Periodic Labs emerged from stealth mode on Tuesday, announcing a massive $300 million seed investment round. The funding comes from a roster of top tech investors, including Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt, and Jeff Bezos.
The company was established by Ekin Dogus Cubuk and Liam Fedus. Cubuk previously led the materials and chemistry division at Google Brain and DeepMind, where he worked on projects such as GNoME, an AI system that identified more than 2 million new crystals in 2023. Experts believe these materials could eventually enable future technological breakthroughs.
Fedus previously served as Vice President of Research at OpenAI and contributed to the development of ChatGPT. He also led the team responsible for building the first neural network with a trillion parameters.
The compact team at Periodic Labs includes experts who have contributed to significant AI and materials science initiatives, such as developing OpenAI’s agent Operator and working on Microsoft’s MatterGen, an AI system for discovering new materials using large language models.
According to the company, Periodic Labs aims to revolutionize scientific discovery by creating AI-driven scientists. Their vision involves laboratories where robots perform experiments, gather data, refine their methods, and continuously learn and improve.
The lab’s initial focus is on developing innovative superconductors that could outperform current materials and potentially use less energy. With its substantial funding, the startup also intends to discover additional novel materials.
Another objective is to amass comprehensive physical data generated by its AI scientists as they experiment with and manipulate various substances in pursuit of breakthroughs.
In a blog post introducing the company, they note, “Up until now, progress in scientific AI has relied on models trained with internet data,” and that large language models have “reached the limits” of what the internet can provide. “At Periodic, we are developing AI scientists and the autonomous labs they need to operate.”
The company hopes these labs will not only create advanced new materials, but also generate valuable new datasets for AI models to learn from and further advance.
Although this may be one of the most distinguished teams to launch a startup with this mission, they are not alone in the field. The use of AI to automate chemical discovery has been explored in academia since at least 2023. Startups like Tetsuwan Scientific and organizations such as Future House and the University of Toronto’s Acceleration Consortium are also pursuing similar goals.
