EU Carbon €67.42 +2.1%
US REC (National) $3.85 -0.8%
UK Baseload £48.20/MWh +5.3%
DE Grid Load 58.2 GW -1.2%
US Solar Cap 192.4 GW +0.4%
EU Wind Output 142.8 TWh +3.7%
EU Carbon €67.42 +2.1%
US REC (National) $3.85 -0.8%
UK Baseload £48.20/MWh +5.3%
DE Grid Load 58.2 GW -1.2%
US Solar Cap 192.4 GW +0.4%
EU Wind Output 142.8 TWh +3.7%
Wind turbines at dusk

Data-Driven Energy Analysis

How the world's energy systems actually work

Analysis of power grids, data center energy, and renewable infrastructure. No spin, just data.

View latest analysis
AI & Data Centers

Data Centers Are Abandoning the Grid. Here’s Where They’re Going Instead.

A new report reveals the geographic winners and losers as AI infrastructure seeks power-friendly regions.

The relationship between data centers and the electrical grid is fracturing.

According to Bloom Energy’s 2026 Data Center Power Report, released this week, data center operators are increasingly designing facilities to operate independently of utility grids. The shift isn’t ideological—it’s practical. Power availability has become the single largest constraint on data center growth, reshaping where facilities get built and how they operate.

The numbers are stark: total U.S. data center IT load could nearly double from 80 GW in 2025 to over 150 GW by 2028. The grid, much of it built between the 1950s and 1970s, simply cannot add capacity fast enough.

Texas Wins, California Loses

The report identifies clear geographic winners and losers in the race for AI infrastructure.

Winners:

  • Texas is projected to capture nearly 30% of U.S. data center market share by 2028—more than double its current position
  • Georgia’s market share is expected to grow 75%, from 4% to 7% of the national total

Losers:

  • California, Oregon, Iowa, and Nebraska are each projected to lose more than 50% of their relative market share
  • Permitting complexity and power constraints are driving developers elsewhere

The shift to Texas isn’t surprising. The state offers abundant energy resources, a business-friendly regulatory environment, and crucially, the ability to secure power capacity faster than constrained coastal markets.

The Rise of Gigawatt-Scale Campuses

Perhaps more significant than the geographic shift is the scale transformation underway.

The report projects that over half of new data center campuses will exceed 500 MW by 2035. Nearly one-third will exceed 1 GW—each consuming roughly as much electricity as the entire city of San Francisco.

At this scale, relying on utility connections becomes increasingly impractical. Utilities are already projecting delivery timelines 1.5 to 2 years longer than what hyperscalers expect. That gap is driving the move toward onsite generation.

What This Means for Grid Operators

The implications for utilities are significant. Data centers have been a growth engine for electricity demand—one of the few bright spots in an otherwise flat load environment. If major facilities increasingly generate their own power, utilities lose both the revenue and the load-balancing benefits that large, predictable customers provide.

On the other hand, data centers that do remain grid-connected are becoming more sophisticated partners. Many are co-investing in transmission infrastructure and offering load flexibility programs that can help stabilize grids during peak demand.

The Bottom Line

Power availability has moved from a planning consideration to a strategic constraint that’s reshaping the entire data center industry. Developers who can secure reliable power—whether from the grid or their own generation—will capture the AI infrastructure buildout. Those who can’t will watch from the sidelines.

For grid operators and utilities, the message is clear: adapt to serve these customers at speed, or lose them entirely.

Source: Bloom Energy 2026 Data Center Power Report, January 2026

Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Articles

AI & Data Centers

What is PUE? The Complete Guide to Power Usage Effectiveness

Key Takeaway PUE measures how efficiently a data center uses energy. A...

AI & Data Centers

What Is Edge Computing? How Processing at the Network Edge Changes Infrastructure

What Edge Computing Means Edge computing is an architectural approach that moves...

AI & Data Centers

What Is Data Center Water Usage? The Hidden Resource Demand

Why Data Centers Use Water Water is one of the most efficient...

AI & Data Centers

How Data Centers Connect to the Power Grid: Utility Interconnection Explained

Why Grid Connection Is the Bottleneck A data center is only as...