Texas has become one of the clearest places to watch the collision between AI infrastructure and the power grid.
In its August Short-Term Energy Outlook, the EIA cut its forecast for Texas electricity load growth in 2027.
The prior forecast had Texas load growing 14% in 2027.
The August forecast brought that down to 6%.
The reason was not weaker AI demand. It was the policy and grid response to that demand.
In early August, Texas ordered a comprehensive data center audit and paused new data center development approvals while the state reviews grid reliability, interconnection requests, water usage, and emergency preparedness.
The scale of the request queue explains why regulators are paying attention.
According to the Texas governor’s office, ERCOT had 474 gigawatts of grid-connection requests, and roughly 90% came from data centers.
That is an enormous number relative to the state’s power system.
For investors, the chart is useful because it shows AI demand moving beyond earnings calls and construction spending into grid planning.
Data centers need power. They need transmission. They need backup systems, cooling, land, permits, and interconnections. When too many projects line up at once, the constraint may shift from chips to electricity.
Texas is an important test case because it has cheap land, large power markets, fast population growth, and a huge pipeline of industrial and data center demand. It is also a state where grid reliability is already a political issue.
The research question is which companies benefit when the bottleneck moves to power.
Utilities, grid equipment suppliers, power developers, engineering firms, backup-power providers, cooling companies, natural gas infrastructure, batteries, and nuclear developers all sit near this trend.
The risk side is just as important. Some data center projects may get delayed, repriced, resized, or forced into more expensive power arrangements.
AI demand is real. The grid still has to absorb it.