Former OpenAI and DeepMind researchers raise whopping $300M seed to automate science 

Periodic laboratories Came out on Tuesday from Stealth with a $ 300 million war box as a seed round, supported by a technical industry that is who: Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt and Jeff Bezos.

Periodic laboratories were founded by Ekin Dogus Cubuk and Liam Fedus. Cubuk led the material and chemistry team at Google Brain and DeepMind, where one of his projects was, for example, an AI tool called Gnome. That tool discovered more than 2 million new crystals in 2023Materials that can be used on one day to provide new generations of technology with power, researchers say.

Fedus is a former VP of research at OpenAI and one of the researchers who have helped to make chatgpt. He also led the team that the First trillion parameter Neural network.

The small team is also filled with researchers who have worked on other large AI and material scientific projects, from building OpenAi’s agent operator to working on Microsoft’s Mattergen, an LLM Materials Science Discovery AI.

The purpose of periodic laboratories is nothing less than to automate scientific discovery, to create AI scientists, says the company. This means building laboratories where robots perform physical experiments, collect, repeat and try again, learn and improve as they go.

The first goal of the lab is to find out new super conductors who hope to perform better and possibly requires less energy than existing superconducting materials. But the well -funded startup also hopes to find other new materials.

Another goal is to collect all physical world data that its AI scientists produce while they mix and heat and otherwise manipulate different powers and raw materials in their search for something new.

“Until now, scientific AI preliminary output comes from models trained on the internet” and LLMS have “exhausted” the internet as a source that can be consumed, the company says in an introduction Blog post. ‘[A]T Periodic, we build AI scientists And The autonomous laboratories that they can operate. “

The hope is that the laboratories will not only invent the next generation of materials, but they will produce invaluable value that AI models can consume to continue their evolution.

Although this is perhaps one of the most impressive groups of researchers to put together a startup for this goal, this is not the only one working on AI scientists. AI as an aid to automate chemistry discoveries, is one Subject of academic research since at least 2023. It is the pursuit of small startups such as Tetsuwan Scientific, as well as non -profit organizations such as Future house and the University of Toronto’s Gear consortium.