Meet Periodic Labs: The AI-Driven Startup Automating Scientific Discovery with $300M in Funding
Silicon Valley-based startup Periodic Labs debuted this week with a substantial $300 million in funding, drawing backing from tech industry heavyweights such as Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt, and Jeff Bezos.
Founded by Ekin Dogus Cubuk and Liam Fedus, Period Labs aims to revolutionize the scientific discovery process through the development of AI scientists. The duo brings a wealth of experience in cutting-edge AI research – Cubuk previously led the materials and chemistry team at Google Brain and DeepMind, where he oversaw projects including the creation of GNoME, an AI tool that discovered over 2 million new crystals in 2023.
Fedus’s background includes serving as a VP of Research at OpenAI, one of the teams responsible for ChatGPT and the first trillion-parameter neural network. He is joined by other seasoned researchers who have contributed to notable AI and materials science projects, like building OpenAI’s agent Operator and working on Microsoft’s MatterGen, an LLM materials science discovery AI.
Periodic Labs’ objective is to automate scientific discovery, transforming labs into high-tech environments where robots conduct experiments, collect data, iterate, and improve upon their findings. Their initial focus will be on inventing new superconductors with enhanced performance and potentially reduced energy consumption compared to existing materials.
The startup also aims to amass the vast trove of physical world data generated by its AI scientists during the course of experimentation, manipulation, and analysis of various elements and compounds.
In a blog post introducing their venture, Periodic Labs explains that previous advancements in scientific AI have been based on models trained using internet data sources, which are now being exhausted as a viable source for such training. The company positions itself as a pioneer in building AI scientists and autonomous laboratories to carry out research in this evolving landscape.
While Periodic Labs marks an impressive venture for these researchers, it is not the only organization pursuing the development of AI-powered scientific discovery – AI-driven chemistry research has been a subject of academic exploration since at least 2023, with startups like Tetsuwan Scientific and non-profits such as Future House also actively engaged in this field, alongside the University of Toronto’s Acceleration Consortium.
