CuspAI

Cambridge AI-materials startup — generative models + molecular simulation to discover new materials; $450M Series B at $2.6B [1][2].

1. Core Product / Service

CuspAI (founded 2024, Cambridge, UK) applies generative AI and molecular simulation to design new materials from scratch [2]. Its platform, MIRA, runs the full discovery cycle — "from generative materials design through simulation, synthesis-route planning and coordinated experimental validation" — and CuspAI reports it can optimize molecular compositions up to 10× faster than traditional R&D [1][2].

Target sectors are semiconductors, clean energy, carbon capture, and advanced manufacturing, with an explicit goal of reducing or eliminating reliance on scarce metals (iridium, ruthenium) in chip supply chains [2]. Alongside its Series B, CuspAI launched the AI Materials Foundry — a coalition of 45+ companies, labs, and compute partners (including Nvidia, Meta, and Hyundai) pooling compute and scientific expertise for materials discovery [3].

2. Target Users & Pain Points

• Materials-dependent industrials (chipmakers, battery/solar manufacturers, chemical majors) needing new materials faster than empirical R&D allows.
• Frontier AI labs and compute partners joining the Foundry to co-develop materials plus the compute stack that designs them.

Pain solved: industrial progress is bottlenecked on "materials that don't yet exist" — the founders' stated thesis is that without faster materials discovery, the next 50 years of progress stall [2].

3. Competitive Landscape

CuspAI's differentiator is pairing generative materials design with a cross-industry compute coalition (the Foundry) rather than owning wet-lab robotics itself [3].

4. Unique Observations

• "Cusp inside" as the endgame. Max Welling's stated vision is CuspAI as "an autonomous search engine for materials" — the brand-equivalent of "Intel inside" for chips, batteries, and solar cells [3]. That reframes the company as an IP/licensing layer over the materials economy, not just a services lab.
• Sovereign + personal AI royalty stack. The round pulled in the UK government's Sovereign AI Venture Fund and Bezos Expeditions alongside conventional VCs, with Yann LeCun and Geoffrey Hinton on the advisory board — a rare confluence of state, big-tech-figure, and academic legitimacy for an AI4S company [2][3].
• Valuation velocity mirrors lila-sciences. From a $520M valuation (Sept 2025) to $2.6B (mid-2026) in under a year, CuspAI is another AI4S name repricing on potential rather than revenue — consistent with the 2026 AI4S funding wave documented in ai-for-science [1][2].

5. Financials / Funding

• Series A (2025): >$100M; valuation $520M (Sept 2025) [3].
• Series B (2026): $450M at a $2.6B valuation, led by Kleiner Perkins and NEA, with Bezos Expeditions, the UK government's Sovereign AI Venture Fund, Glade Brook Capital Partners, Lux Capital, AMD Ventures, Tru Arrow Partners, StepStone, Invest-NL, and John Doerr; returning investors Temasek and Northzone [1][2][3].
• Use of funds: HPC infrastructure, research expansion, international growth (new Singapore office; teams in Cambridge, Amsterdam, Berlin, Tokyo, and the US) [2].

6. People & Relationships

• Co-founders: Max Welling (professor; ex-Microsoft Research / Qualcomm AI, Amsterdam) and Chad Edwards (CEO) [2][3].
• US operations: John Giannandrea (former Apple SVP / Google AI lead) hired to establish US presence [2].
• Advisory board: Yann LeCun, Geoffrey Hinton [2].
• Foundry partners: Nvidia, Meta, Hyundai, plus 45+ companies and labs [3].
• Peers: orbital-materials, radical-ai, periodic-labs, lila-sciences.