The Real Money in AI Isn't in Chips—It's What Powers Them

The AI boom has everyone talking about Nvidia and GPU shortages. But here’s what most investors miss: the trillion-dollar opportunity isn’t just in the hardware itself—it’s in the unglamorous infrastructure that makes AI actually work.

Why AI Infrastructure is the Real Bottleneck

McKinsey estimates companies need to spend $5.2 trillion by 2030 just on AI infrastructure. That’s not buying GPUs and memory chips from Nvidia and Micron—that’s building the physical foundation for the entire AI revolution.

Think about it: a single AI data center campus can consume over 1 gigawatt (GW) of electricity. That’s enough power to supply 750,000 homes. According to Anthropic, the U.S. alone will need at least 50 GW of power by 2028 just for AI operations. We’re not even close to having that capacity today.

This creates a massive rush for two types of assets: the physical real estate to house AI computers and the power infrastructure to keep them running.

The Data Center Gold Rush: Where AI Actually Lives

AI companies don’t run their operations in the cloud as some abstract concept—they need massive, purpose-built facilities. These aren’t your typical office buildings. We’re talking about specialized data center campuses with advanced cooling systems, redundant power supplies, and custom-designed computing environments.

Several companies are betting big on building these facilities:

Equinix (NASDAQ: EQIX) and Digital Realty (NYSE: DLR) are two of the largest data center REITs making major moves. Equinix formed a $15 billion joint venture in 2024 to acquire land and construct state-of-the-art xScale data centers—massive, AI-optimized facilities. Digital Realty launched its U.S. Hyperscale Data Center Fund in 2025 to deploy up to $10 billion in new capacity and has a $7 billion partnership with Blackstone (announced in late 2023) specifically for large-scale AI data center development.

Brookfield Infrastructure (NYSE: BIPC, NYSE: BIP) operates over 140 data centers globally with 1.6 GW of current capacity. The company sees room to develop an additional 3.4 GW across its platform—and it’s investing in next-generation power solutions, including fuel cell technology from Bloom Energy, to make these facilities even more efficient.

Power Infrastructure: The Unsexy, Essential Play

Here’s where it gets interesting for energy stocks: AI data centers need constant, reliable, massive amounts of electricity. This is forcing a complete rethinking of America’s power grid.

NextEra Energy (NYSE: NEE) is positioning itself as a leader in this space. The company plans to invest over $25 billion in electricity transmission projects to expand the grid’s capacity. It’s also building out its natural gas pipeline network—because gas power plants are expected to supply much of the immediate power boost needed. Beyond that, NextEra is the largest renewable energy developer in the U.S. and has inked strategic partnerships with Google to deploy nuclear power and co-develop multiple large-scale data center campuses.

Williams (NYSE: WMB) is another key player. As a major natural gas pipeline operator, it’s executing multiple projects to increase gas supply across the country—projects expected to go live through 2030. The company is evaluating an additional 30 projects worth over $14 billion that could be built between 2027-2033. It also has $5.1 billion in gas-power generation projects under construction specifically designed to serve data center customers, with several more in the pipeline.

The Real Opportunity

The AI chip shortage gets headlines, but the real bottleneck—and therefore the real investment opportunity—is infrastructure. Building data centers, expanding power grids, and securing energy generation capacity will define which companies capture the most value from the AI era.

From Brookfield Infrastructure’s global data center network to NextEra Energy’s transmission and renewable expansion, these companies are laying the physical foundations that will make AI’s explosive growth actually possible. While everyone chases GPU stocks, the unsexy infrastructure plays could deliver the most consistent long-term returns.