CuspAI raises $450M at $2.6bn to launch AI Materials Foundry

The interesting move at CuspAI this week is not the money, though the money is not small. The Cambridge, UK startup closed a $450 million Series B at a $2.6 billion valuation, led by Kleiner Perkins and NEA with Bezos Expeditions participating, according to the Financial Times and corroborating reports. The interesting move is the shape of what they are building around it: an 'AI Materials Foundry' launched alongside the raise with more than 45 partners, including Nvidia, Meta, Samsung, Hyundai Motor Group and Lam Research.

Materials discovery has been the 'AI for science' pitch that everyone gestures at and few have shipped as a business. CuspAI is trying to route around that by pairing foundation models with physics-based simulation and by pulling the chip industry itself into the tent. ASML, Meta, Samsung and Hyundai are named as existing users of its materials search engine. If that adoption is real customer use rather than letters of intent, it starts to define what a paying customer for scientific foundation models actually looks like: semiconductor and manufacturing supply chains that already burn years on materials trial and error.

The advisory board is where the story gets a little more theatrical. Geoffrey Hinton and Yann LeCun, two of the field's foundational figures, sit alongside Kristin Persson and Verity Harding. That is an unusually stacked bench for a two-year-old startup and reads more as a signal to customers and governments than as day-to-day science management. Co-founder Max Welling did his PhD under Gerard 't Hooft and trained under Hinton at Toronto, so the personal ties are real, but readers should still weigh advisory-board glamour separately from execution risk.

The honest caveat is that most of what is public is the fundraise and the coalition announcement, not benchmarks, materials shipped or independent validation of the MIRA platform CuspAI cites. A valuation up roughly fivefold in nine months, from about $520 million to $2.6 billion, is the kind of jump that rewards asking hard questions about revenue and about how quickly a foundry-style partner network actually produces novel compounds.

What is worth watching from here is whether any of the named corporate partners publish a specific material or process that came out of this pipeline. That, more than the advisory board or the round size, will decide whether AI-designed materials becomes an operating category or stays a very well-funded promise.