What is SES AI and why Hyundai, SK, and LG placed their bets on it

SES AI is a Boston-based company founded in 2012 that started as a startup developing lithium-metal batteries — a technology considered the successor to conventional lithium-ion cells due to higher energy density. South Korean conglomerates including Hyundai Motor Group, SK Group, and LG Group gradually invested in the company. Today, SES AI is listed on the New York Stock Exchange (NYSE) and operates three main business pillars: battery storage for data centers, drone cells, and — most interesting for us — AI-driven materials discovery.

When the electric vehicle market began consolidating under the dominance of a few giants competing primarily on price and manufacturing scale, SES AI stepped back from direct competition and pivoted toward drones, robotics, and materials research powered by artificial intelligence. Its manufacturing plant in Chungju, South Korea, originally built as a joint venture with GM Defense, now produces batteries for drones and plans to increase capacity from 330,000 to 1 million cells per year.

Molecular Universe: From prompt to new material

At the core of SES AI's strategy is the Molecular Universe platform, which maps the "universe" of small molecules relevant to battery chemistry. In May 2026, the company launched version 3.0, whose key innovation is the AI agent StarSeeker. While previous versions required researchers to manually run individual tools, StarSeeker can autonomously control the entire platform — all it needs is a task assigned in natural language.

In practice, it looks like this: a researcher writes a prompt along the lines of "Find the top 20 molecules for low-temperature sodium batteries, generate 10 formulations for each, and rank them by performance." StarSeeker then connects simulations, autonomous labs, and molecular databases, runs the necessary experiments, and returns the results. According to CEO Qichao Hu, who presented the technology in an interview with The Korea Times, battery research is now entering a phase where "AI truly does the research for you. The person enters a prompt, but the platform itself comes up with ideas."

What "vibe research" is and where it came from

The term "vibe research" is an analogy to the popular concept of "vibe coding", where a developer describes the desired functionality in natural language and AI generates the code. In vibe research, the same principle is applied to scientific research: instead of code, AI generates hypotheses, proposes experiments, and tests materials.

The concept is not entirely new — similar platforms have been operating in the pharmaceutical industry for several years, where they have become profitable. SES AI believes the battery industry will follow the same path. Hu stated in the interview that some of the largest battery and automotive companies are already using Molecular Universe — though he did not disclose specific names.

Why robots? The humanoid boom is changing battery requirements

The reason SES AI is betting so heavily on robotics is simple: humanoid robots need fundamentally better batteries than anything that currently exists. Most current humanoids use 21700 cylindrical cells (21 mm diameter, 70 mm length), whose capacity has recently increased from 5 Ah to 7.2 Ah — and it's still not enough.

"Elon Musk mentioned that humanoid robots could be built in greater numbers than cars and even greater than people. That means the market potential is enormous," said Hu. And he added a key insight: for humanoid robots, there are currently no government subsidies or regulatory requirements regarding battery type — whoever delivers the best quality at the lowest price wins.

The speed of discovering new materials is thus becoming the main competitive weapon. The classical approach — several years of research, hundreds of scientists, thousands of manual experiments — is no longer sufficient in the era of rapidly growing robotics.

From teams to individuals: The economics of one-person R&D

Perhaps Hu's boldest vision concerns the very structure of research teams. "The trend may end up with a one-person product team. Instead of a large team, you have one product manager and this platform. That's a huge cost saving and acceleration of development," the CEO claims.

Battery companies globally spend 3 to 5 percent of annual revenues on research and development and employ thousands of researchers. To give an idea — if a battery manufacturer is looking for a way to extend cell lifespan at minus 20°C from 6,000 to 8,000 charge cycles, the classical approach would mean years of work. Molecular Universe promises to solve a similar task in a matter of weeks.

However, the platform remains a cost center for now — SES AI posted revenues of 21 million dollars in 2025 (roughly ten times the previous year, driven primarily by the ESS division) and its operating loss narrowed to 82.61 million from the previous 109.24 million. The company is considering spinning off Molecular Universe into a separate entity, which could attract specialized investments and potentially head toward its own IPO.

What it means for Europe and Czechia

For the European — and specifically Czech — context, the SES AI story is relevant for two reasons. First, the European Union is massively investing in its own battery independence through the European Battery Alliance and the EU Battery Regulation. The ability to rapidly discover new materials using AI can help European manufacturers close the gap with Asian competition.

Second, Czechia ranks among European leaders in automotive and Czech companies such as Škoda Auto are investing billions of crowns in the transition to electromobility, including battery module production. If platforms like Molecular Universe become an industry standard, Czech technology companies and research institutions will need to adopt similar AI tools to remain competitive. As of now, no comparable European tool exists — American and Asian platforms dominate.

It's not sci-fi. But the competition isn't sleeping

SES AI is not the only one betting on AI in materials research. Google DeepMind already published a breakthrough tool called GNoME in 2023, which discovered 2.2 million new crystal structures — of which 380,000 were marked as stable and suitable for experimental synthesis. Microsoft is developing Azure Quantum Elements and the Toyota Research Institute is using AI to search for new materials for EV batteries.

But SES AI has the advantage of also being a battery manufacturer — it can immediately test its AI discoveries in its own labs and production lines. And thanks to its plant in Korea, it has access to a top-tier supply chain and the know-how of Asian giants. Whether Molecular Universe becomes the "ChatGPT for battery scientists" or remains a specialized tool for a few big players, only time will tell — and above all, whether the platform manages to transition from a cost center to profitability.