Particle.news

Download on the App Store

Periodic Labs Launches With $300 Million to Build 'AI Scientists' and Autonomous Labs

Founders from DeepMind and OpenAI plan closed-loop robotic laboratories to generate proprietary experimental data for superconductor discovery.

Overview

  • Periodic Labs emerged from stealth with a $300 million seed round backed by Andreessen Horowitz, DST, Nvidia, Accel, and prominent figures including Jeff Bezos, Eric Schmidt, Jeff Dean, and Elad Gil.
  • Ekin Dogus Cubuk and Liam Fedus lead the startup, bringing experience from DeepMind/Google Brain’s materials work and OpenAI research that helped create ChatGPT and trillion-parameter models.
  • The company aims to automate scientific discovery by deploying AI agents in self-driving labs that run, analyze, and iterate on physical experiments.
  • Its initial targets include discovering higher‑performance superconductors, and reporting indicates early work with an unnamed semiconductor manufacturer on chip heat dissipation challenges.
  • A core objective is to build a proprietary corpus of experimental results—including negative outcomes—to train next‑generation models, positioning the venture within a growing field of AI‑driven materials discovery efforts.