Texas built much of its modern economic identity around abundance: abundant land, abundant energy and a political culture designed to attract capital. Artificial intelligence is now testing how far that model can stretch.
Governor Greg Abbott’s decision to halt new state permits for data centers until regulators complete detailed audits of their electricity and water demands signals a shift from simply attracting AI infrastructure to determining who pays for it.
Abbott directed the Texas Commission on Environmental Quality to stop issuing permits sought by data center projects until the Electric Reliability Council of Texas (ERCOT), the Public Utility Commission and the Texas Water Development Board provide the necessary information.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
The review will examine electricity consumption, water use, grid reliability, infrastructure costs and the effects on surrounding communities. The scale of the underlying challenge is substantial.
ERCOT has been considering more than 474 gigawatts of requests to connect to the Texas grid, according to Abbott’s office, with roughly 90% of those requests associated with data centers. That proposed load is more than five times Texas’ record peak electricity demand.
The numbers explain why the economics of AI infrastructure are becoming inseparable from energy policy. A data center may represent billions of dollars of investment, construction activity and technological capacity.
But it creates a persistent demand for electricity and, depending on its cooling technology, significant water requirements. The central question is therefore moving beyond whether Texas wants AI investment. It is increasingly about how that investment integrates with the physical infrastructure already serving households and businesses.
Abbott has made one principle particularly explicit: data centers should pay their own infrastructure costs rather than shifting them onto residential electricity customers. His June directive instructed regulators to require data centers to fund the electric infrastructure necessary to serve their operations.
ERCOT is expected to complete its broader audit by December. The review is intended to establish a more detailed picture of projects seeking grid connections, including their expected electricity consumption, water sources, cooling systems, onsite generation plans and public incentives.
The politics surrounding the issue are becoming equally important.
President Donald Trump has promoted rapid AI development as part of the United States’ competition with China, framing technological leadership as a strategic priority. Yet Texas is now imposing additional scrutiny on precisely the infrastructure required to expand that AI capacity.
That tension illustrates a broader contradiction in the AI economy. National policymakers can view data centers as strategic infrastructure, while residents experience them through more immediate questions: electricity bills, water availability, noise, land use and whether promised economic benefits justify the costs.
Public opinion adds another layer. Recent polling has found substantial opposition to new AI data centers, with concerns extending across partisan lines. An AP-NORC/University of Chicago survey, for example, found that about 60% of Americans supported limiting the number of new data centers, while concerns about electricity and water consumption were widespread.
For Texas, the permitting freeze is therefore less about abandoning AI than redefining the terms of expansion. The state remains one of America’s most important destinations for technology investment, but the latest policy makes clear that access to Texas’ power and water cannot be treated as unlimited inputs.
The next phase of the AI boom will consequently be measured not only in chips, models and valuations, but also in megawatts, gallons and transmission lines. Texas is forcing that accounting into the center of the AI debate—and the outcome could influence how America builds the physical infrastructure behind its artificial-intelligence ambitions.