Us Data Centers Midterms

Why America's Power Grid Is Struggling to Meet Data Center Demand

Back to Blogs
Blog Img

Why America's Power Grid Is Struggling to Meet Data Center Demand

Key Takeaways

The data center boom is placing enormous strain on America's power grid, with over $5 trillion in planned investments through 2030 colliding with aging infrastructure and rising electricity costs for ordinary households.

  • Data center electricity demand could triple by 2028, growing from 176 TWh (4.4% of U.S. consumption) to potentially 580 TWh, driven primarily by AI workloads expanding at 30% annually.

  • Residential electricity prices surged 11.5% in 2025, with some high-concentration states seeing increases up to 267% over five years as utilities spread infrastructure costs across all ratepayers.

  • New cost-sharing models are emerging nationwide, requiring data centers over 50–100 MW to pay full infrastructure upgrade costs rather than passing expenses on to residential customers.

  • Energy efficiency offers immediate relief: state-of-the-art facilities achieve a Power Usage Effectiveness of 1.03, using 84% less overhead energy than older designs, while advanced cooling cuts consumption by 40%.

  • Grid modernization requires $2 trillion by 2030, yet utilities currently invest only $35 billion annually in transmission — a funding gap that threatens both AI growth and grid reliability.

The challenge is not to choose between technological progress and affordable energy. It is to decide who pays, improve efficiency and invest in the grid quickly enough to support both.

Introduction

Local opposition blocked or delayed at least 120 data center projects worth almost $200 billion in the first half of this year alone1. That figure, striking as it is, only tells part of the story. Tech giants Oracle, Meta, Google, Amazon, and Microsoft collectively expect to invest more than $700 billion into U.S. data centers this year2, part of a broader $5 trillion capital commitment running through 20303. The scale of ambition is remarkable. So is the backlash.

A debate that once sat largely within the infrastructure industry is now a mainstream political issue. Gubernatorial campaigns in several states are being shaped by questions about who pays for the power, how water use should be managed and whether local communities should be able to reject projects altogether. In Texas, regulators were ordered to protect residents from higher electricity bills2 as pressure on the grid became harder to ignore. The midterms have therefore become a test of how the US intends to reconcile its AI ambitions with the limits of an ageing power system. There is still no settled answer.

Why the Midterms Matter for Data Center Expansion

Congress Holds the Purse Strings

The argument over who pays for AI infrastructure has reached Capitol Hill. Brett Guthrie, Chair of the House Energy and Commerce Committee, wants to amend the Public Utility Regulatory Policies Act of 1978 so that data centers and other large electricity users cover the full cost of the infrastructure required to serve them4. Senate Democrats are unconvinced. Minority Leader Chuck Schumer described a Republican-sponsored ratepayer protection bill as “toothless”5. With Congress at an impasse, investors are watching the elections closely: the next Congress will have a significant influence on the policy environment for AI infrastructure across the country3.

Gubernatorial Races as the New Frontier

For now, the more immediate fight is at state level. In August 2026, Pennsylvania Governor Josh Shapiro reversed his position, removed data center projects from fast-track permitting and required developers to fund their own electricity use. The decision followed reports from communities facing more than 100 proposed projects and growing pressure on local grids6. A month later, Texas Governor Greg Abbott ordered regulators to pause permit reviews after polling found that 57% of voters opposed data centers in their communities7. New York Governor Kathy Hochul had already signed a one-year moratorium in July6. These are not isolated moves: gubernatorial candidates in at least sixteen states are now calling for temporary moratoriums on data center development8.

The $5 Trillion Question

The politics are sensitive because the sums involved are so large. Hyperscalers plan to invest more than $5 trillion in technology and data centers by 20309. Goldman Sachs Research puts combined capital spending by the leading technology companies at $5.3 trillion between 2025 and 20309. By the end of the decade, annual funding needs could reach $1.4 trillion, drawing on every part of the capital markets10. JPMorgan also estimates a $1.4 trillion funding gap that may have to be filled by private credit and government support10. Decisions made now in state legislatures and at the ballot box will have a direct bearing on whether that money is deployed or delayed by moratoriums and local opposition.

Ballot Measures and the Ground-Level Revolt

Voters will consider 34 data center-related ballot measures in November across California, Florida, Illinois, Kansas, Michigan, Nevada, Ohio and Wisconsin7. Ohio alone has 20 ballot initiatives spread across 17 jurisdictions7. Local governments adopted 294 moratoriums and bans between January and July, with more than 50 data center projects canceled in 2026 amid mounting community opposition7. Michigan Governor Gretchen Whitmer has called on her legislature to formally codify Michigan Public Service Commission protections requiring data centers to pay full system costs11. The message from communities is becoming increasingly consistent — ambition at the federal level must be matched by accountability at the local one.

Understanding the Data Center Power Debate and Grid Pressure

How Much Electricity Do Data Centers Actually Use?

In 2023, US data centers used about 176 terawatt-hours of electricity, equivalent to 4.4% of national consumption12. Worldwide use reached roughly 415 TWh in 2024, or about 1.5% of global electricity demand13. Those totals are expected to rise quickly. US consumption could reach between 325 and 580 TWh by 202812, while global demand is projected to hit 945 TWh by 203013.

AI is the main source of that growth. Data center electricity use is forecast to increase by 15% a year between 2024 and 2030, more than four times the rate across all other sectors combined13. AI-accelerated servers account for almost half of the net increase, with their electricity consumption rising by 30% annually13. The internet changed how people communicate; AI is now changing the amount of electricity needed to support the digital economy.

The Rising Cost Burden on Residential Ratepayers

For ordinary households, this growth is not an abstract policy question — it is showing up directly on electricity bills. Residential electricity prices rose 11.5% in 2025, outpacing inflation14. In states with high data center concentrations like Virginia, prices increased by up to 267% over the past five years14. Utilities requested more than $29 billion in rate increases during just the first half of 2025 — double the amount from the prior year14.

The outlook is equally troubling. ICF projects that, without new tariff structures, residential electricity rates will rise by 15% to 40% by 2030 and could double for customers of some utilities by 205015. The burden is uneven. Low-income households already spend as much as 20% of their income on energy, compared with 3% for higher-income households14. Spreading the infrastructure costs of industrial-scale computing across all ratepayers therefore hits hardest those with the least room in their budgets.

Why Existing Grid Infrastructure Can't Keep Up

The demand side of this equation is growing rapidly. The supply side is struggling to respond at anything like the same pace. Nearly 2,300 gigawatts of generation and storage capacity are currently stuck in U.S. interconnection queues — a figure that actually exceeds the country's entire installed power capacity16. Interconnection wait times have stretched beyond five years in many regions16.

Equipment shortages add another layer of delay. Transformers can now take up to three years to arrive, and orders for new gas turbines extend into the next decade17. Of the capacity that applied for interconnection between 2000 and 2019, only 13% had entered commercial operation by 202416. The queue is not merely slow; most proposed capacity never makes it through.

Regional Hotspots Facing the Biggest Challenges

The impact varies sharply by region. Data centers already account for more than 25% of Virginia’s electricity use18, an unusually high concentration for a single industry. In Texas, ERCOT estimates that peak demand could rise from 87 GW to 138 GW by 2030 if every proposed data center is built17. That would amount to a major redesign of the grid, not a routine increase in capacity.

These two states alone illustrate why the data center power debate has moved from the pages of trade publications to the centre of state-level political campaigns.

Who Should Pay for Data Center Infrastructure?

A Century-Old Model Meets a 21st-Century Problem

For more than a century, US utilities have spread infrastructure costs across all ratepayers on the basis that everyone benefits from investment in the grid. That approach made sense when demand grew gradually across homes and businesses. Data centers challenge the assumption. Some facilities now under development will use more electricity than large cities19. When utilities spend billions on new generation and transmission to serve them, households can end up sharing the bill. Many operators also negotiate confidential agreements with utilities, sometimes on preferential terms, while regulatory proceedings involve only the utility and the data center company. Ordinary ratepayers are left without a voice1919.

What Legislators Are Proposing

Lawmakers agree that the current arrangement needs to change, although not on the remedy. Senator Adam Schiff’s Energy Cost Fairness and Reliability Act would require data centers above 50 megawatts to secure their own power and pay the full cost of network upgrades, including transmission lines and transformers20. The House passed the Ratepayer Protection Act by 417 votes to 3. It directs states to require loads above 100 MW to cover the full incremental cost of generation, transmission and distribution upgrades2122. Senators Josh Hawley and Richard Blumenthal have proposed the GRID Act as an alternative23. The proposals differ, but they all point towards large users paying more of the costs they create.

What States Are Already Doing

Several states haven't waited for federal consensus. California's State Senate passed SB 886, compelling the Public Utilities Commission to establish a dedicated tariff structure ensuring that grid investments driven by data centers are fully recovered from those operators24. Ohio regulators reached a similar conclusion, determining that data centers must front more of the upfront costs associated with their power demands24. Virginia and New Jersey have both signalled their intention to pursue fair-share cost recovery mechanisms24. The patchwork of state-level action reflects the urgency of the problem — and the frustration with the pace of federal deliberation.

The Incremental Cost Debate

The central question is straightforward: who should pay to upgrade an ageing grid for demand on this scale? In areas with a high concentration of data centers, residents have seen electricity costs rise by as much as 267%2020. Technology companies reasonably point to the jobs, investment and tax revenue their facilities can bring. Opponents argue that those benefits are widely dispersed while higher bills arrive quickly and fall hardest on people least able to afford them. Clearer legislation will be needed to settle the trade-off.

Water: The Overlooked Dimension

The cost debate has tended to focus on electricity, but water is an equally pressing concern that has attracted far less scrutiny. Large data centers can consume up to 5 million gallons per day — comparable to a town of between 10,000 and 50,000 people25. Across the United States, data centers were consuming 449 million gallons of water daily as of 202125. Approximately 56% of facilities still rely on evaporative cooling systems26, though several major operators have committed to water-positive operations by 203027. The more promising direction involves partnerships with utilities to use treated wastewater for cooling rather than drawing on fresh drinking water supplies — an approach gaining traction particularly in regions with established reclaimed water infrastructure28. It is an area where good intentions will need to be backed by operational commitments and regulatory accountability in equal measure.

Building a Smarter Energy Future: Beyond Just More Power

The Efficiency Lesson America Has Already Learned Once

There is a useful precedent. Before the early 2000s, US electricity demand rose by as much as 30%. It then levelled off, not because the country built its way out of the problem, but because improvements in energy efficiency reduced the rate of growth29. The same lesson applies now.

Total energy demand could grow by 15–20% over the next decade29. More generation will be needed, but it cannot be the entire answer. A credible response also requires storage, stronger grid infrastructure, cleaner generation and better management of demand. Previous growth cycles show why efficiency should be treated as essential infrastructure, not an afterthought.

Energy Efficiency: The Fastest Win on the Board

The DOE’s national laboratories show what efficient design can achieve. Their exascale computing facilities recorded Power Usage Effectiveness of 1.0329. Facilities operating at a PUE of about 1.1 can use roughly 84% less overhead energy than those near 2030. In practical terms, design quality makes an enormous difference to power demand.

The hardware choices matter just as much as the building design. Advanced cooling technologies such as Lenovo's Neptune water-cooling system cut power use by up to 40%, while considered server selection alone can reduce energy consumption by as much as 75%31. The DOE's Energy Efficiency Scaling initiative takes the longer view, targeting a thousandfold improvement in microelectronics efficiency over two decades29. These are not marginal gains — they represent a meaningful alternative to simply demanding more from an already strained grid.

Modernising the Grid and Clean Energy Together

Solar, land-based wind, battery storage, and energy efficiency remain the most rapidly scalable and cost-competitive options for meeting increased electricity demand29. The National Council of State Legislatures estimates the U.S. will spend up to $2 trillion on grid modernisation by 203032. The challenge is that utilities currently invest roughly $35 billion annually in transmission infrastructure33 — a significant shortfall relative to what the moment actually demands.

States are beginning to respond with creative tariff structures. Louisiana and Ohio, for instance, are designing frameworks that front-load development costs for urgently needed capacity while protecting residential customers from bearing the brunt of those investments34. These models are worth watching closely, as they may offer a template others can adapt.

Data Centers as Part of the Energy Solution

Data centers do not have to be passive consumers. A growing number are acting as “prosumers”, helping to manage the grid as well as drawing power from it34. That gives operators a potential role in supporting grid stability rather than simply adding to demand.

The Nordic countries offer a striking example of what this can look like at scale: waste heat from data centers is piped directly into district heating networks, warming homes through the winter34. Ireland, having ended its grid connection moratorium, commissioned its first large microgrid installations specifically designed around data center loads. Greece, meanwhile, has repurposed conventional generation assets alongside hydro reservoirs and solar capacity to create hybrid microgrids capable of meeting large-scale computing demand34.

Nuclear power is also returning to the discussion. Small modular reactors are gaining something they have long lacked: customers willing to commit to long-term demand. Data center operators need reliable, low-carbon power and can provide the certainty required to support investment2934. The economics help. Electricity represents 60–70% of data center operating costs, while some facilities achieve EBITDA margins of up to 80%34. That leaves room for operators to shoulder some of the risk attached to newer energy technologies.

The Workforce Dimension

Alongside the infrastructure and technology questions sits a workforce challenge that deserves equal attention. Data center employment rose 60% between 2017 and 202335, and the industry may need an additional 140,000 skilled tradespeople by 2030. These are not entry-level positions — electricians, HVAC technicians, pipefitters, welders, and facilities technicians working in this sector typically earn 25-30% more than counterparts in other industries35.

The supply of that talent is not keeping pace with demand. Some 52% of data center equipment manufacturers, engineers, and construction companies reported business disruptions in 2025 due to staffing shortages35. A recurring theme in the clean energy space is how these transitions can generate real economic opportunity for working communities. Data center expansion is no different — the question is whether workforce development investment arrives early enough to match the pace of build-out.

Conclusion

America’s AI ambitions are running into the physical limits of its power system. Meeting the challenge will require more than political pledges. Data center operators should pay a fair share of the infrastructure costs they create and invest more heavily in efficient hardware, advanced cooling and low-carbon power. Utilities, meanwhile, need faster transmission development and modernised grids. If those responsibilities are clear, the US can expand its digital economy without asking households to underwrite the cost or weakening the reliability of the electricity system.

References

[1] - https://www.bloomberg.com/news/articles/2026-10-10/how-data-centers-are-driving-up-us-electricity-prices
[2] - https://www.pbs.org/newshour/politics/data-center-controversies-have-transformed-midterms-heres-where-different-groups-stand
[3] - https://www.cnbc.com/2026/10/09/the-ai-election-wall-street-braces-for-midterms-impact-on-data-centers.html
[4] - https://www.thompsoncoburn.com/insights/congress-struggles-with-data-center-policy/
[5] - https://www.wired.com/story/midterm-races-decide-data-centers-political-fate/
[6] - https://www.ap.org/news-highlights/elections/2026/governors-races-are-being-increasingly-buffeted-by-the-toxic-politics-of-data-centers/
[7] - https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/100126-data-center-backlash-shapes-midterm-ballots-campaigns-across-us
[8] - https://www.multistate.us/insider/2026/9/18/gubernatorial-candidates-split-on-data-center-policy-in-2026
[9] - https://www.goldmansachs.com/insights/articles/private-markets-expected-to-have-growing-role-in-data-center-financing
[10] - https://www.datacenterdynamics.com/en/news/jpmorgan-global-data-center-and-ai-infra-spend-to-hit-5-trillion-demand-for-compute-remains-astronomical/
[11] - https://www.michigan.gov/whitmer/news/press-releases/2026/07/15/gov-whitmer-data
[12] - https://www.congress.gov/crs-product/R48646
[13] - https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai
[14] - https://www.eesi.org/articles/view/data-center-power-demands-are-contributing-to-higher-energy-bills
[15] - https://www.brookings.edu/articles/the-pledge-to-protect-ratepayers-from-ai-data-center-costs-needs-enforcement/
[16] - https://www.hanwhadatacenters.com/blog/data-center-grid-limitations-the-power-bottleneck/
[17] - https://www.domesticpreparedness.com/articles/the-data-center-dilemma-understanding-americas-new-grid-challenge/
[18] - https://www.electricchoice.com/datacenters/
[19] - https://hls.harvard.edu/today/how-data-centers-may-lead-to-higher-electricity-bills/
[20] - https://www.schiff.senate.gov/news/press-releases/icymi-sen-schiff-introduces-legislation-to-ensure-fair-and-affordable-energy-costs-for-americans-amid-data-center-buildouts/
[21] - https://www.politico.com/news/2026/09/16/house-passes-bill-data-center-power-costs-01076743
[22] - https://www.utilitydive.com/news/house-passes-ratepayer-protection-bill-data-centers/830658/
[23] - https://www.politico.com/news/2026/09/18/senate-competing-data-center-bills-01083204
[24] - https://sd18.senate.ca.gov/news/bills-protecting-ratepayers-data-center-costs-and-incentivizing-sustainable-tech-1
[25] - https://www.eesi.org/articles/view/data-centers-and-water-consumption
[26] - https://www.usatoday.com/story/opinion/columnist/2026/08/19/data-center-water-use-misunderstood/91336968007/
[27] - https://submer.com/blog/ai-data-centers-water-myths/
[28] - https://www.fwpcoa.org/content.aspx?page_id=5&club_id=859275&item_id=130961
[29] - https://www.energy.gov/oe/clean-energy-resources-meet-data-center-electricity-demand
[31] - https://www.ibm.com/think/topics/data-center-management
[32] - https://cleanenergyforum.yale.edu/2025/11/12/grid-modernization-for-data-center-and-ai-loads-how-the-grid-is-gearing-up
[33] - https://www.renewableenergyworld.com/power-grid/data-centers-are-straining-the-grid-can-they-be-forced-to-pay-for-it/
[34] - https://www.mckinsey.com/industries/energy-and-materials/our-insights/blog/data-centers-catalyzing-innovation-in-energy
[35] - https://www.streamdatacenters.com/articles/the-truth-about-data-center-talent/