Archive · 14 Sep 2026
US AI Boom Requires $110 Billion for New Power Plants
Artificial intelligence and cloud computing are causing many more data centers to be built in the United States.
These data centers need a lot of electricity.
Moody’s Ratings estimates that the country may need 45 gigawatts of new power generation by 2030.
Building those power plants could cost about $110 billion.
Most of the new electricity would come from natural gas, with some coming from solar power and batteries.
Nuclear restarts would provide less than 5% of the new supply.
The projects could increase electricity costs by $25 billion to $30 billion each year.
Data centers may pay for some power plants themselves, but regulators will decide how many costs are passed to other customers.
Some people support the expansion, while others worry about higher bills, pollution, water shortages, and land impacts.
Moody’s Ratings estimates the US will need $110 billion for 45 gigawatts of new power generation by 2030.
More than 30 gigawatts is expected to come from natural gas, requiring about 4 billion cubic feet of additional gas supply.
Solar and energy storage will provide most of the remaining capacity, while nuclear restarts will contribute less than 5%.
Data centers could consume 426 terawatt-hours in 2030, doubling their share of US electricity use to 10%.
The new plants could add $25 billion to $30 billion annually to US electricity costs, while data centers may directly pay up to $15 billion.
- Who
- Moody’s Ratings, US utilities and regulators, and companies developing AI and cloud-computing data centers.
- What
- A forecast that the United States will need $110 billion to build 45 gigawatts of new power generation for data centers by 2030.
- When
- The forecast covers development through 2030; the report was published September 14, 2026.
- Where
- The United States.
- Why
- AI and cloud-computing data centers are driving a major increase in electricity demand.
AI Expansion and Grid Investment
Affordability and Environmental Concerns
Need for new generation
AI Expansion and Grid Investment
Supporters of the buildout argue that new power plants are needed to supply electricity for the expanding AI and cloud-computing industries.
Affordability and Environmental Concerns
Critics worry that households and businesses could help subsidize power infrastructure built primarily for data centers.
Cost allocation
AI Expansion and Grid Investment
Data centers may directly fund up to $15 billion of the buildout through power plants located on their campuses.
Affordability and Environmental Concerns
The remaining costs could increase electricity bills, depending on local rate-making and cost-allocation rules.
Energy and land impacts
AI Expansion and Grid Investment
Natural-gas plants can provide electricity on demand, unlike renewable sources that depend on sunlight and wind conditions.
Affordability and Environmental Concerns
Opponents cite concerns about pollution, water shortages, and threats to public lands and private property.
- Estimated investment
- $110 billion for new power plants through 2030
- Required capacity
- 45 gigawatts of new power generation
- Natural gas share
- More than 30 gigawatts, requiring about 4 billion cubic feet of incremental gas supply
- Projected data-center demand
- 426 terawatt-hours in 2030
- US electricity share
- Data centers are projected to account for 10% of US electricity use in 2030
- Potential annual cost increase
- $25 billion to $30 billion in US electricity costs
- Direct data-center payments
- Up to $15 billion for behind-the-meter power projects