Revolutionary AI Startup Periodic Labs Raises $300M to Automate Scientific Discovery, Pioneer Next-Gen Materials
Silicon Valley-based startup Periodic Labs officially emerged from stealth mode yesterday, announcing a groundbreaking seed funding round worth $300 million. The investment was secured by a prestigious consortium comprising tech giants Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt, and Jeff Bezos.
Founded by Ekin Dogus Cubuk and Liam Fedus, Periodic Labs seeks to revolutionize scientific discovery through the application of artificial intelligence (AI). Cubuk, who previously led the materials and chemistry team at Google Brain and DeepMind, played a pivotal role in developing GNoME – an AI tool that discovered over two million new crystals in 2023. These potentially groundbreaking materials could power future generations of technology, according to researchers.
Fedus, on the other hand, brings impressive credentials as a former VP of Research at OpenAI and a key contributor to the development of ChatGPT. He also led the team responsible for creating the first trillion-parameter neural network.
Periodic Labs’ small but highly accomplished team boasts a wealth of experience in AI and materials science, with previous stints at projects such as building OpenAI’s agent Operator and contributing to Microsoft’s MatterGen, an AI designed for materials science discovery.
The mission of Periodic Labs is ambitious: to automate scientific discovery by creating AI scientists. This vision involves establishing laboratories where robots conduct experiments, collect data, iterate, and try again, continuously learning and improving as they progress.
The initial focus of these labs will be on inventing superior superconductors, with the aim of developing materials that offer better performance and potentially reduced energy consumption compared to existing superconducting materials. However, Periodic Labs also plans to explore other promising new materials.
In addition to this research, the lab aims to collect and store all data generated by its AI scientists during the manipulation of various powers and raw materials in their pursuit of innovative discoveries. According to a company blog post, traditional AI advancements have been primarily driven by models trained on internet data, which has now reached its limits as a viable source. By creating AI scientists and autonomous laboratories, Periodic Labs hopes to overcome this limitation.
The hope is that these labs will not only produce next-generation materials but also generate invaluable fresh data that can fuel the ongoing evolution of AI models. While Periodic Labs represents one of the most prominent research teams assembled for this purpose, other groups are also working on AI scientists, with AI automating chemistry discoveries being a topic of academic interest since at least 2023. This includes small startups like Tetsuwan Scientific and non-profit organizations such as Future House, as well as the University of Toronto’s Acceleration Consortium.
