GX Insight | Stop Chasing AI, Bet on Energy

In the midst of the global AI frenzy, investors are flocking to soaring valuations of tech giants, chasing the allure of the next trillion-dollar breakthrough. Yet, beneath this glossy narrative lies an overlooked reality: AI's true fuel isn't algorithms or chips—it's electricity. Drawing from the latest market data and forecasts, the Big 7 tech companies' capital expenditures are surging at an unprecedented pace, but the real winners aren't AI startups. They're in the energy sector. It's time to rethink portfolios: Skip the AI hype and bet on energy to capture tomorrow's growth.

AI's "Capital Thirst": Prosperity's Hidden Peril

Consider the big tech titans—Apple, Amazon, Alphabet, Meta, Microsoft, Nvidia, and Tesla. Data from late 2015 through June 2025 reveals their capital expenditure (Capex) as a percentage of cash flow rebounding to historic highs, far outpacing other non-financial firms. The statistics shows the Big 7's ratio spiking dramatically in 2024-2025, signaling massive pours into data centers, GPU servers, and AI infrastructure. Amazon and Microsoft's cloud arms, for instance, are aggressively expanding facilities to power models like ChatGPT.

This "capital thirst" appears as a badge of AI's boom, but a closer look uncovers structural risks. AI's ROI cycles are protracted: Training a large model can take months or years, with diminishing marginal returns—the next innovation often hinges on vast electricity rather than pure tech leaps

Diving headfirst into AI stocks is akin to wagering on these giants' execution. History warns of tech bubbles bursting on infrastructure choke points—recall the 2000 dot-com crash, when fiber optics couldn't keep up. Today, AI's "fiber" is power.

Diving headfirst into AI stocks is akin to wagering on these giants' execution. History warns of tech bubbles bursting on infrastructure choke points—recall the 2000 dot-com crash, when fiber optics couldn't keep up. Today, AI's "fiber" is power.

Data Centers: AI's Electricity Abyss

AI's explosive growth is reshaping U.S. power dynamics. McKinsey and the International Energy Agency's (IEA) September 2024 forecast projects U.S. data centers' electricity share surging from 3.7% in 2023 to 11.7% by 2030—a threefold leap, propelled by AI. Generative models like the GPT series demand compute power equivalent to millions of households. By 2030, U.S. data centers could devour 606 terawatt-hours (TWh)—nearly twice the UK's 2023 total electricity use.

Picture this: A single large language model in training might guzzle a mid-sized city's worth of power. Google and Microsoft have admitted their AI initiatives are driving 20-30% annual data center power hikes. The trend isn't linear: Shares climb stepwise from 2023 to 2030—4% in 2024, 5.2% in 2025, 6.5% in 2026, 8% in 2027, 9.3% in 2028, 10% in 2029, and 11.7% in 2030. This exponential surge positions AI as the culprit.

AI firms aren't power producers; they're consumers. Operators like Equinix and Digital Realty grapple with shortages—many projects stall due to grid limits. Hotspots like California and Virginia face rationing. Betting solely on chips and software ignores the upstream supply crunch.

Energy's Golden Era: The True Windfall from AI's Surge

Shift to the broader power demand landscape, and America's energy market is undergoing a seismic shift. The U.S. Energy Information Administration's (EIA) December 31, 2024, projection sees total electricity demand rising from 3,938 TWh in 2024 to 5,780 TWh by 2050—a 46.8% cumulative gain. Industrial segments (including data centers) lead with 61.2% growth (1,026 TWh to 2,263 TWh); commercial and transport follow at 43% (1,404 TWh to 2,049 TWh); residential at 36% (1,507 TWh to 2,049 TWh).

This "power renaissance" stems from a multi-engine boom: AI inflating industrial use, EVs and data centers boosting commercial demand, and renewables accelerating the mix. Critically, EIA uses sales as a demand proxy, implying even steeper real growth—especially in Texas and Arizona's data center hubs.

Energy is pivoting from "sunset" to "sunrise." Utilities like NextEra Energy and Duke Energy are capitalizing via substation builds and transmission lines for premium pricing. Nuclear revival—think small modular reactors (SMRs)—is heating up; Microsoft inked a deal with Constellation Energy to restart Three Mile Island for AI power. Renewables leaders like First Solar benefit from data centers' "green" mandates—AI-driven demand could claim over 20% of new capacity by 2030.

The Pivot to Reshape Investment Logic

As of October 2025, AI's roar peaks, but data signals a turning tide. Chasing AI's mythological tales risks missing energy's enduring bounty. The real trajectory isn't tech's hyperbolic rise—it's energy's steadfast foundation.

AI may be the new steam engine, but steam engines ran on coal and water. Today, that foundation is power and the grid. If the AI boom is a gold rush, the most reliable returns won’t come from the nuggets—they’ll come from selling the shovels: electricity itself.