News
AI veterans raise $300 million to build "AI scientists" at Periodic Labs
- By John K. Waters
- 10/02/2025
Former senior researchers from OpenAI and Google DeepMind have launched a startup called Periodic Labs, which has emerged from stealth with $300 million in seed funding from some of the biggest names in the technology industry.
The San Francisco-based company was co-founded by Liam Fedus, one of the creators of ChatGPT and a former vice president of research at OpenAI, and Ekin Dogus Cubuk, who led the materials and chemistry teams at Google Brain and DeepMind. Investors include Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt, and Jeff Bezos.
Automating science with AI
Periodic Labs says its goal is to create "AI scientists" — systems that combine artificial intelligence with robotics to automate experimentation in physics and chemistry. The approach involves building laboratories where robots mix materials, run experiments, and iterate on results. At the same time, AI models learn from both the scientific literature and new data generated by the machines.
The startup's initial research target is new superconductors that could outperform today's materials, potentially reducing energy demands. Over time, Periodic also aims to discover other classes of materials and collect physical-world data to train AI models.
Cubuk previously worked on GNoME, an AI-based deep learning system for materials discovery, which is credited with discovering more than 2 million new crystals in 2023. Fedus led the team that developed the first trillion-parameter neural network and was part of the group behind ChatGPT.
A crowded but ambitious field
Periodic Labs is among a growing number of ventures trying to bring automation to scientific discovery. Universities, nonprofits such as Future House, and startups like Tetsuwan Scientific are exploring similar ideas, while DeepMind's AlphaFold has already shown how AI can accelerate breakthroughs in drug discovery.
The company has hired more than 20 researchers from OpenAI, Meta, Google, and other leading AI labs. Its first dedicated facility, in Menlo Park, California, will house robots capable of running thousands of experiments around the clock.
Market and outlook
Fedus said in a recent interview that the company expects its tools to appeal to engineers and researchers in sectors such as semiconductors, aerospace, and defense, which spend heavily on R&D but lack specialized AI-driven discovery tools.
Analysts say the approach is ambitious but unproven. Running AI models in digital environments has scaled rapidly, but translating those methods to the physical world of materials science is far more complex. "Is this going to solve cancer in two years? No. But is this a good, visionary bet? Yes," said Oren Etzioni, former CEO of the Allen Institute for AI.
Periodic Labs' backers are betting that combining leading AI expertise with automated labs could shorten the timeline for scientific breakthroughs — and create a new source of valuable data to train future AI systems.
About the Author
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at [email protected].