Syensqo's Finelli: AI pushes chips to physical limits

Syensqo's Finelli frames the AI boom as a materials challenge, not just a computing one. He says semiconductors and data centers are approaching physical limits around performance, thermal management, electrical efficiency, and reliability, which raises the number of requirements—what he calls the 'and, and, and' principle—that a material must satisfy at once.

Syensqo is responding on several fronts it says are already in development: materials for high-voltage data center architectures, sealing materials for semiconductor fabs and wafer tools, and heat transfer fluids for direct immersion cooling. Some of this work draws on materials Syensqo already developed for electric vehicles, such as insulating polymers for bus bars and a battery binder for lithium-ion cathodes, which the company says can transfer to data centers facing higher voltage and energy density.

On discovery, Syensqo says it has used AI agents with Microsoft for about two years to digitally synthesize millions of molecular combinations, predict their performance and sustainability, and rank them down to roughly a hundred for lab testing. Finelli describes a possible reinforcing cycle—AI helping develop materials that improve AI infrastructure, which then enables better AI for materials discovery—though he says he is curious how fast that loop will move. Whether the approach delivers at scale likely depends on how quickly lab-validated materials reach production.