CuspAI’s $450M Bet: AI Materials Discovery Hits a $2.6B Valuation

A Fivefold Valuation Jump in Ten Months

CuspAI announced on July 20, 2026 that it closed a $450 million Series B, taking its valuation to $2.6 billion. The jump is unusually steep even by 2026 AI-funding standards: according to TechFundingNews, the company was valued at $520 million as recently as its Series A close in September 2025 — a nearly fivefold increase in roughly ten months. That Series A, more than $100 million, was co-led by New Enterprise Associates (NEA) and Temasek.

CuspAI was founded in 2024 by CEO Chad Edwards and Max Welling, a machine-learning researcher based in Amsterdam, according to Yahoo Finance’s coverage of the round. The company applies AI to predict how candidate materials will perform before anyone runs a physical experiment — its platform, MIRA, is described by Sifted as managing “the end-to-end discovery process, from designing candidate materials to simulation and experimental validation.” The pitch is that materials R&D — historically a multi-year, trial-and-error cycle for semiconductor and clean-energy components — can be compressed by having a model filter an enormous search space of candidate compounds down to the handful worth testing in a lab.

This year, semiconductors account for roughly 80% of CuspAI’s research bandwidth, per Yahoo Finance, with a specific push to eliminate rare metals such as ruthenium and iridium from chipmaking processes. That focus matters because both metals face supply concentration risk and price volatility, and a materials substitute that performs equivalently would directly de-risk a chunk of the semiconductor supply chain — precisely the kind of problem that has made “AI for science” attractive to strategic corporate investors over the past year.

The Round, the Foundry, and Who’s Actually an Investor

The Series B was led by Kleiner Perkins and NEA, with participation from Bezos Expeditions — the personal investment vehicle of Amazon founder Jeff Bezos — plus Glade Brook Capital Partners, Lux Capital, AMD Ventures, Tru Arrow Partners, StepStone, the UK’s Sovereign AI Venture Fund, and the Netherlands’ Invest-NL. Returning backers include Temasek, Basis Set Ventures, Giant Ventures, Touring Capital, Prosus, Phoenix Court, and Northzone, alongside a personal check from Kleiner Perkins chairman John Doerr.

It’s worth being precise about a distinction that several headlines blur: Nvidia and Samsung are not listed among the Series B’s disclosed financial backers. They are founding members of the separate “AI Materials Foundry” that CuspAI launched the same day — a coalition the company describes, per Sifted, as more than 45 organizations pooling compute, lab access, proprietary data, and scientific expertise. Alongside Nvidia and Samsung, the Foundry’s founding members include Meta’s Fundamental AI Research (FAIR) team, Hyundai Motor Group, Applied Materials, Tokyo Electron, Lam Research, Henkel, and Merck. That distinction matters for anyone reading the deal as “Nvidia and Samsung bet on CuspAI’s equity” — what they’ve actually committed to, based on public reporting, is participation in a shared materials-discovery network, not a disclosed cash stake in the company.

CuspAI’s advisory board has also grown to include AI research figures Yann LeCun and Geoffrey Hinton, plus former Apple and Microsoft AI executive John Giannandrea, according to Yahoo Finance. The company is expanding lab operations across Cambridge, Amsterdam, Berlin, Tokyo, Singapore, and the San Francisco Bay Area to support both the funding round’s growth targets and the Foundry’s multi-region structure.

What Founders and Investors Should Do

1. Separate “who invests” from “who joins the coalition” before you pitch the deal to your board

CuspAI’s own announcement blurs equity investors and Foundry members into one 45-plus-name list, and most press coverage repeated that framing uncritically. If you’re building an AI-for-science company, a hardware consortium, or any multi-party “network” alongside a funding round, structure the announcement — and your internal cap table story — so investors, customers, and strategic partners are never presented as interchangeable. Conflating them looks impressive for one news cycle and creates confusion the first time a journalist or acquirer asks who actually owns equity.

2. Price AI-for-science rounds on unit economics, not on the multiple alone

A near-5x valuation jump in ten months — $520 million to $2.6 billion — is a headline, not a due-diligence conclusion. Before benchmarking your own Series B ask against CuspAI’s multiple, verify what changed operationally between rounds: new commercial contracts, a working consortium of paying enterprise partners, or simply a hotter AI-for-science category. CuspAI’s jump coincided with landing dozens of named enterprise partners in its Foundry — that’s a revenue-pipeline signal investors can underwrite, unlike valuation momentum alone.

3. Recruit strategic corporate partners around a shared pain point, not a generic “AI platform” pitch

Nvidia, Samsung, Hyundai, and the materials/semiconductor names in CuspAI’s Foundry didn’t join because CuspAI pitched “AI for materials” abstractly — they joined around a specific, expensive problem (eliminating rare-metal dependency in chipmaking, per Yahoo Finance’s reporting on the 80%-semiconductor focus). If you’re recruiting strategic partners for a coalition, lead with the exact cost or supply-chain risk their engineering teams already lose sleep over, not a category-level thesis.

4. Treat advisory-board additions as a signal-stacking tool, not a vanity metric

CuspAI added Yann LeCun, Geoffrey Hinton, and John Giannandrea to its advisory board around the same funding cycle. For early-stage AI founders, a credible technical advisory board is one of the cheapest ways to reduce investor due-diligence friction on deep-tech claims that are otherwise hard for generalist VCs to evaluate — but it only works if the advisors have a genuine, checkable relationship to the technology, not just a marquee name attached after the fact.

Where This Fits in 2026’s Funding Cycle

CuspAI’s round is part of a broader pattern: 2026 has seen a wave of “AI for science” startups — covering materials, drug discovery, and climate modeling — attract valuations that would have been reserved for foundation-model labs two years ago. The distinguishing feature of CuspAI’s raise isn’t just the size of the check; it’s the parallel construction of an industry consortium that gives the company distribution and validation from hardware incumbents (Nvidia, Samsung, Applied Materials) without those incumbents necessarily writing equity checks. That’s a capital-efficient way to buy credibility, and it’s likely to be copied by the next wave of vertical AI-for-science startups trying to raise growth rounds without yet having enterprise revenue that matches their valuation. Whether the underlying science delivers commercially usable materials at the pace the funding implies is the open question the next 12-18 months will answer.

Frequently Asked Questions

Sources & Further Reading

• CuspAI lands $450m round to accelerate AI materials discovery — Sifted
• From $520M to $2.6B in 9 months: Bezos Expeditions co-leads CuspAI’s $450M Series B — Tech Funding News
• CuspAI raises $450 million Series B for AI materials discovery — Yahoo Finance
• CuspAI Raises $450M Series B — The SaaS News
• CuspAI secures Series A investment co-led by NEA and Temasek — UK Tech Investment News