U.S. Data Center Load Forecasts Exposed: The Hidden Risks Behind the 90-GW Power Narrative (2026 Guide)

Reading Time: ~6 minutes

---

Why U.S. Data Center Load Forecasts Suddenly Matter So Much

Few planning assumptions have shifted as fast, or as dramatically, as U.S. data center load forecasts. In just three years, long-range utility projections ballooned from manageable increments to eye-watering numbers that threaten to reshape capital plans, bond schedules, land use, and grid reliability strategies.

At the center of the debate sits one headline: 90 gigawatts (GW) of new data-center demand by 2030. For municipal utilities, this single number can trigger billions in over- or under-investment if accepted uncritically.

But is 90 GW real, or is it a modeling mirage?

Recent analysis suggests the truth is more nuanced. For cities balancing rate stability, resilience, and economic development, understanding what's behind these forecasts is no longer optional, it's essential.

Why U.S. Data Center Load Forecasts Suddenly Spiked

From 24 GW to 166 GW: A 6× Jump Explained

Between 2021 and 2024, five-year peak-load forecasts reported by U.S. utilities surged from 24 GW to 166 GW, according to industry tracking. The primary driver wasn't population growth or electrification, it was data-center announcements, particularly hyperscale and AI-oriented facilities.

Utilities, under pressure to respond quickly, incorporated these announcements directly into long-range plans. The result was a wave of aggressive projections that now dominate boardrooms and council chambers.

The Load-Factor Assumption That Changed Everything

A recent Grid Strategies study pinpointed a critical modeling flaw: unrealistically high load-factor assumptions.

Many utilities assumed:

• 95 to 100% utilization
• Near-constant, year-round demand
• Minimal downtime or phase-in delays

In reality, most data centers:

• Ramp up in phases
• Experience workload variability
• Do not operate at full capacity continuously

This single assumption inflates projected 2030 demand from a market-based ~65 GW to the now-familiar 90-GW headline.

Translation for Municipal Leaders

If your long-range plan relies on the 90-GW assumption, you may be:

• Over-allocating capital by up to 28%

• Reserving excess land for substations

• Over-building transmission headroom years too early

For rate-sensitive municipal systems, that margin matters.

Zooming Out to 2035: Are 106 GW U.S. Data Center Load Forecasts Realistic?

BloombergNEF's Bullish Outlook

BloombergNEF's December outlook raised eyebrows by projecting 106 GW of U.S. data-center demand by 2035, a 36% jump from its own April estimate.

The drivers include:

• Roughly 150 mega-projects

• About 25% exceeding 500 MW

• Geographic spread beyond Northern Virginia into Texas and the rural Midwest

On paper, the momentum looks unstoppable.

The Caveats Behind the Headline

Even BloombergNEF tempers its forecast with major qualifiers:

• AI hype cycles may cool

• Chip supply constraints could slow deployments

• Developers often shop the same project across multiple utilities

This means a single proposed facility may appear in multiple forecasts, but get built only once, or not at all.

What This Means for Long-Range Planning

For municipal utilities, treating 106 GW as inevitable could:

• Lock in stranded assets

• Force premature bond issuances

• Crowd out funding for resilience and cybersecurity

The smarter move is scenario planning, not headline chasing.

What U.S. Data Center Load Forecasts Mean for Municipal Utilities

Avoid Single-Scenario Budgeting

Municipal utilities should fold three demand cases into integrated resource plans:

• Low case: ~65 GW (market-based)

• Mid case: ~90 GW (utility consensus)

• Stretch case: ~106 GW (aggressive AI adoption)

This approach preserves flexibility while keeping policymakers grounded.

Stress-Test Capital Expenditure Timing

Hyperscale projects often slip 6 to 8 months due to permitting, financing, or supply chains. That delay can:

• Push substation upgrades into more favorable bond markets

• Defer peak-load exposure

• Reduce near-term rate pressure

Timing is as important as capacity.

Leverage Flexible Capacity Instead of Fixed Assets

Traditional generation takes 4 to 7 years to deliver. By contrast:

• Battery peakers

• Demand-response contracts

• Modular microgrids

...can add capacity in 6 to 12 months, aligning better with uncertain load growth.

Press Developers on Transparency

Before issuing firm interconnection agreements, require disclosures on:

• Competing utility negotiations

• Phase-in schedules

• Cancelation clauses

This protects cities from planning around phantom load.

Align IT and OT Security Early

More compute means more cyber exposure. Municipal utilities should:

• Embed NIST 800-171 or CJIS controls

• Align IT and OT security requirements

• Include cybersecurity language in power-purchase and pole-attachment contracts

Early alignment avoids costly retrofits later.

Action Checklist for City Leadership

Frequently Asked Questions (FAQs)

1. Why are U.S. data center load forecasts so different across reports?

Because assumptions vary widely, especially around load factors, utilization rates, and project certainty.

2. Is the 90-GW forecast wrong?

Not necessarily, but it reflects aggressive assumptions that may not hold universally.

3. Why is 65 GW considered more realistic?

It's based on observed utilization patterns and phased deployments rather than theoretical maximums.

4. Should municipal utilities plan for the 106-GW scenario?

Yes, but only as a stretch case, not the baseline.

5. How can cities avoid stranded infrastructure?

By prioritizing modular, flexible assets and delaying irreversible investments.

6. Do data centers really pose cybersecurity risks to utilities?

Yes. Increased interconnection expands the attack surface, making early IT/OT alignment critical.

Flexibility Is the New Forecasting Advantage

Forecasting electric load is getting harder just as the stakes get higher. Municipal utilities that treat U.S. data center load forecasts as ranges, not certainties, will be better positioned to:

• Stabilize rates

• Protect reliability

• Remain attractive to responsible data-center investment

Cities that build flexibility into both kilowatt projections and procurement playbooks will win, regardless of which forecast proves closest.

Need a sounding board?
Our municipal-focused MSP practice helps cities model realistic load, secure grant funding, and deploy NERC-ready IT/OT architectures without blowing past staffing limits.

Contact us to get the conversation started

616-828-4416 option 2

Sales@Ipconsultinginc.com