U.S. Data Center Power Demand Set to Triple by 2035 as AI Expansion Strains Electrical Grid

Reviewed byNidhi Govil

3 Sources

Share

BloombergNEF forecasts data center electricity demand will reach 106 gigawatts by 2035, driven by AI expansion and massive new facilities. Despite record grid spending, transmission bottlenecks and reliability risks are mounting across major U.S. power markets.

Explosive Growth in Data Center Power Demand

The United States is experiencing an unprecedented surge in data center electricity consumption, with new forecasts showing demand will triple by 2035. BloombergNEF's latest analysis projects data center power requirements will reach 106 gigawatts by 2035, up from the current 40 gigawatts

1

2

. This represents a dramatic 36% upward revision from the firm's April outlook, illustrating the rapid pace of sector expansion driven primarily by artificial intelligence applications.

Source: Axios

Source: Axios

To put this growth in perspective, one gigawatt can power approximately 750,000 to 1 million homes

1

. The surge reflects both the increasing number of facilities and their dramatically larger scale, with AI workloads requiring massive compute density and consistent power availability.

Source: Benzinga

Source: Benzinga

Shift Toward Massive Rural Facilities

The data center landscape is undergoing a fundamental transformation in both location and scale. Development is moving away from urban areas as facility sizes grow exponentially

1

. While today's U.S. data centers are typically located in suburban areas within 30 miles of major cities, new construction is increasingly targeting rural locations to accommodate larger builds away from densely populated zones

2

.

The scale difference is striking: currently, only 10% of existing data centers exceed 50 megawatts of capacity, yet most facilities in development are north of 100 megawatts

1

. Nearly 25% of new facilities will exceed 500 megawatts, with some gigawatt-scale sites coming online in the coming years

2

3

.

Regional Concentration and Grid Strain

New capacity planning is concentrating in specific U.S. regions, particularly within the PJM Interconnection area covering Virginia, Pennsylvania, Ohio, Illinois, and New Jersey, as well as Texas's ERCOT grid

2

. These regions are attracting development due to available land and energy resources, but the concentration is creating significant challenges.

PJM, which operates the grid across 13 states and supplies nearly 65 million people, could see data center capacity reach 31 gigawatts by 2030

3

. This nearly matches the 28.7 gigawatts of new power generation expected in the same period, highlighting the strain on the system.

Massive Investment but Persistent Bottlenecks

The boom is driving unprecedented investment levels. Global investment in data centers reached $580 billion this year, exceeding worldwide expenditures on developing new oil supplies

2

. Major tech companies are leading this charge, with Barclays estimating that Meta, Google, Amazon, Microsoft, and Oracle will have roughly $390 billion in combined capital expenditure this year, representing a 71% year-over-year increase

1

.

Despite record grid spending of $115 billion in the U.S. alone as part of $470 billion globally, transmission bottlenecks persist

3

. Supply-chain constraints, permitting delays, and labor shortages continue to slow transmission expansion, creating a mismatch between demand growth and infrastructure capacity.

Today's Top Stories

TheOutpost.ai

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

© 2025 Triveous Technologies Private Limited
Instagram logo
LinkedIn logo