Ambrosia Energy to Build 10 GWh Battery Factory in Texas for Off-Grid Data Centers

Ambrosia Energy has started work on a 10 gigawatt-hour battery factory near Austin, Texas, to supply its own off-grid power plants for data centers. Co-founder and CEO Ben Longmier announced the project on stage at TechCrunch Disrupt in San Francisco on Oct. 14, according to TechCrunch.

The factory will cover 150,000 square feet. Ambrosia plans to reach full production in two to three years. At that rate, the company said the plant will provide enough batteries to build about one 300-megawatt power plant per year at its field sites.

Ambrosia's model is islanded by design, meaning its plants run separate from the main power grid. The company emerged from stealth in June with a plan to build power for data centers at $100 per megawatt-hour. Its setup uses solar to carry the load by day and batteries to discharge at night and on cloudy days, for round-the-clock operation.

Ambrosia will start work on its first commercial plant in West Texas later in 2026. That site will pair 300 megawatts of solar with 660 megawatt-hours of batteries to deliver 30 megawatts of continuous power. A megawatt measures current output, while a megawatt-hour measures stored energy. The company targets 99.3% uptime for the site.

Sections of its smaller pilot plant have run continuously for more than 100 days. That pilot can generate 200 kilowatts of electricity around the clock.

On supply, Ambrosia buys ready-made battery cells and builds its own battery packs in-house. A pack is the finished unit, with cooling, wiring and control electronics around the cells. The Texas factory scales up that approach. The company states it has developed a battery system that makes 24-hour storage economically viable, according to its blog. It describes itself as building vertically integrated solar-plus-battery plants in West Texas, meaning it controls most steps itself, ideal for data centers and industrial operations.

Ambrosia was founded by Ben Longmier and Sara Spangelo, who previously founded satellite startup Swarm, which they sold to SpaceX.

The broader context here is why those numbers look uneven. Three hundred megawatts of solar and 660 megawatt-hours of storage for only 30 megawatts of steady output is heavy overbuilding by normal grid standards. For a standalone system, that extra size is the point. The array is oversized to keep charging through winter light and multi-day clouds, and the battery is sized to shift daytime energy into night with margin. In that design, cells are commodities. Packs, cooling, battery management systems and dispatch software are not.

In my view, the factory choice is the more revealing detail. A 10 GWh pack plant avoids the cost and chemistry risk of making cells from scratch, while keeping in-house the integration work where uptime and cost are decided. It also links factory output directly to construction pace. One plant per year tied to 300 megawatts of incoming solar gives investors and customers a clear unit of scale, even if results will vary by site.

For operators, one detail is worth flagging. A 99.3% uptime target fits many AI training and batch jobs that can save progress and restart, especially across sites. It leaves a gap for loads that must never stop, unless paired with extra backup or flexible scheduling. The pilot's long run suggests the control challenge can be solved. The commercial site will test whether the economics still work at a scale roughly a thousand times larger.

If that test succeeds, the payoff is freedom on location. Plants that can be built where land and sunlight are available, without waiting for a grid connection, change where capacity can go and how fast it can be added. That helps a sector now limited more by power delivery timelines than by chips.