Data-center power demand
31 GW in 2025 to 66 GW by 2027
The report forecasts this increase in U.S. data-center power demand, linking expansion plans to the need for physically deliverable electricity.
Abstract
The primary headwind for data center expansion is no longer the availability of silicon chips or land — it is the sheer inability of an under-invested electrical grid to physicalize and transmit power. U.S. data center power demand is forecast to climb from 31 GW in 2025 to 66 GW by 2027. Transformer lead times have stretched from 50 weeks in 2021 to over 160 weeks in 2026. The interconnection queue has swelled to 2,600 GW. This report outlines the grid bottleneck triad — the transformer crisis, T&D backlogs, and the shift to on-site microgrids — and profiles four US-listed companies capturing institutional capital as the grid chokes under demand.
Key Highlights
Global rubber market valued at $51.82B in 2025, projected to reach $89.51B by 2034 (CAGR 6.26%)
EV tire sub-segment growing at 14.3% CAGR — the fastest in the entire rubber industry
Asia-Pacific commands 54.18% market share, led by Thailand (supply) and China (demand)
Seven stocks profiled across tire manufacturing, specialty chemicals, and upstream production
Supply chain concentration in Southeast Asia creates structural vulnerability and pricing opportunity
Reader-facing analysis
Research Brief
This on-page brief distills the report's analytical frame, supporting evidence, and limits. The downloadable PDF remains the complete research record.
Thesis & framing
AI and data-center expansion depends on the physical system that delivers electricity, not only on computing hardware or developable land. The report frames grid infrastructure as a binding constraint because new load must be interconnected, transformed, and moved through transmission and distribution networks before facilities can operate. Its central evidence connects rapidly rising U.S. data-center power demand with extended transformer lead times and an enlarged interconnection queue. The resulting analytical lens is a bottleneck triad: transformer availability, T&D backlogs, and greater attention to on-site microgrids.
Market structure
The market structure described here is an interdependent chain rather than a single equipment market. Data-center demand enters a grid whose ability to connect and carry power is shaped by transmission and distribution capacity, interconnection processing, and the availability of transformers. A delay at any link can defer the usable power needed for a project. The report identifies a developing response in on-site microgrids, which are considered alongside—not as a substitute for—the need to expand grid infrastructure. It also discusses four U.S.-listed companies for educational purposes, without providing an individualized selection, allocation, or trading conclusion.
Key data & evidence
Transformer lead times
50 weeks in 2021 to over 160 weeks in 2026
The report presents the lengthening lead times as evidence that transformer availability can delay the conversion of planned demand into operating capacity.
Interconnection queue
2,600 GW
The report identifies the size of the interconnection queue as evidence of the scale of projects seeking access to the power system.
Risks & limits
- The projected increase in data-center power demand is a forecast, so actual demand and the timing of new load can differ from the report's stated path.
- Long transformer lead times and transmission-and-distribution backlogs can delay access to power even where data-center demand and project sites are present.
- A large interconnection queue does not by itself establish when, or whether, queued capacity will become operational.
- On-site microgrids may alter how some projects address nearer-term power needs, but they do not remove the report's broader focus on grid interconnection and power delivery.
Countercase
A countercase is that the grid constraint need not remain equally binding if transformer supply conditions improve, transmission-and-distribution backlogs ease, and interconnection capacity is advanced more quickly than assumed in the report. In that setting, the gap between data-center demand and accessible power could narrow, while on-site microgrids could address a portion of nearer-term needs. The report's evidence still calls for distinguishing these potential improvements from a conclusion that the physical power-delivery system has already caught up.
Research conclusion
Taken together, the report treats power delivery as a practical gating condition for AI and data-center buildout. The relevant questions extend beyond computing demand to whether electricity can be interconnected, transformed, and delivered on the required timetable. The data points on demand, transformer lead times, and the interconnection queue provide a compact way to monitor that relationship. This brief summarizes research and analysis drawn from the report's stated methodology and sources; it is not personalized financial, investment, or trading advice.
Research transparency
Sources & Methodology
This report draws on public energy-system data, national-laboratory research, industry disclosures, and company filings. Its structural thesis and public-company discussion are educational research, not investment advice.
Queued Up: 2026 Edition
Lawrence Berkeley National Laboratory
U.S. transmission interconnection queue data and analysis.
Energy and AI
International Energy Agency
Data-center and AI electricity-demand analysis.
Distribution Transformer Demand Through 2050
National Renewable Energy Laboratory
Distribution-transformer demand and supply-chain research.
Grid Infrastructure Investments Drive Utility Spending
U.S. Energy Information Administration
Transmission and distribution investment context.
Source links are provided to support independent review. They do not constitute endorsements, and external sources may update their materials after publication.
Access the Full Report
The complete 12-page report includes the full analytical framework, sources, and disclosures.
Download Full PDF — 12 pages