Periodic Labs, a new startup co-founded by former OpenAI researcher Liam Fedus and former Google Brain colleague Ekin Dogus Cubuk, has emerged from stealth operations with a $300 million seed funding round. The company aims to automate scientific discovery, with an initial focus on identifying new superconductor materials.
Felicis led the seed round, which also included investments from Andreessen Horowitz, DST, NVIDIA’s venture capital arm NVentures, and Accel. Notable angel investors such as Jeff Bezos, Elad Gil, Eric Schmidt, and Jeff Dean also participated. The funding round followed an initial tweet from Fedus regarding his departure from OpenAI, which triggered significant venture capital interest.
According to Cubuk, the venture is timed with key advancements in several areas. He cited the recent reliability of robotic arms capable of powder synthesis, the process of mixing and creating new materials. Additionally, machine learning simulations have reportedly achieved sufficient efficiency and accuracy to model complex physical systems relevant to material development. Cubuk further noted the enhanced reasoning capabilities of large language models (LLMs), an area where Fedus contributed significantly at OpenAI.
Cubuk previously co-authored a 2023 paper documenting a Google research project that utilized a fully automated, robotic lab to create 41 novel compounds based on language model suggestions. Periodic Labs has established its laboratory and is currently working with experimental data and simulations, testing predictions as part of its mission to discover new superconductor materials. The company has hired over two dozen experts spanning AI and scientific disciplines, including Alexandre Passos and Eric Toberer.
The founders contend that even unsuccessful experiments will generate valuable data for AI training, potentially redefining the traditional scientific motivation system that often prioritizes successful outcomes for publication and grants. Fedus stated, "Making contact with reality, bringing experiments into the [AI] loop — we feel like this is the next frontier." While the core lab operations are active, the robotic components are not yet fully integrated, with Cubuk indicating they will "take a bit to train."
The broader industry is also exploring AI's role in scientific advancement. OpenAI's VP, Kevin Weil, recently announced the launch of an "OpenAI for Science" unit within the company, focused on building an AI-powered platform to accelerate scientific discovery.
