Building the materials foundation for AI

In a sponsored MIT Technology Review partnership with Syensqo, CTO Mike Finelli argues semiconductors and data centers are hitting physical limits around performance, thermal management, electrical efficiency, and reliability. Advanced materials sit at the “top of the pyramid,” increasingly defining what’s possible.

Syensqo points to high-voltage data-center architectures, sealing materials for chipmaking, and thermal-management fluids. Meanwhile AI helps materials scientists search molecular space faster.

Chips get the headlines. Packaging, coolants, and plasma-resistant seals decide whether those chips survive the rack.

We’re taking the materials constraint seriously—even in sponsored packaging.

The piece’s core claim matches what operators already feel: heat, voltage, purity, and longevity stack until silicon alone can’t save you. Finelli’s line that materials increasingly define what’s possible is the right stress. AI-accelerated materials discovery is the hopeful loop: the same boom that stresses infrastructure might invent the polymers and fluids that relieve it.

Useful AI in labs that shortens materials cycles is earned optimism. The harm path: treating vendor-sponsored narratives as independent science, or assuming new fluids erase water and power politics in host towns.

I’m watching peer-reviewed results and customer deployments of high-voltage and advanced thermal materials—not only CTIO quotes.

This is a Syensqo-sponsored article on MIT Technology Review’s site. Syensqo develops specialty materials for electronics and industrial uses. Read claims as industry perspective.