Autonomous AI scientists for materials discovery.

Periodic Labs builds autonomous AI scientists paired with real, automated physical laboratories, running hypothesis generation, experiment design, robotic experimentation, and data collection in a continuous loop to accelerate discovery in materials science, physics, and chemistry. The company's first target is next-generation superconductors, with a broader ambition to find novel materials across the periodic table. Its core bet: internet-scale text has been exhausted as a source of new scientific knowledge, so the next frontier is AI systems that generate their own real-world experimental data.

Periodic Labs was founded in 2025 by Ekin Doğuş Cubuk, a former Google Brain / DeepMind researcher behind GNoME, a landmark model for materials discovery, and one of the creators of one of the first fully autonomous robotic labs to synthesize new compounds from AI-suggested recipes, and Liam Fedus, a former VP of Research at OpenAI and one of the key architects of ChatGPT.

LLMs trained purely on internet-scale text are running out of new ground to cover, and the next real gains in both science and AI will come from models grounded in physical, experimental data. Liam and Dogus are two of the very few people with both the AI research pedigree and the materials science depth to build that loop for real, and the caliber of researchers and investors who moved to back them before the company even had a name confirmed the strength of the thesis instantly.