Radical AI’s – In a Midtown Manhattan lab, Radical AI is letting an “AI scientist” run experiments nearly autonomously—reading thousands of papers in seconds, generating hypotheses in real time, and pushing toward 100 tests a day. The startup says its active learning loop co
In midtown Manhattan. a robotic arm moves with the calm certainty of repetition: it lifts small glass bottles. mixes and weighs pellets of iron and other elements. then passes the batch to another machine that melts it into an alloy. Nearby. separate systems analyze what’s been made—composition. structure—and then push the material through tests for hardness and how well it resists oxygen and heat.
What makes the scene unusual isn’t the lab equipment. It’s the fact that the experiment was designed by AI, and the “self-driving” lab is running nearly autonomously, so that if a new idea arrives at 4 a.m., the process simply starts again.
The work is happening inside Radical AI. a startup betting that faster materials discovery can ripple outward across some of the hardest problems in industry. Radical says it is using AI to find new materials that could eventually be used to make jet engines last longer. help enable fusion energy. and address other bottlenecks in how today’s materials perform under extreme conditions.
Joseph Krause, the startup’s CEO, described the shift in terms of scale. “This AI-driven process is really a radical shift to allowing us to scale scientific discovery,” Krause said. “You go from 10 scientists focused on one problem to one scientist focused on 10 problems at a time.”
The company’s argument is built around time. A typical process to develop a new material is slow—often taking 20 years or more. Scientists form hypotheses. produce a material. characterize it. test it. and then cycle back to new hypotheses when the results don’t land. At the same time. there’s a “greater need for new materials. ” Radical says. not only to enable solutions like clean energy. but to confront challenges already tied to existing materials. including shortages and the environmental footprint of extraction and production.
Radical’s pitch is that its AI scientist can accelerate each step and work on multiple steps in parallel. When the AI system needs to review the scientific literature, Krause says it can read 10,000 papers in five seconds. Then. when a human team begins defining a new problem for a company or industry. it starts by giving the AI a list of the specific properties the material should have.
From there, the AI agent references 380,000 papers and 57 million data points from the lab. Krause emphasized why internal lab data matters—failures usually aren’t published in scientific papers—so the lab’s own record can be especially useful. From that pool. the system may propose anywhere from a dozen to a few hundred materials to try in the lab.
The lab itself uses standard materials science equipment. but Radical says it’s almost all automated and run by the AI. The company says the system can run up to 50 experiments a day. and it is aiming to increase to 100 experiments a day by the end of the summer. Krause contrasted that output with what a human materials scientist might achieve. saying such a scientist might do 50 experiments in a year.
Krause also described the lab as inherently parallel—one that can keep working even while new results come in. “This is parallel in nature. meaning I can look at scientific publications. I can run quantum chemistry. I can look at my past experimental results. and I can generate new hypotheses simultaneously. ” Krause said. “And I can do those things at very very large scale.”.
As experiments run, the AI keeps learning and pushes forward on other steps at the same time. Human materials scientists give the AI notes on results—for example, that they observed cracks in a material—something Krause said helps the system begin to learn scientific intuition.
Radical says it has built an “active learning loop,” with the lab capturing data and the AI studying it to make a new hypothesis in real time. Krause believes Radical is the first to be running such a loop.
The idea is drawing attention beyond Radical. A growing number of startups are developing similar systems. Lila Sciences raised $350 million in a Series A round last year. Orbital uses AI to make critical hardware for data centers. CuspAI, another startup, is using AI to develop a new material designed to remove PFAS from water, among other products.
Radical is still early. The lab was built last year, and the company is now moving to a larger space at Brooklyn Navy Yard. In the new location, Radical says it will add new equipment to make the process even more automated.
Krause declined to name what new materials the AI scientist has created so far. citing that the company is still in the process of filing patents. Still, Radical says one recent campaign discovered around 300 novel compositions over 16 weeks. Some of the most promising materials were sent to Purdue Applied Research Institute. which verified that they outperformed the leading material of its type.
For now, Radical says it is focused on metal alloys, with applications such as jet engine parts. Designing new materials that can better handle intense heat is intended to help components last much longer. The company also connects the work to efficiency: it says materials that endure higher heat could potentially allow fuel to burn hotter and more efficiently. saving money and emissions.
Beyond aviation, the startup is also working on materials for the defense industry and fusion energy. Krause called fusion “a serious materials problem,” saying many parts of a reactor require novel materials to exist.
Later, Radical plans to move beyond alloys to other material classes and other applications. Krause summed up the company’s broader bet in a way that points to how fundamental materials are to the next wave of tech. “There is not a single system on Earth that will move forward without a novel material,” Krause said. “Semiconductors, robotics, all the way through energy generation. Materials are the gateway to the next age of innovation.”.
Radical AI self-driving lab materials discovery AI scientist metal alloys jet engines fusion energy Purdue Applied Research Institute Brooklyn Navy Yard seed round automation active learning loop