Texas approved massive AI data center adjacent to wind farm with new grid integration rules buried in deal.

The Public Utility Commission of Texas approved a net metering arrangement for a 260-megawatt artificial intelligence data center co-located with a wind farm of approximately 265.5 megawatts. This pairing makes the site one of the most closely evaluated power projects in the nation. Developer Crusoe teamed up with Ensign, the large-load customer operating the compute facility.

This represents the second phase of development at the site. Combined, the two compute developments will demand roughly 525 megawatts—an amount exceeding the generation output of the wind farm. That physical imbalance transformed what might have been a routine net metering application into a high-stakes regulatory precedent.

The case serves as one of the initial operational tests of recent state statutory frameworks authorizing grid operator ERCOT to enforce load shedding on energy-intensive facilities during grid stress. How regulators resolved this application signals how Texas will govern colocation arrangements moving forward.

Crusoe and Ensign argued that because the aggregate load exceeds wind output, the secondary facility should operate under distinct curtailment parameters. Regulators rejected that argument.

Commissioners adopted strict reliability protocols, incorporating recommendations from Chairman Thomas Gleeson. Under the ruling, the facility must maintain the capability to shed its full load within thirty minutes during grid emergencies through direct physical breaker disconnection rather than software-managed power reductions. The regulatory order further prohibits the project from receiving financial compensation through grid service demand response programs for this capability. Regulators classified emergency load shedding as a standard baseline requirement for grid access rather than a reimbursable service.

The core reasoning remains straightforward: allowing a secondary data center to remain operational while primary systems shed load would undermine the grid operator's ability to maintain reserve capacity during regional supply shortages.

While the ruling establishes firm operational obligations, it offers procedural timelines. Regulators instructed grid managers to provide sixty minutes of advance notice when feasible, while granting developers the option to implement ten-minute rapid disconnect systems. However, the requirement to drop the full load on demand remains binding.

By establishing these parameters, state regulators signaled that future co-located compute projects cannot rely on unbacked grid power during supply constraints. Developers building similar behind-the-meter facilities must integrate complete operational redundancy from the start.

Energy analysts noted that the decision provides crucial clarity rather than halting development. Grid Tracker co-founder Chris Talley observed that the framework does not eliminate co-location models within the Texas market. Instead, the decision establishes that off-grid architectures require dedicated auxiliary infrastructure capable of sustaining operations through sustained grid disconnections. Market experts caution that maintaining full compute operations independently during extended grid emergencies represents a significant technical requirement that will require proven execution at scale.

Power generation associations and independent producers filed comments expressing concern that mandatory curtailment terms could raise capital costs for future projects. While procedural questions remain regarding how future petitions will address these constraints, the baseline standards for co-located operations are now firmly established.

Upcoming decisions will test how regulators handle baseload generation. A pending proposal involves co-locating a data center complex alongside the Comanche Peak nuclear power plant near Fort Worth. Industry observers view that proceeding as the definitive test for co-location economics. Nuclear facilities generate constant baseload power, creating a different grid dynamic than variable wind generation. Whether regulators apply identical curtailment mandates to constant nuclear power or offer alternative provisions based on grid stability will test state energy policy even further.

Grid operators nationwide face similar surges in compute energy demand. Decisions made by Texas regulators establish an influential model for how regional power grids will balance technological expansion with grid reliability.