.

Rather than relying on a narrow group of mega-cap growth stocks to carry the market, investors may find more opportunities scattered across sectors tied to physical assets and real-world investment cycles.

This dispersion creates new chances to generate alpha—excess returns versus the broader market—by picking the right sectors and companies rather than just riding a handful of tech giants.

If Goldman's analysis proves correct, the real AI winners may not all sit in Silicon Valley. Some might be running utilities in the Midwest or building transmission equipment in industrial parks you've never heard of.