The latest AI infrastructure debate is moving from chip supply to electricity allocation.
That shift matters because grid investment is shared, while large computing loads are concentrated.
The question is becoming who pays, who receives priority, and who carries the reliability risk.

The U.S. Energy Information Administration expects national electricity sales to reach 4,135 billion kilowatt-hours in 2026.
That would be an increase of almost 2 percent from 2025.
The agency forecasts another increase of almost 2 percent in 2027, reaching 4,211 billion kilowatt-hours.
EIA identifies data center development and manufacturing activity as major drivers of commercial and industrial growth.

A PM can summarize research, draft a PRD, and mock up a prototype before lunch. The hard part starts when the team has to decide what actually gets built.

Jira Product Discovery gives product teams one place to capture insights, prioritize ideas with consistent frameworks, and build living roadmaps stakeholders can rally around.

And because it’s connected to Jira, the context behind every decision stays with the work—so developers and their agents know not just what to build, but why.

AI helps PMs move faster. Jira Product Discovery helps the whole team build with confidence.

Get Started for Free

The regional pattern is more revealing than the national total.
EIA expects the West South Central region to provide nearly 20 percent of nationwide electricity-sales growth in 2026.
The region could provide nearly 40 percent of that growth in 2027.
Texas has paused new data center connections while regulators audit proposed projects.
Even with that pause, the region remains the largest contributor in the forecast.

The generation mix adds another layer to the story.
EIA expects natural-gas generation to rise 2 percent in 2026 and 1 percent in 2027.
Coal generation is forecast to fall 8 percent in 2026 and 6 percent in 2027.
Solar generation is forecast to grow 21 percent in 2026 and 18 percent in 2027.
Wind generation is forecast to grow 7 percent in 2026 and 5 percent in 2027.
These figures describe a system adding supply while also changing its fuel mix.
They do not prove that AI alone caused any single generation increase.

The cost question became more visible in reporting from an Axios Live event.
The report said the House passed the Ratepayer Protection Act after postponing an earlier vote.
The bill aims to prevent existing customers from absorbing costs linked to new data centers.
The report also described Duke Energy agreements designed to make data centers cover infrastructure costs.
That is a company approach, not evidence of a nationwide rule.

This guide distills 10 AI strategies from industry leaders that are transforming marketing.

Outpace your competitors by mastering AI.

Get Your Free Guide

The same report said the EIA is preparing a weekly electricity report.
That would make changes in electricity use easier to track as AI workloads expand.
More frequent reporting could improve public review of forecasts, connection requests, and regional stress.
It would not by itself resolve disagreements over cost recovery or project approval.

Interconnection delays are another constraint on the computing buildout.
AWS reported that more than 2,000 gigawatts of generation and storage sought U.S. transmission interconnection at the end of 2025.
The company cited Lawrence Berkeley National Laboratory for that figure.
AWS also said projects entering operation in 2025 had spent a median of more than five years in the process.
These figures describe the queue, not future capacity.

The operational response is also changing.
AWS described a grid-planning system where agents coordinate data preparation and simulations.
Engineers still direct studies and make final engineering decisions, according to the company.
A versioned record can make automated steps easier to inspect after a study is complete.
That benefit depends on accurate grid models, controlled system access, and meaningful human review.

The current evidence points to a governance problem with technical roots.
AI data centers create concentrated loads, but grid upgrades affect wider customer groups.
Forecast transparency, project-level cost rules, and public reporting therefore matter alongside new generation.
The next useful signals will be updated load data, regulatory decisions, and evidence about who actually pays.
This issue is a research briefing, not a forecast of any single market outcome.

The next wave of millionaires will be people who figured out how to make AI work for them.

The window to get ahead is still open. But not for long.

Here are 200+ proven ways to make money with AI in 2026.

Sign up for Superhuman AI, the free daily newsletter read by 1M+ professionals, and get instant access to all 200+ ways to profit from AI this year.

Claim your free list