The Cambridge startup closed a $450 million Series B on July 21, 2026, valuing it at $2.6 billion and pulling in Bezos Expeditions, Kleiner Perkins, Nvidia, and Meta as backers of its AI Materials Foundry.
Nine months ago, CuspAI was worth $520 million. It's now worth $2.6 billion. That kind of jump usually signals a company that has found a problem worth solving at exactly the right moment, and in CuspAI's case, that problem is one the entire semiconductor industry is quietly losing sleep over: the physical materials that chips are made from are running out of room to improve.
The Series B, led by Kleiner Perkins and NEA and co-led by Bezos Expeditions, brings CuspAI's total funding past $650 million. The investor list reads like a who's who of people who believe the next hardware bottleneck is a chemistry problem: AMD Ventures, Glade Brook Capital, Lux Capital, StepStone, Temasek, Prosus, the UK's Sovereign AI Venture Fund, and Invest-NL from the Netherlands all joined the round. John Doerr signed on personally. When that many serious people write checks in the same direction, it's worth paying attention to why.
Founded in 2024 by Dr. Chad Edwards and Professor Max Welling, CuspAI runs an agentic AI platform called MIRA that models how new materials would behave before anyone spends months testing them in a lab. The pitch is simple in theory and staggering in practice: instead of chemists grinding through thousands of candidate compounds over years, MIRA compresses that search into months by simulating material properties at scale. The company is directing roughly 80% of its 2026 efforts at semiconductors specifically, and one of its stated goals is reducing chipmakers' dependence on rare metals like ruthenium and iridium, both of which carry real supply-chain risk.
The launch product is an AI Materials Foundry, a network that links corporate research labs, academic institutions, and high-performance computing clusters through CuspAI's platform. More than 48 organizations have already signed on as founding members. Nvidia, Meta's Fundamental AI Research team, Samsung, Hyundai Motor Group, Applied Materials, Tokyo Electron, and Lam Research are among them. That coalition is not window dressing. Getting chip equipment giants like Lam Research and Tokyo Electron in the same consortium as Meta's research arm tells you that this isn't a startup pitching to industry from the outside. It's being invited in.
Frankly, the materials science bet is overdue. The semiconductor industry has spent the better part of a decade squeezing more performance out of the same basic toolkit, and the marginal gains are getting harder to find. Finding new materials that conduct, insulate, or switch more efficiently than existing ones could do more for chip performance than any architectural trick, but the traditional discovery process is slow and expensive. The entire value proposition of CuspAI is that AI can search that space far faster than humans working in wet labs.
Why the timing matters
The funding lands at a moment when governments and chipmakers alike are treating supply chains as a national security issue. Not merely a procurement problem. The UK and Dutch sovereign funds participating here aren't just writing checks out of enthusiasm for the science - they're hedging against a world in which the materials critical to chip manufacturing are concentrated in a handful of geographies, and the political cost of that concentration is becoming impossible to ignore. CuspAI's Cambridge base helps, and its new offices planned for Singapore, Amsterdam, Berlin, Tokyo, and the US give it an unusual spread for a company that's barely two years old.
The valuation jump from $520 million to $2.6 billion in nine months does deserve scrutiny. CuspAI hasn't shipped a chip, hasn't announced commercial revenue figures, and the AI materials discovery space is still nascent enough that no one can point to a proven case of AI-discovered materials reaching high-volume chip production. What investors are paying for is the potential to compress timelines, and the cost of being wrong is steep at a $2.6 billion price tag.
Still, the coalition CuspAI has assembled is a stronger signal than most early-stage rounds produce. When Nvidia and Meta's research lab join as foundry partners rather than just writing a check, they're betting their own R&D time on the platform. That's a different kind of endorsement, and it suggests CuspAI isn't just raising capital. It's becoming infrastructure.
Also read: Synthesia turns its AI avatars into performance coaches with Roleplay Sessions • Omen AI raises $31M to detect the bacterial threat quietly killing GPU clusters • AI data center debt has climbed to the top of Wall Street's credit risk watchlist
