Generative AI has accelerated the discovery of new materials capable of removing PFAS (per- and polyfluoroalkyl substances), commonly called "forever chemicals," from drinking water. In a first-of-its-kind commercial partnership, Kemira, a global water treatment chemicals company, and CuspAI, an AI materials science startup, announced they designed over 5,000 novel material candidates in just six months, a timeline that would traditionally take years to achieve.

What Are Forever Chemicals and Why Do They Matter?

PFAS are synthetic chemicals that have been used in various industries since the 1940s. They earned the nickname "forever chemicals" because they do not break down in the environment and can accumulate in the human body over time. These persistent compounds are now found in drinking water worldwide and are subject to increasingly strict regulations, including new US EPA maximum contaminant limits announced in 2024 and the EU Drinking Water Directive.

The challenge of removing PFAS at trace concentrations, measured in parts per trillion, has made this a pressing environmental problem. Today's leading remediation technology relies on granular activated carbon, which Kemira already works with in its activated carbon regeneration business. The company initiated this AI collaboration to explore whether AI-driven design could create more selective and longer-lasting alternatives.

How Did AI Compress Years of Research Into Six Months?

The partnership leveraged generative AI to design entirely new metal-organic frameworks, or MOFs, from scratch. MOFs are nano-porous crystalline materials whose structure and chemistry can be precisely tuned for specific filtration and adsorption applications. The design space for possible MOFs is vast, estimated at hundreds of trillions of candidate structures, making traditional trial-and-error approaches impractical.

CuspAI's platform explored approximately 300 trillion possible material structures and delivered over 5,000 novel material designs with full property data for three priority PFAS molecules: GenX, PFBS, and PFOS. This massive search was narrowed to approximately 20 selected priority candidates now advancing to further development and testing.

"Kemira challenged us with finding new solutions to one of the most pressing environmental problems of our time, and in six months our partnership delivered," said Dr. Chad Edwards, CEO and Co-Founder of CuspAI.

Dr. Chad Edwards, CEO and Co-Founder of CuspAI

What Makes This Breakthrough Different From Previous AI Materials Research?

While academic groups and technology companies have explored AI for materials screening in the past, this is the first commercial partnership to apply generative AI end-to-end to design entirely new material structures from scratch against industrial performance criteria. Previous efforts typically focused on screening existing materials rather than generating novel ones at this scale and speed.

The collaboration also uncovered new functional group chemistries with potential for development into broader adsorption products, opening pathways for future applications beyond PFAS remediation. The success of this initial phase has already prompted further programs across additional material classes to be scoped under the partnership's framework agreement.

Steps to Understanding AI-Driven Materials Discovery

• Define the Problem: Kemira specified exact industrial requirements, including the ability to remove specific PFAS molecules at sub-parts-per-billion concentrations using water-stable, environmentally compatible, synthesizable, and cost-effective chemistry.
• Explore the Design Space: CuspAI's generative AI platform searched approximately 300 trillion possible material structures, a task that would be impossible for human researchers to complete manually.
• Evaluate and Narrow Candidates: The AI generated over 5,000 novel material designs with full property data, which were then evaluated against real industrial requirements and narrowed to approximately 20 priority candidates for further testing.
• Advance to Development: The selected candidates now move into the next phase of development and testing, where their performance will be validated in real-world conditions before potential commercialization.

"This project combined Kemira's deep expertise in water treatment chemistry with CuspAI's capabilities, and the results speak for themselves," stated Sampo Lahtinen, Executive Vice President, Research and Innovation at Kemira.

Sampo Lahtinen, Executive Vice President, Research and Innovation at Kemira

What's Next for AI-Designed Water Treatment Materials?

The partnership has demonstrated a credible path toward next-generation PFAS remediation products. Antti Salminen, President and CEO of Kemira, emphasized that the rigor applied to evaluating each material candidate against real industrial requirements gives the company confidence in advancing these candidates forward.

The success of this initial phase suggests that AI-driven materials discovery could accelerate solutions to other pressing environmental and industrial challenges. As regulatory pressure on PFAS intensifies globally, the ability to rapidly design and test new remediation materials could prove critical for water treatment companies seeking alternatives to traditional activated carbon approaches.

CuspAI, founded in 2024 by Professor Max Welling and Dr. Chad Edwards, has already raised over $200 million in Series A and Seed funding from investors including NEA, Temasek, NVIDIA, Samsung, and Hyundai. The company specializes in using generative AI and molecular simulation to accelerate the discovery of breakthrough materials for industrial applications across semiconductors, energy, and climate.