For decades, American households have paid roughly the same share of electricity costs regardless of which industries were expanding. AI data centers have broken that arrangement. In 2025, regions with concentrated data center activity saw wholesale electricity prices rise as much as 267% over five years, with the PJM grid operator—serving 65 million people across 13 states—projecting $100 billion in extra consumer costs through 2033 unless something changes.
Trump's announcement that Microsoft will 'make major changes' to shield ratepayers from AI power costs marks the administration's first attempt to address this politically dangerous gap. With tariff-driven inflation already hurting Republicans in off-year elections, rising electricity bills in data-center-heavy states like Virginia and Ohio threaten to become a midterm liability. The question now: Can voluntary corporate commitments actually constrain an industry racing to deploy power-hungry AI infrastructure faster than the grid can accommodate it?
Images from Openverse under Creative Commons licenses.
Videos from YouTube.
Key Indicators
267%
Wholesale electricity price increase
Maximum price increase in areas near data centers over five years, per Bloomberg analysis
$16.6B
Consumer cost for data center demand
Amount PJM consumers will pay for power supplies to meet data center demand from 2025-2027
4,165
U.S. data centers
Total number of data centers in the United States as of November 2025, the most of any country
$600B+
Hyperscaler capex in 2026
Projected capital expenditure by Amazon, Google, Microsoft, Meta, and Oracle, up 36% from 2025
People Involved
Donald Trump
President of the United States (Seeking voluntary commitments from tech companies on energy costs)
Satya Nadella
CEO, Microsoft (First tech executive to commit to consumer electricity cost protections)
Chris Wright
U.S. Secretary of Energy (Leading administration efforts to accelerate data center grid connections)
Elizabeth Warren
U.S. Senator (D-MA) (Leading congressional inquiry into tech company energy practices)
Organizations Involved
MI
Microsoft
Technology Company
Status: First company to commit to consumer electricity cost protections
The world's second-largest company by market cap is building massive AI infrastructure, including a $7+ billion data center campus in Wisconsin.
PJ
PJM Interconnection LLC
Regional Transmission Organization
Status: Managing grid stress from data center expansion
PJM operates the largest electric grid in North America, serving 65 million people across 13 states including Virginia's data center corridor.
FE
Federal Energy Regulatory Commission (FERC)
Federal Agency
Status: Developing new rules for data center grid connections
FERC regulates interstate electricity transmission and has emerged as a key battleground for data center policy.
Timeline
Trump Announces Microsoft Consumer Cost Commitment
Policy Announcement
President Trump announces Microsoft has agreed to 'major changes' to ensure Americans don't pay higher electricity bills due to AI data centers, with more tech company commitments expected.
Democratic Senators Demand Tech Company Energy Transparency
Congressional Action
Senators Warren, Van Hollen, and Blumenthal send letters to Amazon, Google, Microsoft, and Meta demanding detailed responses about data center energy practices by January 12.
FERC Orders PJM to Create New Colocation Rules
Regulatory Order
FERC directs PJM to establish transparent rules allowing data centers to colocate at power plants, with new transmission services to be filed by February 16, 2026.
Virginia Creates Data Center Rate Class
Regulatory Action
Virginia's State Corporation Commission approves special electricity rate class for large data centers, requiring them to pay at least 85% of contracted demand starting January 2027.
Republicans Lose Virginia and New Jersey Governor Races
Political Event
Democrats win gubernatorial elections in Virginia and New Jersey, with affordability concerns—including rising electricity bills—cited as key voter motivations.
DOE Proposes 60-Day Data Center Approval Timeline
Regulatory Proposal
Secretary Wright proposes FERC rule to reduce interconnection approval times for large data centers from over a year to 60 days for curtailable projects.
Microsoft Announces Wisconsin Ratepayer Protection
Corporate Commitment
Microsoft reveals its $7+ billion Wisconsin data center will include utility rate structures ensuring the company pays for infrastructure built to serve it, preventing cost spillover to residential customers.
FERC Initiates Data Center Colocation Review
Regulatory Action
FERC opens show cause proceeding questioning whether PJM's rules for data center colocation with power plants are just, reasonable, and non-discriminatory.
Chris Wright Confirmed as Energy Secretary
Appointment
Senate confirms former Liberty Energy CEO Chris Wright as Secretary of Energy with a 59-38 vote, putting a fossil fuel industry executive in charge of AI infrastructure policy.
Trump Announces $500B Stargate Project
Policy Announcement
President Trump unveils Stargate, a joint venture between OpenAI, SoftBank, and Oracle promising $500 billion in AI infrastructure investment over four years, with $100 billion to deploy immediately.
Microsoft-Constellation Three Mile Island Deal
Corporate Agreement
Microsoft announces agreement with Constellation Energy to restart the Three Mile Island nuclear reactor to power AI data centers, signaling tech industry's willingness to pursue unconventional energy sources.
PJM Capacity Auction Shocks Market
Market Event
PJM's capacity auction clearing price increases 833% from the previous year, with data centers responsible for an estimated 63% of the $9.3 billion price increase.
Scenarios
1
Tech Giants Accept Cost Burden, Bills Stabilize
Discussed by: Energy Department officials, Microsoft corporate communications, industry analysts at S&P Global
Major tech companies follow Microsoft's lead and negotiate utility rate structures nationwide requiring them to prepay for infrastructure and bear capacity costs. Combined with FERC's accelerated interconnection rules and new nuclear/natural gas capacity, electricity price growth moderates. Trump claims victory heading into midterms, and the voluntary framework becomes the de facto industry standard without need for federal legislation.
2
Voluntary Commitments Prove Unenforceable, States Regulate
Corporate commitments lack binding mechanisms and monitoring. As data center construction accelerates—$600B+ in hyperscaler capex projected for 2026 alone—regional electricity prices continue climbing. States like Virginia, Maryland, and Ohio pass legislation mandating data center cost allocation, creating a patchwork of regulations. Federal-state jurisdiction battles reach the courts, delaying resolution while consumers keep paying.
Data center demand outpaces generation and transmission capacity. By summer 2026, PJM falls below reliability standards, triggering rolling blackouts in Virginia and Maryland. Congress passes emergency legislation giving FERC expanded authority over large load interconnections, with mandatory cost-allocation requirements. Tech companies face choice between accepting strict utility-style regulation or relocating infrastructure overseas.
4
AI Demand Plateaus, Pressure Eases
Discussed by: Goldman Sachs analysts, Bloomberg Intelligence, tech industry skeptics
The AI infrastructure buildout slows as companies like Stargate struggle to raise capital and enterprise AI adoption disappoints. Data center construction decelerates, easing grid capacity pressure. Electricity price increases moderate naturally, and the political urgency fades. Voluntary commitments remain in place but become largely symbolic as underlying demand growth subsides.
Historical Context
California Electricity Crisis (2000-2001)
June 2000 - September 2001
What Happened
California's deregulated electricity market collapsed when Enron and other traders manipulated supply, creating artificial shortages. Wholesale prices spiked from $45 to over $1,400 per megawatt-hour. Rolling blackouts hit millions. Governor Gray Davis declared a state of emergency in January 2001.
Outcome
Short Term
The state spent $27 billion on wholesale electricity in 2000 alone, up from $7.4 billion in 1999. Pacific Gas & Electric declared bankruptcy. Davis's political standing collapsed.
Long Term
California recovered $7 billion through litigation. Federal regulators gained authority to impose price caps. The crisis demonstrated that market failures in electricity can rapidly become political crises.
Why It's Relevant Today
Today's data center energy conflict shares key features: concentrated demand growth outpacing supply, cost increases hitting consumers, and questions about whether market mechanisms or government intervention should determine who pays. The California crisis showed how quickly electricity affordability can become a political liability.
AT&T Kingsbury Commitment (1913)
December 1913
What Happened
Facing federal antitrust action over its telephone network monopoly, AT&T voluntarily agreed to divest Western Union stock, allow competitors to interconnect, and limit acquisitions. The 'Kingsbury Commitment' was negotiated directly with the Attorney General, not mandated by law or court order.
Outcome
Short Term
AT&T avoided breakup and maintained its network dominance. Interconnection requirements created the appearance of competition while cementing AT&T's control of core infrastructure.
Long Term
The voluntary framework lasted decades but ultimately proved insufficient. AT&T was eventually broken up in 1984 through antitrust action, demonstrating the limits of voluntary industry commitments when underlying market power remains concentrated.
Why It's Relevant Today
Trump's tech company commitments follow the Kingsbury model: voluntary corporate pledges negotiated under threat of political or regulatory action. The historical precedent suggests such arrangements can delay but not resolve fundamental tensions between concentrated industry power and public interest concerns.
AI Safety Voluntary Commitments (2023)
July 2023
What Happened
Seven leading AI companies—Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI—announced voluntary commitments at the White House to manage AI risks. Commitments included security testing, watermarking AI-generated content, and sharing safety research.
Outcome
Short Term
Companies formed the Frontier Model Forum nonprofit. The commitments generated positive press coverage and provided political cover against calls for binding regulation.
Long Term
MIT Technology Review's one-year review found limited concrete changes. Watermarking commitments went largely unfulfilled. No enforcement mechanism exists. The episode illustrated the gap between voluntary pledges and operational changes.
Why It's Relevant Today
The AI safety commitments provide the most recent template for Trump's data center cost approach: voluntary pledges from tech companies under political pressure, announced with White House fanfare, lacking binding enforcement. The parallel suggests skepticism about implementation is warranted.
