Why the Real AI Infrastructure Opportunity Might Be Hiding Beyond the Chip Makers

The Chip Story Has Already Been Told

Everyone knows Nvidia (NASDAQ: NVDA) has been the star of the artificial intelligence revolution. The numbers speak for themselves—a staggering 25,000% surge over the past decade and commanding roughly 8% of the S&P 500 (SNPINDEX: ^GSPC). Wall Street has certainly noticed.

But here’s the thing: success breeds risk. Nvidia may dominate the chip market today, but nothing guarantees this position will last forever. During the internet era, Yahoo! ruled web search until Alphabet’s (NASDAQ: GOOG, GOOGL) Google came along and reshaped the entire landscape. The chip industry could follow a similar path as AI technology matures and competition intensifies.

Even more telling: Nvidia’s price-to-earnings ratio sits around 55x currently—high on an absolute basis, despite being below its five-year average. Compare that to the S&P 500’s average P/E of 29x (itself historically elevated), and the valuation picture becomes clearer. The risk-reward equation for late-stage chip investors may not be as compelling as headlines suggest.

Looking Upstream: Where AI Actually Operates

Here’s what gets less attention: AI doesn’t run in the clouds—it lives in buildings. Specifically, data centers. And that’s where the investment chain gets interesting.

The explosive growth in data center construction has benefited data center REITs, companies like steel manufacturer Nucor (NYSE: NUE), which supplies building materials, and industrial powerhouse Eaton (NYSE: ETN), which provides critical power management infrastructure. These businesses have positioned themselves at the intersection of AI expansion and physical infrastructure requirements.

But there’s a catch. Technology cycles eventually mature. Data center demand, while booming now, could eventually stabilize as the buildout completes. Past technology revolutions offer a cautionary tale—explosive growth followed by plateau.

The Unglamorous but Durable Play: Electricity

This is where the analysis goes one layer deeper. AI systems require constant, uninterrupted power. Every data center that opens permanently increases the power grid’s load. Unlike the temporary nature of construction booms, electricity consumption remains as long as these facilities operate.

This fundamental shift has profound implications. NextEra Energy (NYSE: NEE), a utility giant, projects electricity demand will surge 55% between 2020 and 2040—compared to just 9% growth during the prior two decades. That’s a structural change reshaping the entire power industry, not a cyclical bump.

Capturing the Trend Without Individual Stock Risk

Investing directly in regional utility companies introduces geographic concentration risk, since most utilities operate within defined service territories. A more efficient approach involves sector-level exposure through exchange-traded funds.

Two solid alternatives stand out: the Vanguard Utilities ETF (NYSEMKT: VPU) and the Utilities Select Sector SPDR ETF (NYSEMKT: XLU). These funds provide nearly identical characteristics—both charge roughly 0.9-0.8% expense ratios annually, generate approximately 2.6% dividend yields, and include holdings like NextEra Energy. Performance outcomes are essentially equivalent, backed by trusted financial institutions.

The Real Opportunity in AI Infrastructure

The narrative surrounding AI has focused heavily on semiconductors and data centers, but the true infrastructure boom extends far beyond. The projected 55% increase in electricity demand between 2020 and 2040 represents an often-overlooked but exceptionally durable investment thesis.

Unlike chip makers facing competitive pressures or data center REITs subject to cyclical construction patterns, utility providers benefit from inelastic demand. Power remains essential whether the AI market expands, contracts, or stabilizes. This structural advantage makes utilities a genuinely underrated way to gain exposure to the artificial intelligence revolution—one that will likely persist throughout the decade and beyond.

The AI infrastructure story isn’t finished. It’s simply evolved beyond the headlines into the foundational systems that power the technology itself.