Texas Governor orders statewide audit of AI data center projects and pauses all new grid connections pending completion. Audit applies to all data centers seeking ERCOT grid access.

Texas Gov. Greg Abbott ordered a statewide audit of data center projects in ERCOT's interconnection queue on Monday, tightening oversight of AI infrastructure as proposed electric demand continues to accelerate. In a directive to the Public Utility Commission of Texas (PUCT) and the Electric Reliability Council of Texas (ERCOT), Abbott mandated a "comprehensive verification and audit" of every data center interconnection request. Any project failing to comply with state law or regulatory requirements must be denied connection to the Texas grid.

"Our top priority is to protect Texans' safety and quality of life," Abbott said. "Any project that fails to comply with the requirements set forth by the PUCT and ERCOT, and by state law, must be denied connection to the Texas grid. Simply put, Texans must come first."

ERCOT said it will work with PUCT to implement the directive, including postponement of the Batch Zero transmission planning study. PUCT Chairman Thomas Gleeson endorsed the order, stating the commission would continue working with ERCOT to ensure the requirements are "fully carried out on behalf of the people of Texas."

For developers, the consequences are immediate. ERCOT will postpone its Batch Zero transmission planning study while executing the governor's directive. More broadly, the order signals that securing a place in the interconnection queue is no longer sufficient. As AI campuses grow into gigawatt-scale loads, regulators are increasingly asking whether proposed projects are financially committed, physically executable, and likely to become long-term customers before planning transmission and generation around them.

Texas, which has become a hotspot for hyperscale AI infrastructure representing hundreds of gigawatts of proposed new electric demand far exceeding the state's current peak load, is the latest—and arguably most aggressive—example of this regulatory shift. State leaders have expressed growing concern that speculative projects could distort long-term grid planning and drive unnecessary infrastructure investment.

Energy analyst Neil Osnato of Persistence Analytics Group argues the directive's implications extend beyond simply cleaning up an oversized queue. "Texas is redefining what counts as credible demand," he said. The state is moving away from a conventional interconnection process toward an evidence-based approach requiring developers to demonstrate project maturity before the grid commits planning resources. That means distinguishing between those that have merely requested service and those with financing, site control, permitting, equipment, construction plans, power and water availability, and a realistic path to long-term operation. "A place in the queue is not proof that the grid should build around the request," Osnato said.

Abbott's directive raises the stakes by making broader verification a prerequisite for continued advancement through the interconnection process. For developers, this could translate into a higher evidentiary burden—stronger proof that projects are real, financeable, and likely to reach commercial operation before grid planners commit resources.

Texas is not alone in reassessing how to evaluate massive new AI loads. PJM has approved reforms separating physical interconnection from firm resource adequacy for large new loads while tightening viability requirements elsewhere. In Virginia, regulators are debating whether hyperscale customers should bear greater transmission costs. According to Osnato, these represent different responses to the same challenge: "Large-load planning can no longer rely on forecast magnitude alone. It must test probability, timing, deliverability, durability and cost responsibility."

Abbott's directive does not specify the audit's timeline or exactly how it will affect projects already moving through interconnection. It also leaves unanswered how PUCT and ERCOT will divide responsibility and what additional information developers may need to provide. Those details will determine whether the initiative becomes a targeted effort validating the state's largest proposed loads or a broader procedural hurdle affecting AI development timelines.