> ## Content Index > Fetch the complete content index at: https://nexi.fund/llms.txt > Use this file to discover other available public pages before exploring further. # AI Grid Voltage Defence — Sandia DERMS and the National Security Case for Distributed Energy Control - URL: https://nexi.fund/sandia-ai-grid-voltage-defence/ - Published: 2026-07-20T06:00:18.000Z - Updated: 2026-07-20T06:00:18.000Z - Description: Sandia's AI-driven DERMS stabilizes grid voltage in real time as data center demand surges. The same system is being positioned to protect military installations from grid instability. - Author: Nexi.fund Labs - Tags: AI & Infrastructure, #mode-4, #hook-paradox, #track-F, #brand-heavy The same technology driving the grid to the edge of instability is now being deployed to pull it back. 🎯 AI data center power demand is growing at 40% annually, pushing local distribution grids into voltage swings that mechanical equipment can no longer handle fast enough. Sandia National Laboratories has field-tested an AI-driven Distributed Energy Resource Management System (DERMS) that coordinates grid-connected devices in real time and demonstrated measurable voltage improvements at two Texas sites. The same system is being positioned as a solution for military installations where energy resilience is a national security requirement, not just an operational preference. Data center electricity demand is on track to reach 945 terawatt-hours by 2030 — roughly Japan's entire annual consumption. Each new AI training cluster adds hundreds of megawatts of load that can swing by 100 MW or more in minutes as GPU jobs start and stop. Local distribution grids, designed for gradual, predictable load growth, were not built for this. The result is voltage instability. When demand surges faster than mechanical capacitor banks and line regulators can respond, voltage sags propagate through the distribution network. For data center operators, a 5% voltage drop can trigger cascading server failures. For military installations sharing the same grid, the consequences are strategic. ## The Limits of Mechanical Regulation Utility voltage management has relied on the same physical principle for decades: capacitor banks that switch on and off, line regulators that tap up and down, and substation transformers with fixed setpoints. These devices work on timescales of seconds to minutes, which was acceptable when load changed gradually over hours. Data center power demand changes nothing like that. An AI training run can draw 700 MW one moment and 400 MW the next as batch jobs complete. The grid sees this as a voltage transient that mechanical equipment cannot track. Sandia engineer Rachid Darbali-Zamora puts it directly: "The way we generate electricity and the loads being placed on the grid are evolving, but the backbone of the grid that connects these is staying the same." Upgrading every distribution feeder with new capacitor banks and voltage regulators would cost billions and take years in permitting alone. It is not a realistic solution at the scale required. ## The Case for AI-Driven Coordination The Sandia team's approach exploits hardware already in the ground: smart inverters. Millions of inverters connect solar panels, batteries, and backup generators to the grid. They are already deployed, already communicating. What they lack is a central intelligence to coordinate their actions. The DERMS platform fills that gap. It forecasts changes in electricity demand and available generation, then issues coordinated setpoints to every connected inverter in near real time. Instead of each device reacting independently to local voltage readings, the entire distributed fleet operates as a single responsive system. Field tests at two sites in Lubbock, Texas, the SWiFT wind farm facility and the Texas Tech GLEAMM microgrid, produced the first real-world validation. At GLEAMM, which includes an active data center, voltage was running roughly 5% above the utility target. With DERMS coordinating inverter actions, voltage moved measurably closer to the optimal value. The team ran a full day with the controller active and a full day without. "When you compare the voltage graphs from those two," Darbali-Zamora said, "you can visually see that the voltage in the system is improved." 945 TWh projected data center demand by 2030 ↑ 2× vs 2025 #### Data Center Electricity Demand AI training clusters are the fastest-growing source of new grid load. IEA projects demand will equal Japan's entire electricity consumption by decade-end. · *IEA, 2026* 5% voltage improvement at GLEAMM nearer to utility target #### DERMS Voltage Regulation The lab's side-by-side field test showed the AI controller brought voltage measurably closer to optimal levels compared to the uncontrolled baseline. · *Sandia National Laboratories, Jul 2026* ## The National Security Dimension The DERMS work was funded by the Department of Energy's Office of Electricity, but the project's framing has always included a parallel mandate. Senior manager Charles Hanley is explicit: "In scenarios of national conflict or war, adversaries will target energy infrastructure to disrupt both military and civil functions. The Sandia-developed DERMS system helps to defeat such adversarial attacks through the application of agile and secure technologies that adapt to real-time developments and keep critical systems operating." Military installations in the United States and abroad share grid connections with surrounding civilian populations. When voltage drops on the shared feeder, missile defense radars, command centers, and secure communications platforms experience the same instability as the data center next door. The FY2026 defense budget already allocates $63 million for a new microgrid at Naval Base Guam, where power quality failures could compromise submarine support facilities and strategic communications terminals. The distinction between grid reliability and national security is eroding as fast as data center load is growing. The same infrastructure serves both purposes. An AI controller that stabilizes voltage for a data center is, with no hardware changes, the same controller that keeps a military installation's sensitive loads within operating tolerances. | Approach | Traditional | AI DERMS | | ------------------------- | --------------------------------------- | -------------------------------------------------- | | **Response time** | Seconds to minutes | Near real-time (ms) | | **Hardware required** | New capacitor banks, line regulators | Existing smart inverters | | **Cost profile** | High CAPEX, years of permitting | Software + coordination (Energy I-Corps) | | **Adaptability** | Fixed setpoints, manual reconfiguration | AI retrains as grid conditions evolve | | **Defence applicability** | Limited to base perimeter | Coordinates all connected devices on shared feeder | Comparison of traditional voltage regulation vs Sandia's AI-driven DERMS approach · *Sandia National Laboratories, 2026* The technology is moving toward commercialization. Its Energy I-Corps program is working with utilities and microgrid operators to refine the operational requirements for broader deployment. The platform has advanced to Phase III, and the lab is actively seeking corporate partners. The team has also joined the Open Power AI Consortium, led by the Electric Power Research Institute, to scale AI-for-grid tools across the industry. Challenges remain. The DERMS approach requires a minimum density of smart inverters on a given feeder to be effective. Neighborhoods still running on dumb hardware gain nothing from software alone. Utilities must also contend with cybersecurity risks: an AI controller that can coordinate thousands of devices is, in the wrong hands, an attack surface of equivalent scale. The lab's earlier work on autoencoder-based intrusion detection for grid systems, published in 2025, directly addresses this vector, but the security posture of every connected inverter is only as strong as the weakest firmware version still in the field. The irony is not subtle. AI training clusters created the voltage instability problem. AI inference at the grid edge is now being proposed as the solution. The question is whether the deployment timeline for DERMS can keep pace with the load growth from the very industry that made it necessary. [ As data-center demand grows, Sandia advances AI controls to keep voltage steady in real time Sandia National Laboratories' DERMS platform uses AI to coordinate grid-connected devices for real-time voltage regulation, with field tests at two Texas sites. Sandia National Laboratories ](https://newsreleases.sandia.gov/as-data-center-demand-grows-sandia-advances-ai-controls-to-keep-voltage-steady-in-real-time/?ref=nexi.fund) Primary source: official Sandia news release with full technical details, field test data, and senior manager quotes on the national security application. [ AI Tool Could Keep Military Installations Protected From Grid Instability Coverage of Sandia's DERMS from a defense-sector perspective, emphasizing the direct military application of AI-driven grid voltage controls. NextGen Defense ](https://nextgendefense.com/ai-tool-military-systems/?ref=nexi.fund) Defence-focused analysis connecting Sandia's technology to military installation resilience requirements. [ Sandia field-tests AI-driven voltage control system at two Texas grid sites Detailed coverage of the SWiFT and GLEAMM field demonstrations with specific voltage improvement data and PHIL testing methodology. Energies Media ](https://energiesmedia.com/sandia-ai-voltage-control-texas-energies?ref=nexi.fund) Energy-sector journalism with practical detail on the field deployment methodology and results.