🗼

Explore roles, hiring trends, and funding at AI companies on breakout trajectories. Updated daily.

Periodic Labs

Periodic Labs is an AI company that builds "AI scientists" — software models that can propose, run, and learn from real physics and chemistry experiments. Instead of only training AI on text from the internet, Periodic runs an automated laboratory in Menlo Park where robots and instruments make new materials around the clock. The data from those experiments (both successes and failures) is fed back to train frontier AI models that get better at predicting which materials will work. The company was founded in 2025 by Liam Fedus, a former VP of Research at OpenAI who helped build ChatGPT, and Ekin Dogus Cubuk, a former Google DeepMind materials-science researcher. Their lab handles powder synthesis, thin-film deposition (PVD and PLD), materials characterization, and device-level electrical testing. Customers are primarily in the semiconductor industry, where Periodic sells access to its AI-plus-simulation capabilities to help chipmakers design better materials, memory devices, and manufacturing processes. Longer term the company aims to serve energy, photonics, and advanced manufacturing as well. The business model appears to be two-sided: (1) build a proprietary scientific dataset and models by running experiments at scale, and (2) sell those models and simulation services to industrial customers through "forward deployed engineers" who embed on-site. In short: a vertically integrated materials-discovery lab that turns its own experiments into training data for AI that can then be sold back to industry.

$1.3B post-money (Sep 2025 seed); reportedly $7.5B on pending round valuation · 62 employees · 33 open roles

Careers

Periodic Labs

Open roles