CuspAI

CuspAI is a British artificial intelligence company headquartered in Cambridge, United Kingdom, focused on generative AI for materials science. The company was founded in 2024 by Chad Edwards (a chemistry PhD and former Quantinuum executive) and Max Welling (a Dutch machine-learning researcher and former vice president of technology at Qualcomm). CuspAI develops generative foundation models for the design of new chemical compounds and engineered materials, with primary applications in carbon capture, water purification, and adjacent industrial chemistry. As of May 2026, CuspAI is one of the principal AI-for-materials startups in Europe, with a $100 million Series A in September 2025 at a $520 million valuation co-led by New Enterprise Associates and Temasek, and reported follow-on discussions targeting a unicorn-class valuation.

At a glance

• Founded: 2024 in Cambridge, United Kingdom, by Chad Edwards and Max Welling. Additional offices in Amsterdam, Berlin, and Tokyo.
• Status: Private. Series A in September 2025 at a $520 million post-money valuation. Reported follow-on discussions in 2026 targeting a unicorn-class valuation, per Yahoo Finance / Sky News coverage.
• Funding: Approximately $130 million cumulative across two disclosed rounds. $30 million seed (June 2024) led by Hoxton Ventures, Basis Set Ventures, and Lightspeed. $100 million Series A (September 2025) co-led by New Enterprise Associates and Temasek.
• CEO: Chad Edwards, Co-Founder and Chief Executive Officer. Chemistry PhD; former senior executive at quantum-computing company Quantinuum.
• Other notable leadership: Max Welling, Co-Founder and Chief Scientist. Professor of Machine Learning at the University of Amsterdam; former Distinguished Scientist at Microsoft Research and former vice president of technology at Qualcomm.
• Open weights: Limited. CuspAI develops commercial generative-materials models accessed through customer partnerships rather than as standalone open releases.
• Flagship products: MOFGEN (a generative model for metal-organic frameworks); the broader CuspAI search-engine-for-materials platform serving Hyundai, Meta, Kemira, and other industrial customers.

Origins

CuspAI was founded in 2024 by Chad Edwards and Max Welling. Edwards had spent the prior decade in deep-technology operating roles, most recently as a senior executive at the quantum-computing company Quantinuum. Welling had been a long-standing leader in machine-learning research, with academic appointments at the University of Amsterdam, prior research-lab roles at Microsoft Research and Qualcomm, and a record of advisory and technical work in generative modeling and equivariant neural networks.

The founding thesis combined Edwards's deep-technology operating experience with Welling's research direction in geometric deep learning and generative models. The premise was that materials discovery, like drug discovery before it, sits in a structural position where laboratory cycles run on multi-year timescales and computational generative methods could compress those cycles by orders of magnitude. Materials applications cited in the company's framing range from carbon capture to semiconductor manufacturing to water purification to energy storage, with each category representing both a commercial customer base and a public-policy demand for accelerated discovery.

The June 2024 seed round of $30 million, led by Hoxton Ventures with Basis Set Ventures and Lightspeed, established the initial capitalization. The September 2025 Series A of $100 million at a $520 million valuation, co-led by New Enterprise Associates and Temasek, was the company's most consequential public-facing transition and supported the expansion of its model platform and customer-engagement programs.

In parallel, CuspAI assembled a scientific advisory board including Geoffrey Hinton, Yann LeCun, Kristin Persson (the Lawrence Berkeley scientist behind the Materials Project database), and Verity Harding. The advisory composition has been characterized in industry coverage as one of the company's principal credibility signals, given the breadth of senior AI and materials-science figures involved.

Mission and strategy

CuspAI's stated mission is to build a generative AI search engine for materials. The strategic premise is that customers can specify the precise physical and chemical properties they require, and the platform produces synthesizable candidates approximately 10x faster than traditional materials-discovery methods.

The strategy combines three threads. First, foundation-model research on generative architectures specialized for chemistry and materials science, with the MOFGEN model for metal-organic frameworks as the early flagship. Second, industrial customer partnerships in which CuspAI's models are applied to specific materials-discovery problems with target chemistry partners. Third, scientific collaborations with research organizations including Meta and the Georgia Institute of Technology, which produce open datasets such as OpenDAC for direct air capture.

The competitive premise reflects CuspAI's positioning at the intersection of frontier generative-model research credentials (Welling's research and the advisory board) and industrial-chemistry application focus (Edwards's operating background). The company's reported MOFGEN benchmarks, which characterized the model as achieving 49 percent valid-unique-novel candidate generation rates against benchmarks of 10 percent for Microsoft's models and 16 percent for Meta's, have been the most commonly cited capability signal.

Distribution channels include direct industrial-customer partnerships, sector-specific commercial-development programs, and continued participation in scientific data-release collaborations.

Pipeline and programs

• MOFGEN. Generative model for metal-organic frameworks. Reported to achieve a 49 percent valid-unique-novel candidate generation rate, compared to 10 percent for prior Microsoft research models and 16 percent for prior Meta models. The model emphasizes synthesis-aware generation, producing candidates that can be physically manufactured rather than just simulated.
• Carbon-capture program. Joint work with Meta and Georgia Institute of Technology on OpenDAC, characterized in coverage as the world's largest direct air capture database with more than 100 million data points. The SkyVault project demonstrated end-to-end progression from generative design through laboratory synthesis and experimental validation in approximately six months.
• Hyundai partnership. Multi-year collaboration on sustainable-energy applications, including battery and energy-storage materials.
• Kemira partnership. Collaboration with the Helsinki-listed chemicals company on materials for removal of PFAS "forever chemicals" from water systems.

Distribution channels include direct industrial partnerships, scientific data-release collaborations, and customer-engagement programs across sustainable energy, semiconductors, climate, and water sectors.

Benchmarks and standing

CuspAI's evaluation framework focuses on materials-discovery-specific metrics: generation-rate measures (valid, unique, novel candidates per unit of compute), synthesis success rates (the fraction of generated candidates that can be physically produced), and target-property accuracy (how closely synthesized materials match the specified property profile). Standard horizontal-LLM benchmarks (Artificial Analysis Intelligence Index, LMArena, GPQA Diamond) do not apply to the materials-design domain.

The MOFGEN benchmarks reported by CuspAI and characterized in NEA's investment thesis indicate a meaningful capability lead over peer research efforts at Microsoft and Meta, though the comparison points are publicly disclosed research models rather than commercial offerings. Industry coverage has consistently characterized the MOFGEN result as one of the more credible quantitative capability signals among 2024 to 2025-vintage AI-for-materials startups.

The OpenDAC and SkyVault end-to-end synthesis demonstrations have been characterized as the principal proof points that CuspAI's generative outputs translate to manufacturable materials, addressing a persistent failure mode in computational materials science where promising candidates prove impossible to produce at scale.

Leadership

As of May 2026, CuspAI's senior leadership includes:

• Chad Edwards, Co-Founder and Chief Executive Officer. Chemistry PhD; former senior executive at Quantinuum.
• Max Welling, Co-Founder and Chief Scientist. Professor of Machine Learning at the University of Amsterdam; former Distinguished Scientist at Microsoft Research and former vice president of technology at Qualcomm.
• Senior research and engineering staff across the MOFGEN, partnerships, and customer-engagement lines.

The advisory board, which includes Geoffrey Hinton and Yann LeCun, has been characterized in industry coverage as one of the principal recruiting and credibility signals for the company.

Funding and backers

• Seed (June 2024): $30 million led by Hoxton Ventures, Basis Set Ventures, and Lightspeed.
• Series A (September 2025): $100 million co-led by New Enterprise Associates and Temasek, at a $520 million post-money valuation.

Cumulative capital approximately $130 million across the two disclosed rounds. Industry coverage in early 2026 has reported follow-on discussions targeting a unicorn-class valuation, with a potential $200 million round in negotiation, although terms had not been confirmed at the time of the most recent reports.

Industry position

CuspAI occupies a structurally distinctive position among 2024-vintage AI-for-science insurgent labs. The combination of European founding base, materials-science vertical focus (rather than the more crowded biology-AI segment), industrial-customer partnerships across multiple sectors, and the senior advisory board produces a profile differentiated from the more biology-focused peers in the AI-for-science cohort.

Strategic risks include the open question of whether materials discovery proves to be a separable commercial market with the unit economics that biology-AI customers have demonstrated, the competitive pressure from larger labs (Google DeepMind on GNoME, Meta on its FAIR Chemistry initiative) that have published peer materials-AI research, and the dependence on industrial-customer partnership cycles that run on enterprise rather than software-as-a-service timeframes.

Strategic strengths include Welling's research credibility and the advisory-board composition, the MOFGEN benchmarks that distinguish CuspAI from peers without published quantitative results, the partnerships with Hyundai, Meta, and Kemira that establish multi-sector reference customers, and the geographic diversification across Cambridge, Amsterdam, Berlin, and Tokyo.

Competitive landscape

CuspAI competes with several AI-for-science and materials-AI peers across research and commercial dimensions:

• Periodic Labs. Autonomous-laboratory AI-for-science startup with materials-discovery focus and superconductor-research lead application. Direct application-overlap peer.
• Lila Sciences. Autonomous-laboratory AI-for-science peer with broader biology focus.
• Google DeepMind. GNoME (Graph Networks for Materials Exploration) materials-AI research line. Research peer with comparable computational coverage but no commercial product.
• Meta AI / FAIR. FAIR Chemistry research initiative including OpenDAC collaboration. Strategic partner and research peer.
• Microsoft Research. Materials-AI research initiatives including MatterGen and MatterSim. Research peer.
• Citrine Informatics, Schrödinger, Materials Project. Established commercial and research peers in materials informatics with longer commercial track records.

Outlook

• The progression of the reported $200 million 2026 funding discussions toward a unicorn valuation.
• The expansion of industrial-partnership reference customers beyond Hyundai, Meta, and Kemira.
• The cadence of new model releases beyond MOFGEN, including potential extensions to semiconductor and energy-storage chemistry.
• The translation of OpenDAC and SkyVault synthesis demonstrations into commercial-scale carbon-capture deployments.
• The competitive dynamics with Google DeepMind GNoME, Meta FAIR Chemistry, and Microsoft Research's materials initiatives across 2026 and 2027.
• The trajectory of senior-talent recruitment leveraging the Welling and advisory-board credentials.

Sources

• CuspAI official site. Company reference.
• Fortune: CuspAI raises $100 million Series A at $520 million valuation. Series A funding context.
• SiliconANGLE: CuspAI raises $100M to build AI search engine to transform materials science. Capability and partnership context.
• Hyundai Motor Group partnership announcement. Sustainable-energy collaboration.
• Tech Funding News: AI search engine for new materials nears $200M raise. Reported 2026 follow-on discussions.
• Sourcery VC: CuspAI investment thesis. MOFGEN benchmarks and competitive positioning.
• HumanX: CuspAI startup analysis. Company background and advisory-board composition.