MW These 5 infrastructure stocks are gaining smart-money interest as data centers strain the U.S. power grid
By Jurica Dujmovic
Institutional investors are pivoting toward the 'picks and shovels' of the power sector - and opening a new frontier for AI-driven growth
While investors obsess over semiconductor stocks and hyperscalers, a quieter capital rotation is under way.
U.S. utilities are faced with something they haven't seen in a generation: sustained, growing electricity demand, driven by artificial-intelligence data centers and widespread electrification.
The debate over AI's energy appetite typically centers on power generation: oil versus renewables, nuclear versus natural gas. But a closer look at infrastructure constraints reveals a different binding factor: the physical grid itself.
And while investors obsess over semiconductor stocks and hyperscalers, a quieter capital rotation is under way. Utilities are expanding multiyear spending plans to build transmission lines, upgrade substations and procure transformers. The equipment manufacturers and construction firms executing this buildout may represent a more durable investment theme.
Demand surge
The International Energy Agency estimates that global electricity demand will grow 3.6% annually through 2030.
U.S. electricity consumption remained essentially flat from 2005 to 2020, according to the U.S. Energy Information Administration. Efficiency gains had offset population growth and industrial demand stayed subdued. All in all, the grid wasn't stressed. Now that era is over.
The International Energy Agency estimates that global electricity demand will grow 3.6% annually through 2030, driven by industrial activity, electric vehicles, air conditioning and data centers. In the U.S. specifically, data centers are expected to account for roughly 50% of power demand growth for the rest of the decade.
Unlike temporary spikes, structural-load increases from AI training clusters, EV-charging infrastructure, industrial heat pumps and manufacturing reshoring persist.
Developers want to build. The issue is connection capacity.
The problem is not the lack of proposed generation. The issue is connection capacity. As of late 2024, close to 2,300 gigawatts of wind, solar, storage and other projects were waiting in U.S. interconnection queues, far exceeding the capacity that can realistically be added in the near term.
Typical project wait times now stretch three to five years. Transmission lines require multiyear permitting processes. Substation upgrades involve long procurement cycles. Physical infrastructure cannot scale at the pace software and semiconductors can.
In November 2025, Grid Strategies' annual load-growth analysis found that the aggregated utility and regional grid forecasts for U.S. peak load growth over the next five years had climbed to about 166 gigawatts, more than six times the roughly 24 gigawatts that were forecast just three years earlier. This represents a fundamental shift in planning assumptions, driven largely by AI data-center commitments that arrived faster than infrastructure timelines allow.
The choke point
Hyperscalers can order servers and deploy them quickly, but the transformer that steps down high-voltage transmission to usable power cannot be rushed.
The most acute bottleneck may be large power transformers. They are custom-built, weigh hundreds of tons and require specialized manufacturing.
A 2024 U.S. Department of Energy report titled "Large Power Transformer Resilience" notes that lead times have become "exceptionally long," with 36-month delivery schedules now common and maximum lead times extending to 60 months in some cases. (See the report here.)
Data centers can be constructed in 18 to 24 months. Hyperscalers can order servers and deploy them quickly, but the transformer that steps down high-voltage transmission to usable power cannot be rushed, and this creates a fundamental timing mismatch.
Utilities facing surging loads from AI campuses are competing for limited transformer supply. Aging grid infrastructure compounds the problem, as replacement cycles accelerate alongside new capacity needs.
Follow the money
Major utilities are responding with expanded capital plans. This spending reflects signed interconnection agreements, contracted load commitments and regulatory filings that lock in returns. Unlike commodity exposure, transmission and distribution investments earn regulated rates of return, typically in the 9% to 11% range.
Grid modernization also enjoys rare bipartisan support. Transmission buildout does not trigger the ideological battles that surround generation sources. Both political parties support infrastructure investment, which reduces policy risks that renewable subsidies or fossil-fuel regulation face.
Several categories of companies stand to benefit from sustained grid investment:
Transmission construction and engineering. Quanta Services $(PWR)$ specializes in building high-voltage transmission lines and upgrading utility substations. The company maintains multiyear project backlogs and benefits directly from utility capital-expenditure increases.
Power management and switchgear. Eaton $(ETN)$ manufactures electrical-distribution equipment for both utilities and data centers. The company has dual exposure: grid modernization on the utility side and electrical infrastructure for AI campuses on the data-center side.
Grid-equipment platforms. GE Vernova (GEV), recently separated from General Electric, focuses on power generation and grid solutions. The spinoff created a more focused grid-equipment platform positioned for transmission expansion.
Transformer and component manufacturers. SPX Technologies (SPXC) produces electrical components including transformer-related platforms. When supply is constrained and lead times grow, component suppliers often gain pricing leverage.
Broader industrial conglomerates including ABB (SE:ABB) and Schneider Electric (FR:SU) $(SBGSY)$ also have significant grid automation and electrification exposure, though it represents a smaller portion of their total business mix.
The investment case here avoids the volatility inherent in energy commodity exposure. Oil prices can collapse. Natural gas can spike. Solar subsidy programs can change with election cycles. Nuclear policy remains contentious.
Transmission spending is infrastructure capex driven by physical demand and regulated returns. It does not depend on breakthrough technologies or policy shifts. AI creates load growth whether it's powered by renewables, natural gas or nuclear. Regardless of the generation mix, electrons require delivery infrastructure.
This also differs from earlier grid investment themes tied primarily to renewable integration. While wind and solar interconnections contributed to queue backlogs, AI data centers represent dense, constant loads that strain local distribution networks in ways intermittent renewables do not.
Risks and short-circuits
Several factors could slow implementation. AI efficiency gains may reduce power consumption per compute unit faster than anticipated. A recession could delay industrial capital spending and reduce data-center buildout. Transformer supply could normalize if manufacturing capacity expands, compressing margins for equipment suppliers.
Some hyperscalers are exploring on-site generation, including small modular nuclear reactors and natural-gas peakers, to reduce grid dependence. However, these solutions also require interconnection and backup transmission capacity.
Additionally, valuations matter. If equipment stocks already reflect multiyear backlogs and margin expansion, upside may be limited. Investors should evaluate whether current multiples price in the growth trajectory or leave room for further appreciation.
AI-infrastructure investment is often reduced to semiconductors, cloud platforms and software. But physical infrastructure represents a parallel layer of capital deployment that may prove more stable and less cyclical.
For investors looking beyond the semiconductor cycle, grid infrastructure offers exposure to AI growth without the volatility of tech multiples or commodity prices. It is the unglamorous foundation that determines whether lofty AI ambitions can succeed.
More: Google just sucker-punched these highflying tech stocks - don't let the relief rally fool you
Also read: Big Tech's AI fantasy hits a nuclear wall: No fuel, no welders - and no Plan B
-Jurica Dujmovic
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
April 08, 2026 15:42 ET (19:42 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments