Top OpenAI and Google brain researchers spark $300 million rush for new startup Periodic Labs

openai

Periodic Labs, a new venture co-founded by former OpenAI researcher Liam Fedus and ex-Google Brain scientist Ekin Dogus Cubuk, has emerged from stealth with a massive $300 million seed round led by Felicis, drawing participation from major names including Andreessen Horowitz, NVentures, Accel, DST, and top angel investors such as Jeff Bezos, Eric Schmidt, Elad Gil and Jeff Dean.

The idea for Periodic Labs began about seven months ago when Fedus and Cubuk discussed the growing potential of AI in scientific discovery. Cubuk, known for his research in material science at Google Brain, and Fedus, one of the creators of ChatGPT and head of OpenAI’s post-training team, saw an opportunity to merge AI reasoning, robotics, and simulation to accelerate material discovery. Recent progress in robotic arms capable of handling powder synthesis, improved machine learning simulations, and advances in large language models convinced them that the timing was right.

Their vision is to create an AI-driven lab where language models propose new compounds, robotic systems execute experiments, and simulations refine the results — effectively automating material science. Cubuk had already helped demonstrate this concept in 2023 at Google, where an AI-powered lab produced 41 new compounds from model-generated recipes. Fedus believes this integration of real-world experimentation into AI workflows represents “the next frontier” for science.

Interestingly, OpenAI itself is not backing the startup, despite early speculation. After Fedus announced his exit from OpenAI, investors swarmed the project, sending pitches and even what he described as “love letters.” The first serious conversation came from Peter Deng, a former OpenAI executive now at Felicis. Deng decided to invest after an impromptu walk through San Francisco’s Noe Valley, where Fedus explained that true scientific breakthroughs require AI systems capable of testing their hypotheses in the real world.

Periodic Labs now employs more than two dozen top researchers from AI and material science backgrounds, including Alexandre Passos, Eric Toberer and Matt Horton. The startup has already established its lab and begun testing simulations and experimental data. Its first mission is to discover new superconductors — materials that could significantly reduce energy use and advance future technologies.

While expectations are high, both founders acknowledge that scientific discovery is inherently uncertain. Still, the blend of robotics, simulation and AI positions Periodic Labs as one of the most ambitious attempts yet to fuse artificial intelligence with experimental science.