Sandia National Laboratories is field-testing an artificial intelligence-driven grid voltage control system at two sites in Texas, Energies Media reported, framing the deployment against rising electricity demand from data centers.

The reported activity extends a U.S. Department of Energy national security lab into the small but growing cohort of research organizations exploring machine learning for distribution-level voltage regulation.

Sandia, headquartered in Albuquerque, runs grid resilience and power systems research. Field tests at two Texas locations would move that work from simulation onto live distribution assets.

What the source covers

The Energies Media item identifies Sandia as the operator, names AI-driven voltage control as the technology, and specifies two Texas sites. It does not name host utilities, feeder types, control architecture, or measured results. Performance specifications have not been published in the materials cited.

Why data center load sits behind the test

Voltage stability on feeders serving large single loads is a recognized distribution-engineering concern. Conventional load tap changers, line voltage regulators, and switched capacitor banks operate on local measurements and time delays calibrated for the original feeder.

A large step change in demand from a new data center substation can cycle regulators and push them outside their tap range. AI-driven control, in principle, draws on phasor measurements, weather, and historical load patterns to anticipate cycling and shift setpoints earlier.

How this slots into current portfolios

Volt-var optimization is already a commercial category, with offerings from established grid-software vendors and a small set of startups. The open question the Sandia tests — if performance data is eventually published — would help answer: can a research deployment match or beat vendor performance on the same feeders?

A national-lab result, if reproducible, would create a credible benchmark for utility R&D groups weighing procurement against incumbent vendor offerings.

What the report does not yet support

Without test duration, funding source, vendor relationships, sample size, or any performance metric, the Energies Media item functions as a market signal. Researchers and utility R&D leaders tracking AI volt-var control will want peer-reviewed results, a Sandia technical report, or a public dataset before treating the deployment as more than a proof-of-concept milestone.

The form of publication Sandia ultimately chooses — a technical report, a journal paper, or a partner handoff — will determine how much the two-site test shifts procurement conversations in AI-driven voltage control over the remainder of the decade.