Price of a Kilowatt-Hour Is Quietly Deciding Where AI Gets Built
A 100-megawatt data center consumes roughly as much electricity as 80,000 homes. At $0.07 per kilowatt-hour, its annual power bill is roughly $61 million. At $0.22 per kilowatt-hour, that same facility costs nearly three times as much to run. The bill tops $190 million per year.
For hyperscalers building at gigawatt scale, the difference is measured in billions. That is the gap between cheap and expensive power over a facility's life. That math is reshaping the geography of AI. Data centers are following electricity, not customers.
The Spread That Reshapes Site Selection
Electricity prices in the United States vary by more than a factor of four. New Mexico's industrial rate sits around 5.43 cents per kilowatt-hour. Texas hovers near 8.26 cents. California averages 21.53 cents — nearly four times New Mexico and nearly three times Texas.
For a 100 MW facility, the annual power cost in New Mexico is about $736,000 per megawatt. In California, the same facility would cost $2.873 million per megawatt. That difference can determine whether a project breaks even or loses money.
The five cheapest states for data center electricity in 2026 are New Mexico, Louisiana, Oklahoma, Texas, and Tennessee. All run below $720,000 per megawatt per year. The most expensive states are California, Hawaii, Connecticut, Rhode Island, and Massachusetts. The pattern is unmistakable: data centers are clustering where power is cheap.
The Old Logic vs. The New Reality
For decades, data center location was driven by proximity to users. Low latency meant being close to population centers. That logic worked when data centers served web pages and streamed video. It does not work for AI training.
Training a frontier model does not require low latency. It requires hundreds of megawatts of electricity, delivered reliably, 24 hours a day, for months at a time. The user is not waiting for the result in real time. The model is training in the background. The location of the compute is irrelevant — except for the electricity bill.
As CBRE noted in its 2026 analysis, "energy cost and availability become the key criteria" in data center site selection. One industry analysis put it even more bluntly: "In 2026, site selection is fundamentally a power strategy."
Developers are shifting away from traditional urban hubs toward smaller metros in states with lower electricity costs. They are choosing locations based on which regions offer the cheapest power. Power availability has become "the primary criteria driving site selection as traditional location and cost factors take a back seat."
Texas: The Clear Winner
Texas is the clearest example of this trend. The state's industrial electricity price sits at roughly 8.26 cents per kilowatt-hour. That is below the national industrial average of approximately 8.9 cents. It is far below California's 21.53 cents.
ERCOT, Texas's grid operator, faces a massive interconnection backlog of over 40 gigawatts. Data centers and cryptocurrency mining facilities are driving much of the demand. Wholesale prices in ERCOT could increase up to 79% by 2027. That assumes high-demand scenarios. It also assumes capacity additions fail to keep pace.
Despite the strain, Texas remains the destination of choice. The state's deregulated electricity market allows sophisticated industrial buyers to hedge against volatility using financial instruments. Commercial rates in Texas average around 9.13 cents per kilowatt-hour. Low rates, available land, and business-friendly regulation made Texas the undisputed leader in AI infrastructure.
Ohio: The Rapid Riser
Ohio tells a different story — one of rapid transformation driven by AI demand.
Central Ohio's explosive data center growth has severely constrained regional transmission boundaries. From February 2025 to February 2026, Ohioans' electricity bills rose 22% — second only to Virginia. Average industrial electricity prices rose 26% in Ohio in the year to December 2025. The national increase was 7%.
The impact is visible across the state. One manufacturer reported its monthly electricity bill jumping from just over $1,600 to more than $12,000. PJM, the regional grid operator serving Ohio and much of the industrial Midwest, has made itself clear. All Ohioans will pay higher rates unless data centers get a separate rate class. Duke Energy has requested a monthly rate increase for residential customers to fund power distribution activities.
Ohio is not the cheapest state for power. Its industrial average sits at 8.73 cents per kilowatt-hour, slightly above the national average. But it offers something Texas does not: proximity to major population centers and existing fiber infrastructure. The trade-off is real. And developers are making it.
The Hidden Costs of Cheap Power
Cheap electricity is not free. It comes with trade-offs.
Texas faces grid reliability challenges. ERCOT is isolated from the rest of the U.S. grid, meaning it cannot import power during emergencies. The state's 2021 winter storm blackouts are still fresh in memory. A 100 MW data center cannot afford to lose power for days. The cost of backup generation and battery storage must be factored into the total cost of ownership.
Ohio faces rising rates. The same data center demand that attracts developers is driving up electricity costs for everyone else. The political backlash is already visible. Ohio gubernatorial candidate Vivek Ramaswamy has proposed a plan. It would require new data centers to give residents free electricity and property tax relief. Ohio House Democrats have urged the governor to act before consumers face higher bills.
New Mexico offers the lowest rates in the country at 5.43 cents per kilowatt-hour. But it lacks the fiber infrastructure and talent pool of Texas or Ohio. The trade-off is real, and developers are weighing it carefully.
Beyond the U.S.
The U.S. is not alone in this shift. In Europe, the gap between cheap and expensive power is even wider. Data center developers in Europe face industrial costs that vary widely by region. Some markets exceed 20 cents per kilowatt-hour. Germany and parts of the U.K. are examples. Ireland's rates have also climbed above 20 cents. Singapore retains the highest pricing in Asia-Pacific at $330 to $475 per kilowatt-month.
China is experiencing a similar dynamic. Zhejiang's new time-of-use electricity pricing policy shows how power cost gaps reshape AI site strategy. Data center operators now evaluate locations by off-peak electricity prices. They also look at solar-plus-storage integration.
The geography of AI is being rewritten by the price of electricity. The math is simple: developers follow the cheapest electrons.
What This Means
The AI industry has spent billions on chips, data centers, and talent. It has spent comparatively little attention on the cost of the electricity that runs it all. That is changing.
A 100 MW facility at $0.07/kWh spends $61 million per year on power. Over a 15-year lifespan, that is nearly $1 billion — just for electricity. The difference between cheap and expensive power can determine whether a project is viable or not.
Developers are responding. They are choosing locations based on power cost, not proximity to users. They are moving to states with low industrial rates, abundant renewable energy, and business-friendly regulation. They are building where the electrons are cheap.
Electricity pricing is now a primary factor in data center site selection. And it is reshaping the geography of the AI industry in ways that will be visible for decades.
Sources: Axis Intelligence U.S. Data Center Electricity Cost by State 2026 (July 29, 2026); Site Selection Group Power Costs for Data Centers Across the U.S.: 2026 Edition (August 11, 2026); EIA Electric Power Monthly (July 23, 2026); Oxford Economics tech companies shift metro strategy for new data centers (February 23, 2026); CBRE Key Factors to Consider for Effective Data Centre Site Selection (February 19, 2026); Enerdatics US Data Center Market Report 2026 (February 6, 2026); The Next Web AI data centers drive Rust Belt factory power bills (July 7, 2026); Dispatch.com Ohio electric bills rising (May 17, 2026); ChargedUp Pro state power bill gap (August 12, 2026); Area Development 2026 Data Center Outlook (January 27, 2026).
Disclaimer
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations
This analysis is based on reporting and public data available as of the article date; figures may be revised as sources update.
Forecasts from third-party analysts can change with market conditions.
Cost and pricing examples are point-in-time estimates; actual rates vary.
Country and company comparisons rely on public reporting, not operational data.
This sector moves fast; timelines and deal terms may be updated later.
Company deals and regulatory rulings may evolve; verify current status.
AI infrastructure is changing quickly; claims can become outdated soon.
Sources
- Axis Intelligence U.S. Data Center Electricity Cost by State 2026 (July 29, 2026)
- Site Selection Group Power Costs for Data Centers Across the U.S.: 2026 Edition (August 11, 2026)
- EIA Electric Power Monthly (July 23, 2026)
- Oxford Economics tech companies shift metro strategy for new data centers (February 23, 2026)
- CBRE Key Factors to Consider for Effective Data Centre Site Selection (February 19, 2026)
- Enerdatics US Data Center Market Report 2026 (February 6, 2026)
- The Next Web AI data centers drive Rust Belt factory power bills (July 7, 2026)
- Dispatch.com Ohio electric bills rising (May 17, 2026)
- ChargedUp Pro state power bill gap (August 12, 2026)
- Area Development 2026 Data Center Outlook (January 27, 2026).
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations: This analysis is based on reporting and public data available as of the article date; figures may be revised as sources update.; Forecasts from third-party analysts can change with market conditions.; Cost and pricing examples are point-in-time estimates; actual rates vary.; Country and company comparisons rely on public reporting, not operational data.; This sector moves fast; timelines and deal terms may be updated later.; Company deals and regulatory rulings may evolve; verify current status.; AI infrastructure is changing quickly; claims can become outdated soon.