Cambridge, and a room full of competitors
On 20 July 2026 CuspAI announced something stranger than a funding round. The Cambridge company, founded in 2024 by Dr Chad Edwards and Professor Max Welling, confirmed a 450 million dollar Series B, roughly 415 million euros or 335 million pounds, led by Kleiner Perkins and NEA. Then it launched the AI Materials Foundry, and the guest list mattered more than the cheque.
More than 45 founding partners signed on across the US, Asia-Pacific and Europe. Nvidia and Meta are in it. So are Samsung, Hyundai Motor Group, Henkel, Applied Materials, Tokyo Electron and Lam Research. The Foundry pools data, laboratory facilities, computational resources and scientific expertise into a single global network aimed at speeding up materials discovery.
The framing at the launch was blunt: "If we don't make progress fast, the next 50 years of industrial progress will be constrained by a single challenge: the world needs materials that don't yet exist." Read that as a recruitment pitch and the structure makes sense. A company does not raise 450 million dollars to sell a licence. It raises it to convene.
What the 45 actually agreed to share
The unusual part is not the technology, it is the consent. Applied Materials, Tokyo Electron and Lam Research compete head-on in semiconductor equipment, and all three agreed to put data, laboratory facilities and compute into a shared venue. Rivals do not do that casually.
They do it when the discovery problem has grown larger than the rivalry over it. A pre-competitive venue lets each member contribute inputs that are punishing to generate alone, such as characterisation data, lab hours and computational capacity, and draw back a search capability none of them would have funded on its own timetable. The pooling is the product. The software sits on top of it.
CuspAI disclosed no figures on cost reduction or time reduction in materials discovery, and that absence deserves respect rather than filler. The efficiency gain is not quantified. What is quantified is the commitment behind it: more than 45 organisations, three regions, a 2.6 billion dollar valuation, and a balance sheet of 450 million dollars behind the entity doing the convening. Those numbers describe conviction, not results.
Discovery just became a membership good
The materials your product will specify in five years are being defined inside a consortium you may not belong to. That is the shift, and it is not priced into anyone's supplier scorecard yet.
Until now, a buyer of coatings, adhesives, battery chemistries or semiconductor inputs could treat materials as a market: several vendors, comparable specifications, negotiation on price and lead time. A Foundry-scale pool changes the sequence. Direction of travel gets set by the members who contribute data and lab time, and the output reaches everyone else later, as a catalogue rather than a conversation. A buyer outside the network becomes a price-taker on what emerges instead of a participant in the roadmap that produced it.
Note what membership actually confers. Not a discount. Visibility. Knowing eighteen months early which material classes are being explored is what lets an engineering team design around a component that does not exist yet. Roadmap visibility is now an asset with an access list, and access lists get harder to join as valuations climb, which this one did, from 520 million dollars in September 2025 to 2.6 billion dollars now.
European public money bought a seat. European buyers did not.
Invest-NL, a Dutch state investor, is on the cap table of a UK-based company, which means European public money has bought exposure to the Foundry while most European industrial buyers have bought nothing. That asymmetry is the detail worth carrying into a board meeting.
The rest of the round reads as a global capital list: Bezos Expeditions, Glade Brook Capital Partners, Lux Capital, AMD Ventures, Tru Arrow Partners, StepStone, Sovereign AI Venture Fund and John Doerr joined the new money, with Temasek, Basis Set Ventures, Giant Ventures, Touring Capital, Prosus, Phoenix Court and Northzone returning. Henkel is the clearest European industrial name among the founding partners.
For a European manufacturer, the practical reading is simple. State participation in a financing round does not transmit any advantage to your plant, your bill of materials or your qualification timeline. The seat that matters is the partner seat, not the investor seat, and those are separate transactions. Treating a national investment agency's involvement as national capability is the mistake to avoid this year.
The question to put to your suppliers this quarter
If materials appear anywhere in your bill of materials, put one question to every supplier before your next sourcing decision: are you a Foundry partner, and what does your membership give you that we do not currently receive?
Semiconductors, batteries, coatings and adhesives are the obvious exposures, but the test is broader. Any component whose performance ceiling is set by a material rather than a design is now a component whose future you cannot see without membership somewhere in the chain. Ask in writing, record the answer, and repeat the question at the next contract review, because the partner list will move.
Then decide deliberately whether you want to be inside. Some buyers are large enough to negotiate participation directly. Others will be better served by concentrating spend with a supplier that is already in, and writing roadmap disclosure into the contract as a term rather than a favour. Both are strategies. Discovering in 2029 that nobody in your supply chain had a seat is not.
Read next: Europe Paid Five Times the VCs in This Chip Deal | Chip Stocks Fell 20%. Your Compute Bill Did Not
