Archive — US — techcrunch
Former OpenAI and DeepMind researchers raise whopping $300M seed to automatescience
Periodic Labs has raised $300 million in seed funding to automate scientific discovery through AI. Founded by former OpenAI and DeepMind researchers, the company aims to build labs where AI-driven robots conduct experiments autonomously. Their initial goal is to invent superconductors that outperform current materials, leveraging extensive data generated in the process.
Key points
- Periodic Labs was founded by Ekin Dogus Cubuk and Liam Fedus, both experienced in AI and materials science.
- The startup raised $300 million from notable investors including Andreessen Horowitz and Jeff Bezos.
- Cubuk led a team at Google that developed GNoME, an AI tool that discovered over 2 million new crystals.
- Fedus is a former VP of Research at OpenAI, having contributed to ChatGPT and the first trillion-parameter neural network.
- Periodic Labs seeks to create 'AI scientists' capable of conducting experiments and generating new data for further AI training.
- The first research focus is on developing efficient superconductors that may consume less energy than current technologies.
Why it matters
- This venture aligns with the increasing trend of using AI for scientific exploration, with parallels in startups focusing on chemistry and materials science.
- The growing investment in AI-driven technologies reflects a broader industry shift towards automating complex research tasks, particularly post-2023 advancements in AI models.
- Periodic Labs mirrors ongoing developments like those at Tetsuwan Scientific and initiatives at academic institutions seeking to integrate AI into scientific research.
Periodic Labs represents a significant leap in the integration of AI in scientific research, potentially revolutionizing material discovery and offering a glimpse at the future of autonomous laboratories and AI scientists.