Buyer Chains · CASE STUDY 001 · Discovery stage
CuspAI × Kemira: 300 trillion structures down to 20 candidates, and not yet an experiment
The question · From the buyer's brief to 20 AI-designed candidates: who decides which experiment gets paid for next?
At a glance · About 300 trillion structures, 5,000-plus designs, about 20 priority candidates; synthesised or tested: none.
One of the Buyer Chains series. Each piece follows one chain, the AI company, its customer and sometimes the customer's customer, through five roles in the public record: who states the need, who sets the research direction, who runs the experiments, who validates the result, who decides to keep paying. It records only what the public record states and says who said each thing; a missing step is marked not yet public rather than filled in, and a company's own figure is written as the company's. The timeline at the end stays open: when the chain's next step reaches the public record, it goes on as a new line.
Kemira is a Finnish water-treatment chemicals company, EUR 2.8B revenue in 2025, already in the activated-carbon regeneration business. In September 2025 it signed a partnership with Cambridge-based CuspAI to find materials that filter PFAS, the "forever chemicals", out of water.[6] On 21 May 2026 Kemira announced the result of the discovery phase: CuspAI's platform designed metal-organic frameworks from scratch across a space of about 300 trillion possible structures, delivered over 5,000 novel designs with full property data for three target molecules, GenX, PFBS and PFOS, narrowed to approximately 20 priority candidates "now advancing to further development", in six months.[7] That is a clear discovery-phase result and only that: the announcement records no candidate synthesised or tested.
Who states the need · Kemira. The release says Kemira defined the brief: remove specific PFAS molecules at sub-parts-per-billion concentrations, with chemistry that is water-stable, environmentally compatible, synthesisable and cost-effective.[7]
Who sets the direction · Both. Kemira set the criteria; CuspAI chose MOFs and generative design; Kemira's R&D head Sampo Lahtinen says "every material candidate was evaluated against real industrial requirements".[7]
Who runs the experiments · Not yet public. The 5,000 designs carry computed property data; synthesis and adsorption testing belong to the next phase, and the release does not say who runs it (CuspAI, Kemira's labs or a third party).
Who validates · Not yet public. The next phase of "development and testing" is the first validation; Kemira's CEO describes the outcome so far as "a credible path toward a next-generation PFAS remediation product", not a product.[7]
Who decides to keep paying · Kemira: the next phase is under way and further programmes in other material classes are being scoped under the framework agreement; the financial terms have never been public.[7] On the other side, CuspAI's investors ($450M Series B, July 2026)[8]; Kemira is also a founding member of the AI Materials Foundry announced in October 2026, a list on which nobody commits to buy anything.[9]
What the buyer looks at, concretely
- Kemira's own written bar: effective against GenX, PFBS and PFOS at sub-ppb (the US EPA's 2024 drinking-water limits are in parts per trillion); water-stable; synthesisable; cost-effective.[7]
- The incumbent it wants to replace is granular activated carbon, and Kemira says it is looking for "more selective and even longer-lasting" alternatives. So the real acceptance comparison is adsorption capacity, selectivity and service life against GAC; the release gives none of those three numbers for any candidate.[7]
- What the next phase has to produce: at least one candidate synthesised, its removal rate in real water, its regeneration cycles and cost.
Not yet public · Any experimental data; the deal's money and revenue split; who owns the 20 candidates; whether any third party will test them.
Decision point · The next decision is Kemira's R&D department's: which of the 20 candidates get synthesised first, who runs that work, and what it costs. The release says the next phase is under way and answers none of the three.
The chain's timeline
- 2025-09CuspAI and Kemira sign a framework partnership on PFAS-filtering materials (Fortune)[6]
- 2026-05Discovery milestone: about 300 trillion structures, 5,000-plus designs, about 20 priority candidates (Kemira release)[7]
- 2026-10Kemira becomes a founding member of CuspAI's AI Materials Foundry (Business Wire)[9]
- NextExperimental validation: synthesis, adsorption performance, scale-up. Awaiting public data.
- LaterIndustrial qualification: performance and cost against GAC, customer trials, adoption. Awaiting public data.
Sources
- Fortune (via Yahoo Finance), report on CuspAI's $100M Series A, which mentions the Kemira partnership, 2025-09. finance.yahoo.com ↗ (the Kemira passage read)
- Kemira Oyj press release, "New AI-Designed Materials Show Promising Potential to Remove 'Forever Chemicals' from Drinking Water in Industry-First Breakthrough", 2026-05-21. kemira.com ↗ (read in full); timeline entry kemira-cuspai-pfas-20-priority-candidates (added 2026-11-03, graded 2).
- Timeline entry cuspai-series-b-450m-ai-materials-foundry, 2026-07-20.
- CuspAI's announcement of the AI Materials Foundry's founding members (Business Wire, 2026-10; read earlier for the record).