Human progress is written in materials. Stone, Bronze, Iron, Silicon; we name our eras after them. Every leap forward, from steam engines to semiconductors to lithium-ion batteries, came from a material that didn't exist in usable form before.
Our lab generates a continuous stream of real-world experiments, producing grounded reward signals that feed back into our models and sharpen their designs.
However, material discovery is broken. Physical experimentation is painfully slow, low-throughput, human-curated trial and error-prone. And while computation can now predict new materials at scale, those predictions happen in a vacuum, with no real-world feedback on which candidate materials actually work. The result: a process that is either too slow to matter or too disconnected from reality to trust.
That's the bottleneck we're built to break.
Diffractive Labs is building an AI-driven materials discovery system that moves past the limits of human-curated data. We pair frontier AI with a purpose-built, high-throughput wet lab to run a closed discovery loop: the lab generates a continuous stream of real-world experiments, producing grounded reward signals that feed back into our models and sharpen their designs. We target materials critical for the energy transition.
Our team includes former Isomorphic Labs, DeepMind, and Meta FAIR researchers, one of the world's most-cited condensed matter physicists, and a pioneer of uncertainty in deep learning, backed and advised by leading researchers from across the frontier labs. We've raised $60M, led by GreenOaks, Khosla Ventures and Index Ventures.
