In a bold shift across the AI industry, more than 20 leading researchers have left roles at OpenAI, Google DeepMind, and Meta to join Periodic Labs, a new startup aiming to blend artificial intelligence with scientific discovery.
Periodic Labs emerged from stealth with a $300 million seed round, backed by prominent investors including Andreessen Horowitz, Nvidia, Accel, and Jeff Bezos. The startup was founded by Ekin Dogus Cubuk, formerly of Google Brain and DeepMind, and Liam Fedus, who worked as a VP of Research at OpenAI and was involved in early efforts to build ChatGPT.
The company’s stated mission is sweeping: to automate scientific discovery in fields like physics and chemistry. Periodic Labs envisions AI systems that conduct physical experiments, analyze results iteratively, and accelerate breakthroughs in materials and chemicals. Its immediate goal includes developing new superconductors with higher performance or lower energy requirements.
According to TechCrunch, the startup argues that traditional large language models and AI systems trained on internet data have matured in their capability to generate text and code—but that the frontier now lies in applying AI to the physical world through experimentation.
By pulling top researchers from established AI labs, Periodic Labs is signaling a pivot in the AI talent landscape—from purely software or language-driven models to integrated systems that can bridge computing and real-world science. But the challenge is steep: success requires expertise in AI, robotics, lab automation, and materials science, along with access to advanced facilities and funding.
As this story develops, all eyes will be on whether Periodic Labs can deliver on its high ambitions—and whether this trend of research talent migration reshapes how we think about AI’s role in scientific progress.
