The secret ingredient keeping AI running: advanced materials science
Chips and software grab the headlines, but a quieter race is under way in the labs developing the physical materials that let AI hardware exist at all.
Key points
- Specialty materials, including perfluoroelastomers (heat-resistant seals inside chip-making machines) and heat-transfer fluids, set hard physical limits on how powerful AI chips can become.
- Belgian chemicals company Syensqo has developed a next-generation perfluoroelastomer made without fluorosurfactants, a class of long-lasting compounds that are difficult to break down in the environment.
- Syensqo uses Microsoft Discovery, an AI-assisted research platform, to identify candidate molecules for new cooling fluids faster than traditional lab methods allow.
- Cooling techniques borrowed from electric-vehicle engineering are being adapted for high-density AI server racks, which generate intense heat in a small space.
- New materials can take several years to qualify for use in commercial chip-making equipment, so reliability comes before novelty.
Every time a new AI model launches, the conversation turns to algorithms and the billions spent on new chips. Rarely does it turn to the seals inside the machines that print those chips, or the fluids that stop them from overheating. That gap is worth closing.
Manufacturing a modern semiconductor chip requires thousands of tightly controlled steps. Tiny temperature swings or chemically unstable materials create microscopic defects, cutting the number of working chips per wafer and pushing costs up. Each new chip generation packs in more transistors, so those tolerances tighten further.
Perfluoroelastomers sit at the sharp end of this problem. These dense, flexible seals hold the reactive gases and plasma inside chip-fabrication equipment and must survive conditions that would destroy most industrial materials. Syensqo, a specialty chemicals company spun out of Solvay in 2023, now makes a version using a manufacturing process free of fluorosurfactants. The company says the new material performs better and is produced more responsibly, without forcing chip makers to choose between the two.
Does this affect ordinary AI users?
Yes, indirectly but concretely. When materials fall short, chip yields drop and hardware costs climb with them, which pushes up the price of running AI services. Better materials mean more reliable chips, more affordable cloud computing, and AI tools that reach more people.
Data centres running large AI workloads now pack computing hardware so densely that air cooling can't keep up. Higher-voltage power systems are becoming standard, bringing fresh materials challenges around electrical insulation and heat management.
Syensqo says it's drawing on work already done for electric vehicles to address this. Cooling fluid systems designed for EV battery packs share enough engineering DNA with liquid-cooled AI server racks that lessons transfer across, cutting development time. Two weeks before this story, we reported that Cambridge-based CuspAI raised $450 million to build exactly this kind of AI-accelerated materials search, a sign that the funding world is catching up to what chemists have known for years.
How is AI changing the materials discovery process?
Traditionally, chemists would form a hypothesis, synthesise a compound, run tests, then start over, a cycle lasting months per candidate. Syensqo now uses Microsoft Discovery to screen molecular candidates for new heat-transfer fluids early in the process. The AI doesn't replace the chemists. It narrows the field so they spend lab time on the most promising options rather than on long-shots.
None of this moves quickly. A material identified today could take several years to qualify for use in commercial chip-making equipment, and manufacturers change materials only when a new one genuinely solves a problem.
One thing to hold onto: this piece originated as sponsored content published by MIT Technology Review, so it's a company making its case rather than independent analysis. The engineering constraints it describes are real. Whether Syensqo's particular solutions are best-in-class is a question the qualification process will answer, not a press release.
