CuspAI's $450M Series B: The Full Story Behind the AI Materials Startup That's Rewriting Chemistry

The Funding Round: From Seed to $2.6B in Two Years

CuspAI was founded in 2024 and raised a $30 million seed round in June of that year, led by Hoxton Ventures . By September 2025, it had closed a $100 million Series A co-led by NEA and Temasek, at a $520 million valuation, with participation from Nvidia's VC arm, Samsung Ventures, and angel investors including OpenAI co-founder Durk Kingma and Hugging Face co-founder Thomas Wolf .

The new $450 million Series B was confirmed on July 20, 2026, after earlier reports from the Financial Times in June indicated the company was in talks to raise roughly $400 million at a $2.6 billion valuation . The final round was larger than initially reported, and included a roster of new investors — AMD Ventures, Glade Brook Capital Partners, Lux Capital, and sovereign investors from the UK, Netherlands (Invest-NL), and Singapore — alongside the lead investors . The round was also reported at approximately $500 million in some outlets, though most sources cite the $450 million figure .

The AI Materials Foundry: A 48-Member Industrial Coalition

On the same day as the funding announcement, CuspAI launched the AI Materials Foundry, a global consortium designed to pool computing power, proprietary materials data, lab capacity, and manufacturing expertise to accelerate materials design . Founding members include Nvidia, Meta, Samsung, Hyundai Motor Group, Applied Materials, Lam Research, Tokyo Electron, ASMPT, 3M, Henkel, Hitachi, JSR, Merck, AMD, and dozens of others .

Nvidia provides accelerated computing infrastructure, while Meta contributes its Fundamental AI Research (FAIR) team's atomistic chemistry models . Member companies contribute proprietary materials data and lab capacity, and manufacturing requirements . The coalition's stated goal is to compress the traditional 10–20 year materials discovery timeline to six-month cycles .

How the Inverse-Design System (MIRA) Works

CuspAI's core platform, called MIRA, inverts the traditional materials discovery process . Rather than synthesizing compounds and then testing their properties, users specify desired performance characteristics — thermal tolerance, electrical conductivity, CO₂ selectivity, etch resistance — and the generative AI models propose candidate molecular structures optimized for those properties . Critically, the system also evaluates manufacturability, ensuring that proposed candidates can actually be synthesized .

The company describes MIRA as "a search engine for the material world" . The platform screens candidates against massive chemical space, then routes the top hits to coalition partners for rapid synthesis and real-world validation . NEA, a Series A and B investor, described the approach as "inverse design — starting with target properties and working backward to propose candidates — then evaluates stability, performance, and manufacturability through fast feedback loops" .

The Co-Founders

CuspAI was co-founded in 2024 by two individuals with complementary expertise :

• Dr. Chad Edwards — CEO. A chemist with industry experience in technology commercialization at Google, BASF, and most recently the quantum computing company Quantinuum . He also served as a commercial co-founder of Cambridge Quantum Computing .
• Professor Max Welling — Chief AI Officer. A renowned AI researcher, former distinguished scientist and vice president at Microsoft Research and Qualcomm, and a professor at the University of Amsterdam . Welling is a pioneer in modern AI and is also known for his work on graph neural networks and variational inference.

The company is headquartered in Cambridge, UK, with roots stretching back to Amsterdam, where Welling is based . Notable board members include Geoffrey Hinton ("Godfather of AI") as an advisor, Yann LeCun, and former ASML CTO Martin van den Brink .

Competitor Comparison: How CuspAI Stacks Up

CuspAI operates in a fast-growing space alongside other well-funded AI materials startups. Here is a comparison based on available information :

The key strategic difference: CuspAI does not build its own robotic labs. Instead, it outsources physical validation to coalition partners, which theoretically allows it to scale faster and tap into existing manufacturing expertise, but also means it cedes control of the synthesis loop . In contrast, both Lila Sciences and Periodic Labs operate fully autonomous wet labs — a more capital-intensive, end-to-end approach that potentially yields tighter prediction–validation feedback loops . Periodic Labs, co-founded by ChatGPT co-creator Liam Fedus and Google DeepMind scientist Ekin Doğuş Çubuk, has a narrower north star (a room-temperature superconductor) . Lila Sciences has the broadest scope, covering biology, chemistry, and materials .

The Scaling Question: Can Six-Month Cycles Survive?

The AI Materials Foundry's viability hinges on whether its six-month discovery-to-validation cycles hold up as the consortium scales to 48+ partners, each with proprietary data formats, conflicting IP interests, and different manufacturing constraints . The promise is that CuspAI's centralized MIRA platform can route the right candidate to the right partner efficiently, but the risks include coordination overhead, data-sharing friction, and the physical reality that synthesis and testing — no matter how smart the AI — remains slow, expensive, and noisy .

If the coalition cannot demonstrate repeated closed-loop breakthroughs within the six-month cadence, the model risks becoming a large-scale simulation engine rather than a true materials pipeline . Multiple sources note a critical caveat: "currently, no CuspAI initiative has yet produced a commercially deployed material" . For a company valued at $2.6 billion, that gap between promise and proof will be the defining question as the AI Materials Foundry moves from announcement to execution.