GridCARE's AI finds ~20 MW of spare capacity on an SCE Inland Empire circuit

GridCARE said Friday it is working with Southern California Edison to use physics-based AI that identified about 20 megawatts of additional flexible capacity on a constrained Inland Empire circuit, without waiting for a full infrastructure buildout.

AI data centers are slamming into interconnection queues. Here the AI is pointed at the grid itself — finding megawatts on a live California utility circuit so customers can plug in years sooner.

On Friday, 2 October 2026, GridCARE said it is working with Southern California Edison to look for extra room on parts of the grid that are already tight. The Business Wire release is titled “GridCARE Power Acceleration™ to Help California Utility Support Faster Customer Energization.” The line under the headline says the technology identifies additional capacity for constrained grid infrastructure in Southern California. Open copies of the release, on FinancialContent and on Wedbush’s investor page, stamp it October 2, 2026, at 8:00 a.m. Eastern, and both say the item is via Business Wire. The Wedbush copy prints the Business Wire source URL. A megawatt, shortened to MW, is a million watts of electrical power. Twenty megawatts is the size of the extra capacity the study names. A circuit, here, is one section of the local grid, not the whole utility. Energization is the moment a new customer can actually take power. Those lines are the release’s.

What the two companies say they are doing. GridCARE said it is working with Southern California Edison, shortened to SCE, to explore physics-based AI that can identify additional capacity for new customer connections on parts of the grid that would otherwise be constrained. Constrained means that section is already near the limit of what it can carry. Physics-based, in the release’s phrase, means the software searches the grid using the physics of the power system, rather than only guessing from past examples. Power Acceleration is GridCARE’s name for that technology. The release says it may help bring customers online sooner, keep reliability, and improve affordability by getting more capacity from equipment that is already built. “May help” and “explore” are the release’s words. They describe a study and a look at more sites. They are not a statement that the tool is already running across SCE’s whole system. Those lines are the wire’s.

The first circuit, and the number on it. In an initial study, SCE selected a constrained circuit in the Inland Empire, the region east of Los Angeles. Working with GridCARE, the analysis identified a potential of 20 megawatts of additional flexible capacity that could safely support future customer energization while meeting reliability requirements. Flexible capacity, in that sentence, means room to connect customers if some of the new demand can be turned down when the circuit needs it. It is not a new power plant, and it is not power the release says is already flowing to a named customer. A potential of 20 megawatts is the study’s finding, as the release states it. The tilde in the headline is “about.” The release prints 20. It does not print a range, and it does not name the circuit beyond Inland Empire. Those lines are the wire’s.

What SCE’s planning chief said the utility wants. Shinjini Menon, senior vice president of System Planning and Engineering for SCE, said the utility is looking for innovative solutions that can get more out of the existing grid, unlocking capacity for customers without relying on costly and time-consuming infrastructure expansion. Infrastructure expansion, in that sentence, means building new lines, substations, and other equipment. She said that, combined with demand flexibility, the approach has the potential to accelerate meeting new electricity demand and to make the grid more affordable for customers. Demand flexibility means some customers can shift or reduce what they use when the grid is tight. “Has the potential” is her wording. The release does not put a dollar figure on the expansion SCE would otherwise build, and it does not put a date on when the extra 20 megawatts would be in use. That quotation is hers, in the release.

Where else SCE says it is looking. The release says SCE is exploring Power Acceleration at several other locations on its system, to energize customers in electric-vehicle charging, manufacturing, commercial work, and other uses that can benefit the local economy. Exploring means those sites are under consideration. Several is the release’s count. It does not print the number of locations, the cities, or a date for a wider rollout. Those lines are the wire’s.

What GridCARE’s chief executive said the work is for. Amit Narayan, founder and chief executive, said public support for AI will depend on whether communities share in its benefits. “The state that leads in AI will also lead in using AI to make the grid work for everyone,” he said. He said SCE is showing that growth and affordability can go hand in hand, and that every success makes the case for action stronger and the case for delay harder to defend. That quotation is his, in the release. The release does not attach a contract price to it.

The national estimate the release cites, kept as that citation. Under a heading that flexibility has to be qualified before it counts as capacity, the release says Duke University research estimates that existing U.S. power systems could accommodate nearly 100 gigawatts of new load at an annual curtailment rate of just 0.5 percent. A gigawatt is a thousand megawatts, so nearly 100 gigawatts is nearly 100,000 megawatts. Twenty megawatts on one circuit is a sliver of that national figure. The comparison is arithmetic on the two numbers. The release prints 20 megawatts and nearly 100 gigawatts. It does not print the sliver. Curtailment, here, means the new load would be asked to turn down for a small share of the year. Half of one percent of the hours in a year is about 44 hours. That 44 is arithmetic on 0.5 percent. The release prints 0.5 percent. It does not print 44 hours. The release says realizing that potential requires proof, at each circuit and each substation, that the safety margins for reliable operation stay intact. A substation is the yard of equipment where voltage changes and lines meet. A safety margin is the gap between how hard the equipment is working and the point where it fails. Those lines are the release’s account of Duke’s research. They are not a result the release assigns to the Inland Empire circuit, and this filing does not re-check the Duke paper.

How the release says the software fits into that proof. Across a large grid, the release says, the combinations of locations, hourly demand, weather, equipment outages, and flexibility resources produce quadrillions of scenarios to analyze. A quadrillion is a million billion. The word is the release’s, for an enormous set of cases. GridCARE’s physics-based AI accelerates the search for workable solutions. Utility engineers then test those solutions in their own planning tools to qualify them. Qualify, here, means the utility’s own engineers decide the option is safe to treat as real capacity. The release describes a search, then a check by the utility. It does not say the software’s answer is the last word, and it does not say SCE has finished that check at the other locations. Those lines are the wire’s.

What the about box claims, kept as the company’s account of itself. GridCARE says it delivers Power Acceleration for AI infrastructure and for other critical industries. It says the technology finds capacity on existing grids and speeds energization, so utilities can take on growth while protecting reliability and affordability. The box says the company was founded at Stanford’s Doerr School and is supported by AI and energy investors. It says GridCARE has accelerated more than 1 gigawatt of grid capacity. One gigawatt is 1,000 megawatts, fifty times the 20 megawatts in the Inland Empire study. That fifty is arithmetic on the company’s 1 gigawatt and the study’s 20 megawatts. The release does not say the 1 gigawatt includes this SCE circuit. The box also says that by moving energization schedules forward by years, the platform has enabled economic value exceeding $10 billion for data-center developers. More than 1 gigawatt, the years, and the $10 billion are the company’s claims about its own platform. The release does not present an outside audit of those figures. It does not print an SCE contract value. It does not print a statewide rollout date. It does not say Power Acceleration is already in production across SCE’s system. Media contact on the release is pr@gridcare.ai.

What the company site says, kept separate from Friday’s study. The homepage at gridcare.ai describes Power Acceleration as a way to bring AI factories online in months rather than years, and it names an Energize platform with tools it calls Power Finder, Power Activation, and Power Operations. “Months, not years” is the site’s line. It is not a clock the Friday release prints for the Inland Empire circuit. The homepage, as read the same day, does not restate the SCE study or the 20 megawatt figure. Other quotations and capacity figures on that site are earlier partnerships and product claims. They are not this announcement.

The picture is a split card. The left side is GridCARE’s Power Acceleration for AI art: white type on a dark blue field, the word AI in a lighter blue, the GridCARE mark at the lower left, and thin white rings. A teal stripe divides the frame. The right side is a dark panel. Small teal type reads SOFTWARE · ENERGY. White type reads GridCARE × SCE and Power Acceleration. Gray type says physics-based AI finds about 20 MW of flexible capacity on a constrained Inland Empire circuit. A teal button reads SOUTHERN CALIFORNIA EDISON. A line along the bottom reads gridcare.ai · business wire. The frame does not print a calendar date. It is a summary card of the release. It is not a photograph of a power line, a substation, or the Inland Empire circuit.

In plain terms, GridCARE said on Friday that it is working with Southern California Edison on physics-based AI, called Power Acceleration, that looks for extra capacity on parts of the grid that are already tight. On one constrained Inland Empire circuit, the companies say the analysis found a potential of 20 megawatts of flexible capacity that could support new connections and still meet reliability rules. SCE’s planning chief said the aim is to get more from the grid that exists, without a long and expensive build, and that the utility is looking at several other sites for charging, factories, and commercial customers. The nearly 100 gigawatt line is Duke University research as the release cites it, and the release says each circuit still has to prove its safety margins. The more-than-1-gigawatt line and the $10 billion line are GridCARE’s claims about its platform. The release does not print a price for the SCE work, a date for a statewide rollout, or a statement that the tool is already running across the whole system.