Radical AI's self-driving lab produces 1,200 alloys in six months
Joseph Krause, CEO of Radical AI, argues that the real bottleneck in materials science is experiments, not ideas, and that his company's self-driving lab has produced 1,200 alloys in six months — nearly 10x the pace of the best prior DARPA program.
Score breakdown
Radical AI's self-driving lab demonstrates that automating the physical experimentation loop — not just the modeling — can achieve a throughput in materials discovery that prior state-of-the-art programs could not match.
- 01Radical AI was founded roughly 2.5 years ago around a core belief in experimental data and self-driving labs (SDLs).
- 02In six months, Radical produced 1,200 alloys — nearly 10x the pace of the best prior DARPA program.
- 03Of those 1,200 alloys, 300 are novel compositions and 10 are already being developed for commercial applications.
Joseph Krause, founder and CEO of Radical AI, appeared on the Latent Space podcast to make the case that materials science is fundamentally bottlenecked by the speed of physical experimentation, not by a lack of ideas or models. His company, founded roughly two and a half years ago, was built around a deep belief in experimental data — a conviction that in materials, "the ground truth is the material itself." The result is what Krause calls a self-driving lab: a closed-loop system where an AI scientist generates hypotheses, and a fully automated lab synthesizes, characterizes, and tests them at speeds no human team can match. In six months, Radical has produced 1,200 alloys, a pace he describes as nearly 10x that of the best prior DARPA program, with 300 novel compositions discovered and 10 already in development for commercial applications.
He argues this is why no single model can one-shot a new material destined for something like an iPhone or a Starship rocket.
Krause draws a sharp contrast between AI for biology or small molecules — where representations like SMILES strings can encode most of what matters — and inorganic materials science, where factors like microstructure, processing method, supply chain constraints, and cost cannot be captured in a string. He argues this is why no single model can one-shot a new material destined for something like an iPhone or a Starship rocket. The competitive moat, in his view, lies in the lab infrastructure and the proprietary data it generates, not in the underlying model. He also notes that Radical's AI scientist is exploring elemental families that human scientists, constrained by their own biases, would never have considered.
Key facts
- 01Radical AI was founded roughly 2.5 years ago around a core belief in experimental data and self-driving labs (SDLs).
- 02In six months, Radical produced 1,200 alloys — nearly 10x the pace of the best prior DARPA program.
- 03Of those 1,200 alloys, 300 are novel compositions and 10 are already being developed for commercial applications.
- 04Krause argues no single model can 'one-shot' a new material because factors like microstructure, processing, and supply chain cannot be captured in a string.
- 05The self-driving lab is a closed-loop system: an AI scientist generates hypotheses, and a fully automated lab synthesizes, characterizes, and tests them.
- 06Krause contends the competitive moat in this industry is the lab and the data, not the model.
- 07Radical's AI scientist explores elemental families that human scientists, limited by their own biases, would not typically consider.
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 18, 2026 · 10:40 UTC. How this works →