CuspAI: Company Report

CuspAI is a two-year-old Cambridge, UK-based startup building generative AI software to discover new physical materials — what it calls a "search engine for materials." On July 20, 2026, it closed a $450 million ($335–338 million) Series B round at a $2.6 billion (£1.95 billion) valuation, roughly five times its valuation of just ten months earlier. The round was co-led by Kleiner Perkins and NEA, with participation from Jeff Bezos's family office, Bezos Expeditions, and the UK government's Sovereign AI Venture Fund — investing alongside each other in the same round, which is itself part of the story. Alongside the raise, CuspAI launched the "AI Materials Foundry," a coalition of more than 45 organizations including Nvidia, Meta, Samsung, and Hyundai.

Founders and Origins

CuspAI was founded in 2024 (seed round announced June 2024) by:

Dr. Chad Edwards (CEO) — a chemist with a background in deep-tech commercialization at Google and BASF, and most recently a senior executive at quantum-computing company Quantinuum.

Prof. Max Welling (Chief Scientist/CTO) — a Dutch machine-learning researcher, professor at the University of Amsterdam, and former Distinguished Scientist and VP of Technology at Microsoft Research and Qualcomm respectively. Welling is regarded as one of the more heavily cited researchers in modern AI.

The company's premise inverts the traditional materials-science workflow: rather than synthesizing a candidate material and then testing its properties, CuspAI's models take a target set of properties as input — e.g., "a material that selectively binds carbon dioxide under given conditions" — and generate and evaluate candidate molecular structures computationally, before physical validation. Its flagship generative model, referred to as MOFGEN, focuses on metal-organic frameworks; the broader platform combines generative AI with computational/physics-based simulation and synthesis-route planning.

Geoffrey Hinton and Yann LeCun both serve as advisors to the company — Hinton since the 2024 seed round.

Funding History

Round Date Amount Valuation Lead investors

Seed June 2024 $30 million — Hoxton Ventures, with Basis Set Ventures, Lightspeed, LocalGlobe, Northzone

Series A September 2025 ~$100 million ~$520 million New Enterprise Associates (NEA), Temasek

Series B July 20, 2026 $450 million $2.6 billion Kleiner Perkins, NEA

Total raised to date: more than $670 million (£500 million+) in just over two years.

The Series B's other participants included Glade Brook Capital Partners, Lux Capital, AMD Ventures, Tru Arrow Partners, StepStone, the Netherlands' Invest-NL, and John Doerr personally, plus existing backers Temasek, Basis Set Ventures, Giant Ventures, Touring Capital, Prosus, Phoenix Court, and Northzone returning. The UK's Sovereign AI Venture Fund's own stake size wasn't disclosed, though the fund typically writes checks in the £1–10 million range — small relative to the round, but notable for being there at all, discussed further below.

What the Money Is For

CuspAI says the capital will fund international expansion: a new Singapore office, and larger teams across its existing hubs in Cambridge, Amsterdam, Berlin, and Tokyo, plus a US buildout. It has hired John Giannandrea, the former Apple and Google AI executive, to help establish its US operations, and named Abhi Talwalkar — an AMD board member and chairman of Lam Research — to its advisory board, adding semiconductor-industry expertise to a board that already includes Hinton and LeCun.

The AI Materials Foundry

Launched simultaneously with the Series B, the AI Materials Foundry is CuspAI's attempt to build an industry-wide coalition around materials discovery — described as a global network of data, compute, labs, and scientific expertise, orchestrated through CuspAI's platform. Reported founding-member counts range from "more than 45" to "48" organizations depending on the source, and include:

Nvidia, providing compute infrastructure

Meta's Fundamental AI Research (FAIR) team, contributing its Universal Model for Atoms (UMA), a frontier atomistic-chemistry model

Samsung, Hyundai Motor Group, Applied Materials, Tokyo Electron, and Lam Research, among the industrial/semiconductor-supply-chain members

CuspAI CEO Chad Edwards framed the initiative starkly: "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."

The Problem CuspAI Says It's Solving

CuspAI's pitch centers on reducing chipmakers' and other industries' dependence on rare, expensive, hard-to-source metals — iridium and ruthenium are the two most frequently cited examples — by using AI to identify substitute or synthetic materials faster than traditional trial-and-error chemistry allows. Beyond semiconductors, the company points to potential applications in batteries, clean water treatment, carbon capture, and broader energy infrastructure. Its commercial traction so far includes named partnerships with Hyundai (sustainable energy and mobility materials), Meta (carbon-capture-related work), Kemira (removing PFAS "forever chemicals" from water), and ASML as a customer of its materials-discovery platform.

The UK Government Angle

The UK's participation comes through its Sovereign AI Venture Fund, a vehicle sitting within the Department for Science, Innovation and Technology (DSIT), created to invest in early-stage domestic AI companies. UK Secretary of State for Science and Technology Liz Kendall pointed to AI's potential application to energy, climate, and economic-growth challenges in comments on the deal; the government's broader interest is that advanced-materials breakthroughs could strengthen strategic UK industry and reduce reliance on external supply chains. Coverage of the round has noted some irony in the timing: the raise landed in the same period that the UK's incoming prime minister was reportedly considering abolishing DSIT itself — a detail worth flagging as reported political context rather than a settled fact, since government departmental structures are subject to change and this should be verified against the latest news if it matters to your purposes.

Commentators have also noted that the fund's presence alongside Bezos, rather than in place of him, is arguably the more interesting signal: a relatively small sovereign check riding alongside a major US financier and Silicon Valley VCs, in a company that is nonetheless very much the profile the fund was designed to back — Cambridge-founded, deep-technical, already under contract with multinational manufacturers.

Company Scale and Competitive Position

Headcount estimates vary by source and date, ranging from roughly 25 employees around late 2025 to approximately 75 by mid-2026 — consistent with a company mid-scale-up. Named competitors in the AI-for-materials-discovery space include Schrödinger, PhaseCraft, and PASQAL, though these companies differ somewhat in focus (Schrödinger leans more pharma/chemistry software, PhaseCraft and PASQAL more quantum-computing-adjacent).

Assessment

The headline number here is the valuation trajectory: from $520 million in September 2025 to $2.6 billion in July 2026 — a roughly 5x jump in ten months, on a company that's only been operating for about two years. That kind of repricing reflects a broader 2026 investor pattern of chasing AI applied to "physical world" problems (materials, chips, energy) rather than pure software, and Bezos's parallel, much larger bet on AI manufacturing (a separate startup, Prometheus, reportedly raised $12 billion at a $41 billion valuation around the same period) suggests he sees this as a broader thesis, not a one-off. The composition of CuspAI's investor base — chip-industry money (AMD Ventures), sovereign wealth (Temasek, the UK fund), and blue-chip US VCs all in the same round — mirrors the same "sovereignty plus commercial validation" pattern seen elsewhere in deep tech funding this year. Whether the underlying technology can actually deliver materials discovery timelines compressed from years to months at industrial scale is the open question a valuation this size is betting on; the commercial partnerships announced so far (Hyundai, Meta, Kemira, ASML) are a genuine signal of customer interest, but none of the public reporting reviewed here describes a fully validated, at-scale industrial deployment yet — worth keeping in mind as the more speculative part of the story relative to the funding facts themselves, which are well-documented.

Sources: Reuters, Tech.eu, Yahoo Finance/Bloomberg, TechCrunch, EU-Startups, Dealroom, Tracxn, MLQ News, Pulse2, Digit.fyi, Startup Fortune, Results Sense, Giant Ventures and Northzone (investor portfolio write-ups).