AI Data Centers Are Power Hungry. These 6 Power and Cooling Stocks Could Benefit Through 2030.

The IEA projects AI power demand will double from its 2024 level to 945 terawatt-hours by 2030, creating opportunities across data-center power, cooling, generation and utilities. Eaton said AI data centers were expected to account for 21% of its 2026 sales, while Vertiv reported second-quarter 2026 sales growth of 24% year over year and earnings growth of roughly 60%; both raised full-year guidance. Bloom Energy reported fuel-cell demand up 140% year over year to $6 billion at the start of 2026, and GE Vernova ended the second quarter with a $176 billion backlog; the article notes Plug Power remains unprofitable and speculative.

Analysis

The investable bottleneck is not aggregate AI electricity demand but who can monetize scarce equipment, interconnection capacity, and power quickly. ETN and VRT have relatively direct exposure to data-center electrical and cooling spend; the key near-term test is whether orders convert into revenue and margins without delivery costs or customer concentration eroding returns. GEV’s opportunity is further downstream: a large backlog is not earnings until manufacturing capacity, project timing, and customer financing support delivery. Treat backlog quality and conversion as the catalyst, not backlog size alone.

The market may be underpricing substitution and demand elasticity. On-site generation can defer grid connection and benefit BE, but could also weaken the immediate case for regulated utilities; efficiency gains, slower AI deployment, or canceled projects would reduce load growth across the chain. Utilities may grow rate base, yet affordability concerns and regulatory approval can delay cost recovery. PLUG’s broad hydrogen ambitions are not equivalent to BE’s data-center proposition; avoid extrapolating one company’s demand claims to the other.

Over 1–3 months, track order conversion, guidance, equipment lead times, utility interconnection approvals, and data-center project cancellations. Over 6–18 months, watch whether grid and generation additions keep pace with announced load. Verify the reported NextEra–Dominion transaction status independently before assigning any deal value or changing exposure. The thesis weakens if ETN/VRT orders or margins soften, GEV backlog conversion slips, or utility load forecasts and contracted projects are revised down.

Key Decisions for Investors

• Prefer ETN/VRT over PLUG for exposure to nearer-term infrastructure spending, but initiate only after checking current valuation and order-to-revenue conversion; do not assume strong demand guarantees attractive returns.
• Consider a modest quality-versus-speculation pair—long ETN or VRT against a short PLUG—only if relative valuations and borrow are acceptable. Risk: AI capex weakness could hurt both legs, while a hydrogen-specific catalyst could sharply squeeze PLUG.
• Keep GEV on a conditional buy list; add only if capacity expansion translates into delivery and margin progress, and reassess on any backlog conversion or guidance deterioration.
• Treat BE as a higher-risk, project-economics exposure: verify realized deployments, customer commitments, and fuel economics before buying; demand claims alone are not a sufficient trigger.
• Do not trade NEE or D on the stated transaction premise until the deal is confirmed from primary sources; monitor regulatory approvals, rate-case recovery, and revised data-center load forecasts instead.