How did SES.AI use AI to speed up battery material discovery?

SES.AI worked with NVIDIA to map the properties of approximately 100 million possible molecules using AI and computational tools, completing that mapping in 2025. Hu said that without AI, such a task could have taken thousands of years, but with AI it takes about a year. The system can then evaluate material candidates for specific applications, narrowing the search from years to weeks.

What is SES.AI's current production capacity and what does it plan?

SES.AI currently has capacity in South Korea to produce about one million cells annually. The company has publicly described plans to expand its Chungju capacity to approximately one million Li-Metal and Li-ion pouch cells annually to meet demand from U.S. and European drone customers, and is also exploring additional capacity in Japan and Malaysia. However, Hu notes that potential demand is almost 100× the current capacity.

Why are batteries especially critical for drones compared to other applications?

For drones, every gram devoted to a battery is weight that cannot be used for a camera, communications system, medical shipment or other payload. Hu describes the impact simply: an aircraft of the same weight can fly twice as far, or fly the same distance with double the payload, if battery energy density improves—making battery efficiency directly tied to mission capability.

RoboticsDRONELIFEAug 18, 2026Aug 19, 2026, 06:01 JST

Battery startup SES.AI maps the missing link in U.S. drone supply chain

Send on LINE

3 Key Points

What happened
SES.AI, a Massachusetts-based battery maker, has used AI to map approximately 100 million possible molecules for battery development (a process that would have taken thousands of years without AI, now completed in about a year), and operates manufacturing capacity in South Korea with plans to expand in Japan and Malaysia to serve U.S. drone makers seeking alternatives to Chinese components.
Why it matters
U.S. drone manufacturers face an urgent supply-chain problem—federal policy now restricts Chinese components, but American factories cannot yet produce the batteries and critical parts needed at scale. SES.AI's work illustrates a deeper challenge: building not just assembly plants, but entire ecosystems of experienced workers, suppliers, and specialized manufacturing. The company's higher energy-density batteries could allow drones to fly twice as far or carry double the payload at the same weight, making the sourcing shift both feasible and valuable.
What to watch
SES.AI currently produces about one million cells annually from its Chungju facility but says potential demand is almost 100× that, revealing a massive capacity gap. The company is now exploring additional production in Japan and Malaysia. For U.S. drone makers, the bottleneck is no longer aircraft design but the speed and scale at which suppliers can deliver compliant components.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

The U.S. drone industry faces a supply-chain crisis that policy alone cannot solve. Federal restrictions on Chinese components have created urgent demand for alternatives, but as CEO Qichao Hu explains, that demand cannot be met by simply opening assembly factories. The battery industry lost its manufacturing base in the U.S. decades ago as production migrated first to Japan, then South Korea, and finally China. Rebuilding requires not just capital but an entire ecosystem of experienced workers, materials suppliers, equipment makers, and specialized knowledge—a network that takes far longer to recreate than a factory.

SES.AI's journey illustrates how this gap is being addressed. The company's detour into electric-vehicle manufacturing proved valuable, imposing manufacturing standards (more than 3,000 quality checkpoints per cell, compared with about 400 for drones) that have now been brought back to aviation. Its partnership with GM Defense drove capacity development in South Korea, and its use of AI to map a "Molecular Universe" of battery materials compressed what would be centuries of traditional research into a single year. Yet even with these advances, the company produces only about one million cells annually while potential demand is almost 100× that. The gap reveals the real bottleneck: not innovation or design, but the time and capital required to build manufacturing capacity at the scale the drone industry will need.

FAQ

How did SES.AI use AI to speed up battery material discovery?: SES.AI worked with NVIDIA to map the properties of approximately 100 million possible molecules using AI and computational tools, completing that mapping in 2025. Hu said that without AI, such a task could have taken thousands of years, but with AI it takes about a year. The system can then evaluate material candidates for specific applications, narrowing the search from years to weeks.
What is SES.AI's current production capacity and what does it plan?: SES.AI currently has capacity in South Korea to produce about one million cells annually. The company has publicly described plans to expand its Chungju capacity to approximately one million Li-Metal and Li-ion pouch cells annually to meet demand from U.S. and European drone customers, and is also exploring additional capacity in Japan and Malaysia. However, Hu notes that potential demand is almost 100× the current capacity.
Why are batteries especially critical for drones compared to other applications?: For drones, every gram devoted to a battery is weight that cannot be used for a camera, communications system, medical shipment or other payload. Hu describes the impact simply: an aircraft of the same weight can fly twice as far, or fly the same distance with double the payload, if battery energy density improves—making battery efficiency directly tied to mission capability.

DRONELIFERead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.