ROBNEWS
Even SES.AI, the most advanced American drone battery startup, makes its cells in South Korea, exposing the layer NDAA (the annual defense policy bill) driven policy has not yet rebuilt.
US drone policy is funding the visible layer: airframes and NDAA-compliant components. The cell beneath them is the harder rebuild, and the layer the policy has not yet financed at scale.
The United States stopped making lithium battery cells at scale decades ago. The capability moved first to Japan, then to South Korea, and more recently to China. Domestic drone demand is about to scale past the cell-making capacity the country actually has on the ground.
The cleanest evidence of that gap is SES.AI, a Woburn, Massachusetts-based battery developer often cited as the most advanced US drone-cell startup. Its high-volume cell production runs through a facility in Chungju, South Korea, not the United States. The Massachusetts site is described in DroneLife's facility reporting as an R&D and small-format environment, not bulk manufacturing.
SES.AI's founder put the structural point plainly in that same DroneLife report: "If you look at the evolution of the battery industry, we lost a lot of the capabilities in the US decades ago." The quote and the company's own manufacturing footprint point to the same conclusion: rebuilding that capability is not a quarter's work.
SES.AI's Q2 2026 earnings call transcript and the company's August 2026 8-K filing describe current South Korea cell-making capacity at roughly one million cells annually, against a forward drone demand estimate of close to 100 times that figure. The 100X number is an executive forecast, not a contracted order book, but the direction matches the policy push.
SES.AI's drone cells run through about 400 quality checkpoints, compared with more than 3,000 for automotive cells. That asymmetry is not a defect rate. It reflects a drone-specific design and cost envelope: lighter packs and faster cycles, with lower per-unit scrutiny than automotive cells require. That envelope has to be built from scratch, not borrowed from the EV playbook.
SES.AI's longer-term bet is its "Molecular Universe" materials database, which maps roughly 100 million candidate molecules for battery development and was completed in 2025 in collaboration with NVIDIA. That is an R&D-side capability, not a near-term production-scale fix. It tells the reader where the company is placing its optionality, not where the cells will come from this year.
The policy gap is starting to draw independent attention. Manufacturing Dive's coverage of NDAA-compliant drone manufacturing names SES.AI alongside Unusual Machines, 6K Energy, and CRG Defense as US-side component players, but the cell layer remains the structurally weakest link. Commercial pull is building on top of it: on August 3, 2026, SES.AI entered a framework agreement with Doroni Aerospace to design a battery pack for the H1-X personal eVTOL, evidence that drone and eVTOL customers are signing up for cell supply that, for now, still ships from South Korea.
The open question is whether additional capacity inside the United States arrives on a schedule that matches the demand curve. SES.AI has said it is evaluating additional manufacturing in Japan and Malaysia. Both are East Asian cell clusters, not US sites. Until a US gigawatt-hour-scale cell line for drones is permitted, financed, and built, the most advanced American drone-battery company will still be shipping from Chungju.