CuspAI secures $450M Series B to scale AI materials discovery

CuspAI is one of the clearest examples of AI moving from the digital world into the physics of atoms. By treating materials discovery as a search problem powered by generative models and simulation, the company is trying to compress multi-year lab cycles into software loops. A $450M Series B backed by Kleiner Perkins, NEA, Bezos Expeditions, and sovereign capital says investors increasingly see AI for science as an infrastructure bet, not a niche application.

For the race to AGI, this matters on two fronts. First, frontier labs will need radically better materials for everything from chips and optics to batteries and cooling—CuspAI is effectively building a model layer for that upstream supply chain. Second, the AI Materials Foundry network hints at a new pattern: AI-native consortia that stitch together compute, data, and wet labs across borders. If successful, this could become a template for how AGI-era research is organized: not just bigger models, but tightly coupled ecosystems that can turn model output into real-world artifacts much faster than traditional R&D.