---
title: "IEA's Data Center Power Warning: What It Means for Sites | Insights | Build"
description: "The IEA says data center electricity use could roughly double to about 1,000 TWh a year by 2030, and that queues in Texas, Virginia and elsewhere exceed peak demand. Here is what that means for choosing and validating sites."
url: https://build.inc/insights/iea-data-center-power-warning-what-it-means-for-sites
---

[Industry](https://build.inc/insights?category=industry)

# IEA's Data Center Power Warning: What It Means for Sites

The IEA says data center electricity use could roughly double to about 1,000 TWh a year by 2030, and that queues in Texas, Virginia and elsewhere exceed peak demand. Here is what that means for choosing and validating sites.

by Build Team • November 20, 2026 • 6 min read

**Short answer:** On 19 November the International Energy Agency (IEA) warned that data center electricity use could roughly double to about 1,000 TWh a year by 2030, and that tight supply chains, local opposition and limited grid capacity could slow new projects ([CNBC](https://www.cnbc.com/2026/11/19/data-center-electricity-demand-iea.html)). It named Texas and Virginia among the places where the capacity of data centers waiting to connect already exceeds current peak demand. For developers, the message is that a site is only as good as its validated path to power. That is the Build approach to [data center site selection](https://build.inc/learn/ai-site-selection/data-centers): source sites, validate power and permits against primary sources, then bundle the services that carry a site into design and construction.

## What did the IEA say?

According to CNBC's report on the IEA outlook:

-   Data center electricity use could reach around 1,000 TWh a year by 2030 under the IEA's Stated Policies Scenario.

-   Tight supply chains, local opposition and limited grid capacity could slow construction.

-   The IEA cited large data center pipelines in Texas, Virginia, Finland and the U.K. In several regions, the combined capacity of data centers waiting to connect exceeds current peak demand.

-   Without grid upgrades, new projects could face longer waits to connect, and networks could become more congested.

-   Peak electricity demand is projected to rise 40% to 50% by 2035, and supplies of copper, cables and transformers are already tight.

S&P Global reported that the figures come from the IEA's World Energy Outlook 2026 and that grid connection queues are measured in years in several markets ([S&P Global](https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/111926-ai-boom-set-to-double-data-center-power-demand-by-2030-iea)).

## Why does a queue bigger than peak demand matter?

It means most of the queue will not connect on the schedule its sponsors expect. The IEA's earlier analysis notes that connection queues contain duplicative or speculative projects, and that many jurisdictions are reforming queues with stronger entry requirements, maturity assessments, financial commitments and milestone-based release of capacity ([IEA, Powering Artificial Intelligence](https://www.iea.org/reports/powering-artificial-intelligence/executive-summary)).

For a developer, two things follow:

1.  **A queue position is not a power date.** If the queue is larger than the grid, the useful question is where your project sits among credible projects, and what upgrades stand between it and energization.

2.  **Rules are moving toward proof.** Queue reforms reward projects that can show site control, studies and commitments. Evidence gathered early becomes an advantage, not paperwork.

## What does this change for site selection?

Before

Now

Screen for proximity to substations and lines

Confirm deliverable capacity and its date, not only proximity

Treat a queue position as progress

Test the position against queue size, upgrades and reform rules

Assume equipment arrives on time

Plan for long-lead transformers, switchgear and cables

Expect approval to follow power

Treat local opposition and water as schedule risks from day one

This table is Build's framing of the IEA findings, not an IEA table. The gap between nameplate and real capacity is covered in [available vs deliverable capacity](https://build.inc/insights/available-vs-deliverable-data-center-capacity), and the evidence that proves it in [data center power analysis](https://build.inc/insights/data-center-power-analysis-site-selection).

## What about Texas and Virginia specifically?

Both are named in the IEA's examples of large pipelines. In Texas, state agencies have also been reviewing how data centers affect communities; see [Texas resumes data center permits](https://build.inc/insights/texas-resumes-data-center-permits). In PJM, which covers Northern Virginia, new rules for large loads are arriving; see [PJM fast energization](https://build.inc/insights/pjm-fast-energization-data-centers) and [PJM ride-through rules](https://build.inc/insights/pjm-ride-through-rules-data-centers). In both markets, a site with a validated power path is scarcer than a site with land.

## What should developers do now?

-   **Validate power before land spend.** Get the utility's view of capacity, the upgrade path and the date in writing where possible. Our explainer on [will-serve letters](https://build.inc/insights/data-center-utility-will-serve-letter) covers what they do and do not prove.

-   **Price in equipment lead times.** Long-lead items belong in preconstruction, not construction. See [data center construction](https://build.inc/learn/ai-pre-construction/data-centers).

-   **Plan for the community.** Water use, noise and visual impact decide approvals as often as power does.

-   **Keep the shortlist live.** When one site stalls in a queue, the next validated site should already be ready.

## What does Build do?

Build is the development execution engine for infrastructure and data centers, built for quality and speed. We work the way the IEA's warning suggests the market now has to: sites first, then validation of power, zoning, water and fiber against primary sources, then the bundled services that move a passing site into design, engineering and construction. That is what we mean by [AI native development](https://build.inc/learn/ai-native-development).

## Key takeaways

-   The IEA says data center electricity use could roughly double to about 1,000 TWh a year by 2030.

-   In Texas, Virginia and other markets, the capacity waiting to connect already exceeds current peak demand, so most queued projects will wait.

-   Validated deliverable power, not land, is the scarce input. Validate it before committing to a site.

## Sources

-   [CNBC: IEA says grid investment critical as data center electricity demand soars (19 Nov 2026)](https://www.cnbc.com/2026/11/19/data-center-electricity-demand-iea.html)

-   [S&P Global: AI boom set to double data center power demand by 2030, IEA (19 Nov 2026)](https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/111926-ai-boom-set-to-double-data-center-power-demand-by-2030-iea)

-   [IEA: Powering Artificial Intelligence, executive summary](https://www.iea.org/reports/powering-artificial-intelligence/executive-summary)

## Frequently Asked Questions:

What did the IEA say about data center power demand?

On 19 November the IEA said data center electricity use could roughly double to about 1,000 TWh a year by 2030 under its Stated Policies Scenario, and that limited grid capacity could slow new projects.

Which markets did the IEA name?

It cited large data center pipelines in Texas, Virginia, Finland and the U.K., where the combined capacity waiting to connect exceeds current peak demand in several regions.

Does a queue position guarantee power?

No. When a queue is larger than the grid, many projects will wait or drop out. The question is whether a site has a credible upgrade path and energization date.

Why are queues so large?

The IEA notes that queues include duplicative or speculative projects, and many jurisdictions are tightening entry rules to keep only credible projects.

What should developers do in response?

Validate deliverable power before committing to land, plan for long-lead equipment, treat community issues as schedule risks, and keep a ranked shortlist of validated sites.

How does Build help?

Build sources sites, validates power, zoning, water and fiber against primary sources, then bundles the services that move a passing site toward construction.

[Industry ### 60 MW in Under 8 Months: What Fast Data Center Builds Need November 20, 2026](https://build.inc/insights/fast-data-center-builds-60-mw-under-8-months)[Industry ### Data Center Grid Contracts: What 10 to 20 Years Means November 19, 2026](https://build.inc/insights/data-center-grid-contracts-10-to-20-years)[Industry ### Nvidia's Neocloud Guarantees: What Sites Now Need November 19, 2026](https://build.inc/insights/nvidia-neocloud-guarantees-what-sites-need)
