Periodic Labs Inc., an artificial intelligence startup working to speed up scientific research, has raised $300 million in funding.

Andreessen Horowitz led the seed investment with participation from Nvidia Corp., Accel and DST. Periodic Labs said in its announcement of the deal on Tuesday that several prominent tech industry figures contributed as well. The group included Jeff Bezos, Eric Schmidt, Google LLC Chief Scientist Jeff Dean and startup investor Elad Gil.

Bloomberg reported last month that the investment was expected to give Periodic Labs a pre-money valuation of $1 billion. The paper’s sources added that OpenAI had considered participating in the raise. 

Periodic Labs is developing AI models that can automate some of the work involved in scientific research projects. According to the company, its algorithms will operate autonomous powder synthesis labs. Those are facilities where materials are turned into powder, heated and mixed to develop new compounds. The labs will use robots to carry out certain research tasks.

In some cases, a single materials science experiment can produce several gigabytes of data. Periodic Labs plans to use data from projects carried out in its labs to enhance its AI models. The company hopes that the data will enable its algorithms to perform research more effectively than algorithms trained on the public web.

Periodic Labs’ first goal is to develop superconductors that can operate at higher temperatures than current materials in the category. Such superconductors could facilitate the development of more efficient chips and power grid infrastructure.

When electricity travels across a metal wire, some of the electrons turn into heat. That heat is the reason data centers require cooling equipment. Superconductors can conduct electricity without producing any heat, which means they could theoretically be used to build more efficient servers.

In practice, current superconductors don’t lend themselves to making data center hardware. The reason is that they only work in subfreezing temperatures. That’s the challenge Periodic Labs hopes to tackle with its semiconductor research push. 

The company is led by co-founder Ekin Dogus Cubuk and Liam Fedus. Cubuk previously headed Google DeepMind’s materials and chemistry team, where he co-created a material discovery model called GNoME. Fedus is OpenAI’s former vice president of research.

The Periodic Labs team also includes more than two dozen other researchers from major tech firms such as Meta Platforms Inc., Databricks Inc. and Samsung Electronics Co. The company will use its new funding to onboard additional technical professionals. The recruiting drive is set to place particular emphasis on hiring AI researchers, experimentalists and simulation experts. 

“We’re also working to deploy our solutions with industry,” Periodic Labs staffers wrote in a blog post. “As an example, we’re helping a semiconductor manufacturer that is facing issues with heat dissipation on their chips. We’re training custom agents for their engineers and researchers to make sense of their experimental data in order to iterate faster.”

Image: Unsplash