How Radical AI is building a self-driving materials lab

Radical AI, a New York City-based autonomous materials science discovery company, was founded in 2024 and aims to accelerate materials research and development by integrating AI, engineering, and materials science in self-driving labs. Its lab can create and characterize more than 25 alloys a day.

The company’s AI systems screen billions of material compositions to predict structure and physical properties and identify experimental candidates. Synthesis and characterization are then performed to generate data that is fed back to the prediction engine to close the loop.

Radical AI Autonomous Driving Lab

Radical AI’s approach reflects broader industry-wide trends. According to Cypris R&D Intelligence, researchers using AI are producing 44% more material discoveries than traditional methods, and the technology has the potential to compress traditional 10-20 year timelines into 1-2 years in some cases.

“Machine learning is a game-changer in materials discovery because it allows scientists to test new chemicals and manufacture new materials in the lab without having to repeat the same processes over and over again,” said Kristin Persson, director of Berkeley Lab’s Materials Project.

Radical AI’s lab aims to create a complete closed-loop system from discovery to production of new materials. The company’s AI system can read publications, formulate hypotheses, send materials to the lab for synthesis, characterization, and testing, and capture and analyze data that the system uses to design the next loop.

enter the loop

CEO Joseph Krause explained that once a loop is initiated, multiple steps occur simultaneously. “So we may have come up with new experiments by reading publications and characterizing the last one we created to extract real information in real time,” he says.

Krause brings materials science expertise from his doctoral work at Rice University and time at the Army Research Laboratory before joining AlleyCorp as an investor. He co-founded Radical AI with Jorge Colindres (President, also from AlleyCorp) and CTO Gerbrand Ceder, who was a Principal Investigator at Lawrence Berkeley National Laboratory’s Autonomous Laboratory.

The company reportedly raised $55 million in a Seed+ round led by RTX Ventures in July 2025, with participation from NVIDIA’s NVentures, Eni Next, and AlleyCorp, followed by a $60 million Series A.

Competitors such as Lila Sciences, which has raised hundreds of millions of dollars in funding, and Periodic Labs, founded in 2025 by prominent AI researchers, are building similar “self-driving labs.” Academic institutions are also making progress. Argonne National Laboratory’s Polybot screened 90,000 material combinations in a few weeks, a task that would normally take months manually. Meanwhile, Berkeley Lab’s A-Lab has successfully synthesized a material predicted by Google DeepMind’s AI model.

Evolving human-machine dynamics

Radical AI’s long-term goal is an end-to-end automated loop, but the lab is not yet fully manual. Some tests are semi-autonomous, with scientists still guiding the analysis and labeling.

All samples are synthesized autonomously, but tensile tests are performed only semi-autonomously. Other tools such as SEM, XRD, XRF, and oxidation are completely autonomous.

“This is also really important in terms of being able to connect in a complete flywheel. If you can’t bring in data and understand it all and analyze it, you’re not really building knowledge. You’re just reading information. If you can download an image, but can’t analyze it like a scientist, you don’t have scientific knowledge. You just have the ability to bring the image into a model. That’s what scientists are really focusing on today to continue training systems.”

bottleneck remains

“The biggest bottleneck in materials science has always been processing and manufacturing. Creating that know-how is very difficult and takes a very long time,” says Kraus. The 10-20 year timeline for the materials science industry is primarily determined by processing and manufacturing stages.

Radical AI is still working on this challenge. The company has not yet moved to manufacturability, Kraus explained. The company has this stage in mind, but is still in the discovery and testing phase.

AI in materials science is still in its infancy. “How it works today and how we envision it working is still a little different. There’s just a lot of tooling to do,” Ceder said in MIT Technology Review.

Kraus said the company eventually plans to expand beyond discovery into manufacturing and become vertically integrated as a materials supplier.

“If we can vertically integrate and bring AI and autonomy to the manufacturing process as well, we can combine new discoveries and manufacturability testing in weeks versus today’s 10-15 year processes. That’s a great opportunity for this technology,” he added.

fill the data gap

Most experiments fail, but the results of failed experiments are rarely made public to the broader scientific community. This means that scientists from different companies and institutions are likely repeating the same experiments that other companies have already conducted and failed. With Radical AI, all data, including data from failed experiments, is recorded, indexed, and considered when making predictions and hypotheses.

“The ability to capture know-how and build on it is what makes experimental results so important. Typical human science activities today don’t do this,” Kraus said.

This is also important for training AI models, Krause explained. “If you can’t accumulate information so that you can actually learn, those models are only going to be as good as simulations, and that’s where the industry is right now,” he said.

We don’t just use simulated data. We use real experimental data and can build on each other, thus continually increasing our knowledge. That’s why data is so important.

How can Genesis missions fit in?

The Genesis Mission, announced by the Trump administration in November, is a “dedicated, coordinated national effort to unleash a new era of AI-powered innovation and discovery that can solve this century’s toughest problems.” The mission will be conducted within the Department of Energy (DOE), directed by Undersecretary for Science Dario Gil.

The goal is to build an “integrated discovery platform” that will be “the world’s most complex and powerful scientific instrument ever built,” according to a DOE press release.

The Genesis mission will focus on three initiatives aimed at achieving “American energy dominance,” advancing discovery science, and ensuring national security. The mission’s ultimate goal is to “develop an integrated platform that connects the world’s best supercomputers, laboratories, AI systems, and datasets across every major scientific discipline, doubling the productivity and impact of American research and innovation within 10 years,” according to the mission’s website.

On December 20th, Radical AI joined the Genesis mission. “In our opinion, the Genesis mission reflects the current era of science, demonstrates strong leadership and scientific discipline, and is an opportunity to demonstrate what cutting-edge science can deliver,” Kraus said. “This is an opportunity to show the American people how strong American science is, and our ability to actually build the most advanced scientific tools ever built – HPC, quantum AI, self-driving labs and robotic automation – within this single closed-loop system. That’s exactly what we believe as a company, and that’s why we’re so honored to be chosen to be a part of it.”

The next direction for materials research and development

“We believe that what we are building is one of the most important companies in the world, because its impact can be profound for all of humanity.Which industries are you interested in: automotive and aerospace, manufacturing and defense, climate, energy, semiconductors, electronics? It doesn’t matter. The most important industries in the world have direct consequences for materials and R&D. That’s the opportunity we’re seeing and what’s really exciting us when we think about vertically integrated AI and building AI.” Materials manufacturers and suppliers that value autonomy are what the industry needs. ”

Radical AI is touting an end-to-end AI-driven materials pipeline, but the challenges to getting there will require significant physical validation while navigating an increasingly competitive environment. The company has yet to address the manufacturing bottlenecks that are driving the industry’s longest schedules and faces competition from well-funded rivals and established academic research institutions. But with $115 million in funding, a leadership team that includes former Berkeley Lab principal investigator Gerbrand Ceder, and selection for the Genesis mission, Radical AI is positioning itself at the forefront of an industry transformation that could reshape how the world develops new materials.