CuspAI: The Cambridge Startup Rewriting the Rules of Materials Discovery

For those tracking where European AI meets industrial reality, the conversation continues at Human x AI Europe on May 19 in Vienna, where the right people, the right room, and the right day converge.

The Funding Trajectory

Five days ago, Bloomberg reported that CuspAI is in discussions to raise at least $200 million in new funding. No term sheet has been signed, but the round would push the company's valuation above $1 billion, making it Europe's latest clean tech unicorn.

The numbers tell a compressed story. CuspAI raised a $30 million seed round in June 2024, one of Europe's largest that year. By September 2025, the company closed a $100 million Series A co-led by New Enterprise Associates (NEA) and Singapore sovereign wealth fund Temasek, valuing the company at $520 million. New commercial contracts subsequently pushed that valuation to around $800 million before the current round.

The investor roster reads like a who's who of strategic capital: NVentures (NVIDIA's venture arm), Samsung Ventures, Hyundai Motor Group, Prosus Ventures, Northzone, LocalGlobe, and Giant Ventures. Angel investors include Durk Kingma (OpenAI co-founder), Zoubin Ghahramani (Google DeepMind VP Research), and Thomas Wolf (Hugging Face founder).

The Founding Team

CuspAI was incorporated in March 2024 by two co-founders with complementary pedigrees.

Dr. Chad Edwards, CEO, is a chemist turned deep tech entrepreneur. He previously served as Commercial Co-Founder of Cambridge Quantum Computing (CQC) and later as VP of Strategic Partnerships at Quantinuum following CQC's merger with Honeywell. His background bridges laboratory science and commercial scaling.

Professor Max Welling, CTO, is a machine learning researcher at the University of Amsterdam and co-inventor of variational autoencoders (VAEs). He completed his PhD in theoretical physics under Nobel laureate Gerard 't Hooft, then trained under Geoffrey Hinton at the University of Toronto. Before CuspAI, Welling served as Distinguished Scientist and VP at Microsoft Research AI4Science and VP of Technology at Qualcomm. In 2025, he was elected to the Royal Netherlands Academy of Arts and Sciences.

The advisory board amplifies the scientific credibility. Geoffrey Hinton (Nobel laureate, Turing Award winner) and Yann LeCun (Turing Award winner, Meta's chief AI scientist) both serve as advisors. Lord John Browne (former BP CEO) and Martin van den Brink (former ASML President and CTO) joined in September 2025, adding industrial weight.

The Technical Proposition

CuspAI describes its platform as "a search engine for the material world." The core mechanism inverts traditional materials discovery. Rather than synthesizing compounds and testing their properties, users specify desired properties (thermal tolerance, conductivity, CO2 selectivity) and the system generates candidate molecular structures.

The platform combines generative AI models with physics-based molecular simulations. According to NEA's investment thesis, CuspAI's proprietary model MOFGEN achieves a 49% VUN (valid, unique, novel) rate for metal-organic frameworks (MOFs), compared to 10% for Microsoft's models and 16% for Meta's. The company claims its approach generates synthesizable candidates up to ten times faster than traditional methods.

The constraint that matters: these models are "synthesis-aware," meaning they produce materials that can actually be manufactured, not just simulated. This addresses a persistent failure mode in computational materials science, where promising candidates prove impossible to produce at scale.

Industrial Partnerships

CuspAI's commercial traction spans multiple sectors:

Carbon capture: The company partnered with Meta and Georgia Institute of Technology to create OpenDAC, the world's largest direct air capture database with over 100 million datapoints. The SkyVault project demonstrated carbon capture materials moving from generative design through synthesis and experimental validation in six months.

Automotive: Hyundai Motor Group announced a strategic partnership in November 2025 to develop materials for future mobility solutions, including battery efficiency and hydrogen fuel cells.

Water treatment: CuspAI partnered with Kemira, the Helsinki-listed chemicals company, to develop materials for removing PFAS (per- and polyfluoroalkyl substances, known as "forever chemicals") from water supplies.

Semiconductors: The company has undisclosed partnerships in the semiconductor sector, where materials constraints increasingly limit device scaling.

Competitive Landscape

CuspAI operates in a crowded field. Tracxn ranks the company third among 71 active competitors, behind Schrödinger (physics-based simulation for drug and materials discovery) and PhaseCraft (quantum simulation systems). PASQAL, the French neutral atom quantum computing company, also competes in materials simulation.

The competitive dynamics shifted in 2024-2025. Flagship Pioneering spun out Lila Sciences with a $200 million seed round. Periodic Labs, founded by ex-OpenAI and Google DeepMind staff, raised a $200 million seed round from Andreessen Horowitz at a $1 billion valuation. The capital intensity of AI-for-science is escalating.

CuspAI's differentiation rests on three pillars: proprietary datasets (particularly for MOFs), synthesis-aware model design, and closed-loop integration with experimental validation. The company's CB Insights profile notes access to the UK's Isambard supercomputer, enabling training of more powerful models than competitors without equivalent compute access.

The European Context

CuspAI is registered in the UK (Cambridge) with teams in Amsterdam, Berlin, and Tokyo. The company is opening a London Kings Cross office in 2026. Sifted ranked CuspAI first on its AI 100 list of rising European AI companies.

The UK government's AI for Science strategy named CuspAI as one of the frontier AI companies receiving significant access to national compute infrastructure. This positions the company within the UK's broader industrial strategy for AI, though the commercial partnerships span global markets.

For European policymakers, CuspAI represents a test case: can a UK-based AI company with European research roots compete at the frontier of AI-for-science while maintaining its headquarters in Europe? The answer depends partly on whether the next funding round comes from European or American capital, and whether the company's expansion prioritizes US and Asian markets over European deployment.

What to Watch

The reported $200 million round, if closed, would mark CuspAI's transition from well-funded startup to unicorn. The company's 2025 year-end review claimed both team size and contract value grew twentyfold in 21 months.

Three questions will determine whether this trajectory continues:

First, can the platform's speed advantage translate into durable commercial relationships? Six-month discovery cycles matter only if the resulting materials survive manufacturing constraints and regulatory approval.

Second, how will the semiconductor partnerships develop? Materials innovation for advanced chips represents a larger addressable market than carbon capture, but also involves longer validation cycles and more demanding customers.

Third, will European compute infrastructure keep pace? CuspAI's access to Isambard provides a temporary advantage, but US competitors with access to hyperscaler compute may close the gap.