Periodic Labs came out of stealth on Tuesday, and received $300 million in seed funding from a number of prominent investors, including Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt, and Jeff Bezos. Founded by top researchers from OpenAI and DeepMind, the startup aims to use artificial intelligence to “accelerate” discoveries in areas like physics and chemistry.
“The main objective of A.I. is not to automate white-collar work,” said Liam Fedus, one of the startup’s founders. “The main objective is to accelerate science,” he added.
Fedus is a former OpenAI researcher who was part of the team that created ChatGPT. He left OpenAI in March to found Periodic Labs with Ekin Dogus Cubuk, who previously worked at Google DeepMind. Cubuk’s previous projects include an AI tool called GNoME, which discovered over two million new crystals in 2023, materials that could one day be used to power new generations of technology, according to researchers.
Periodic Labs’ main goal is to invent new superconductors that perform better and potentially require less energy than existing superconducting materials. The startup also hopes to discover other novel materials. Another objective is to collect all the physical-world data generated by its AI scientists as they mix, heat, and manipulate various powders and raw materials in their search for breakthroughs.
READ: When AI meets Charlie Kirk: Researcher poses big questions, gets surprising answers (September 24, 2025)
“Until now, scientific AI advances have come from models trained on the internet,” and LLMs have “exhausted” the internet as a source that can be consumed, the company says in an introductory blog post. “At Periodic, we are building AI scientists and the autonomous laboratories for them to operate.”
Periodic Labs has a different approach from most AI companies when it comes to scientific discovery. Leaders in the field often claim that large language models can achieve significant scientific breakthroughs eventually. OpenAI and Meta say their technologies are already pushing toward this goal in areas like drug discovery, math and theoretical physics. However, Fedus and Cubuk believe such companies are not on the path to “true scientific discovery.”
A chatbot “can’t just reason for days and come up with an incredible discovery,” Dr. Cubuk said. “Humans can’t do that, either. They run many trial experiments before they find something incredible — if they even do.”
The startup plans to build its own lab in Menlo Park, California, where robots will conduct scientific experiments on an enormous scale. Company researchers will oversee and guide these experiments while AI systems analyze both the process and the results. Ultimately, the researchers hope the AI will be able to design experiments independently.