AI Data Centers Are Straining America’s Power Grid. Which Stocks Actually Benefit?

KEY TAKEAWAY

PJM transmission congestion costs jumped 43% to roughly $6 billion in the first half of 2026. Data-center load growth is a major part of the long-term pressure, but weather and existing transmission constraints also matter. The opportunity is therefore broader than utilities: electrical equipment, grid construction, power systems and cooling all sit directly in the buildout.

◎

EXPLORE THE STOCKSReview Eaton, Vertiv, Quanta, GE Vernova and Generac in a demo portfolio before putting real money at risk.OPEN A FREE DEMO PORTFOLIO →Virtual funds. No investment required. MarketInsiderLab may receive compensation if you open an account through this link.

Artificial intelligence is turbocharging electricity demand. Hyperscale data centers are coming online faster than transmission can be expanded in many regions. The result is higher congestion, longer interconnection queues and a major infrastructure buildout spanning grid equipment, power systems and cooling.

Why the Grid Is Under Pressure

PJM Interconnection — the largest U.S. regional grid, serving about 67 million people — reported a sharp increase in transmission congestion during the first half of 2026. Its independent market monitor said congestion costs rose 43% to roughly $6 billion. Real-time wholesale electricity costs also rose sharply.Data centers are part of the pressure, especially in Northern Virginia and other high-load corridors, but they are not the only cause. Severe weather and overloaded high-voltage lines were major contributors to the 2026 congestion spike. That distinction matters: the investment thesis is not “AI caused a $6 billion grid crisis.” It is that a grid already facing reliability and transmission constraints is now being asked to absorb unusually fast load growth.

5 Companies With Real Exposure to the Power Infrastructure Boom

Which companies have the cleanest direct exposure?

1. Eaton: one of the cleanest equipment plays

Eaton’s relevance is not simply that it sells electrical products. Data centers require a dense stack of switchgear, distribution equipment, power-quality systems and increasingly thermal-management capability. That makes Eaton an unusually direct way to participate in the physical buildout without depending on electricity prices themselves.In the second quarter of 2026, Eaton reported record sales of $8.5 billion. Electrical Americas organic sales increased 18%, while the rolling twelve-month average of orders rose 41%. Management explicitly identified data centers as a key growth driver, while stressing that demand is broad across end markets.

2. Vertiv: the bottleneck continues inside the data center

Getting power to the site does not solve the entire problem. AI racks are increasingly power-dense, creating demand for power conversion, uninterruptible systems and advanced cooling. Vertiv is therefore exposed to the infrastructure problem after electricity reaches the campus.Vertiv’s second-quarter 2026 sales reached $3.27 billion, up 24% from the prior year. Adjusted operating profit rose 51%, and the company raised full-year guidance across its key metrics. That does not guarantee the stock is attractively valued; it does demonstrate that the infrastructure narrative is translating into current operating results.

3. Quanta Services: monetizing the connection work

One of the least glamorous parts of the AI boom may be one of the most essential: building transmission, substations, site electrical systems and large-load infrastructure. Quanta’s second-quarter filing explicitly ties demand for new and expanded transmission and substation infrastructure to electricity growth from data centers, technology loads, reshoring and electrification.Second-quarter 2026 revenue rose to $9.56 billion from $6.77 billion a year earlier, while total backlog reached a record $53.4 billion. Quanta is not a pure-play data-center stock, which is precisely why the exposure can be more durable than a single-theme narrative.

4. GE Vernova: generation and the grid in one company

GE Vernova has unusually broad exposure because the bottleneck exists both in generating enough electricity and in moving it. Its Electrification business supplies transformers, switchgear, substations and HVDC systems, while Gas Power is benefiting from customers seeking dispatchable generation.In early 2026, GE Vernova said Electrification equipment orders were growing sharply and that data-center development was contributing to grid-equipment demand. Its reported Electrification backlog had increased more than fourfold from 2022 levels by the first quarter of 2026.

5. Generac: a surprising beneficiary of reliability anxiety

Generac is better known for residential backup power, but the data-center buildout has changed the mix of opportunity. Reuters reported that the company has built a roughly $1.6 billion backlog for commercial generators tied to data-center demand and is investing $250 million to expand production.This is a good example of what Market Radar is designed to find: a company whose exposure is visible in actual orders and capacity expansion, rather than merely in an investor presentation that mentions AI.

What the market narrative gets right

• AI data-center load is arriving unusually quickly.
• Transmission and generation expansion cannot always keep pace with campus construction.
• Electrical equipment, power management, cooling and grid construction are seeing real order growth.
• The buildout is broad enough to create second-order industrial beneficiaries well beyond Nvidia and the chip supply chain.

What the narrative can overstate

• Data centers are not the only reason grid congestion is rising. Weather, existing transmission constraints, operating practices and other load growth matter.
• Utilities are not automatic winners. They can face capital intensity, regulatory lag and customer affordability pressure.
• Backlog is not the same as profit. Execution, pricing, supply constraints and project timing still determine returns.
• “AI power stock” is too broad a label. The strongest evidence comes from orders, backlog, contracted projects and segment-level exposure.

MarketInsiderLab conclusion

The AI power story is real, but the investable exposure is more specific than the headline suggests. The clearest beneficiaries are companies selling the equipment and services required to connect, power and cool large-load facilities.Among the names reviewed here, Eaton, Vertiv, Quanta Services and GE Vernova have the broadest direct exposure across the physical infrastructure stack. Generac is a more concentrated second-order example with unusually visible data-center backlog.

Sources

• Reuters — PJM transmission congestion costs, Aug. 21, 2026
• PJM — planning for data-center load growth
• Eaton — Q2 2026 results
• Vertiv — Q2 2026 results
• Quanta Services — Q2 2026 results
• GE Vernova — Q1 2026 earnings presentation
• Reuters — industrial beneficiaries of the U.S. data-center boom, Aug. 19, 2026