AI infrastructure’s next constraint may be advanced materials
Syensqo says it is developing materials for high-voltage data centres, semiconductor manufacturing and cooling while using AI to accelerate molecular research.
The expansion of artificial intelligence is creating new demands on the physical systems that support it, as semiconductors and data centres approach limits involving heat, electrical efficiency, chemical resistance and reliability.
In a sponsored Business Lab discussion published by MIT Technology Review’s Insights division, Syensqo described work on materials for high-voltage data-centre architectures, semiconductor-fabrication seals and thermal-management systems, including fluids for direct-immersion cooling.
The company also points to potential transfers from other industries. Materials developed for electric vehicles, including insulation and battery technologies, may help address the higher voltage and energy-density requirements emerging in data centres. Syensqo is also working on energy-storage systems intended to support renewable power and backup capacity.
AI is being used within Syensqo’s research process to screen millions of possible molecular combinations, predict performance and sustainability characteristics, and prioritise about 100 candidates for laboratory testing. The company says this approach allows scientists to search more broadly and spend more time on engineering challenges.
Sustainability is assessed at the beginning of Syensqo’s research projects, according to the company, which says 88 per cent of its portfolio meets its definition of a sustainable product. The supplied material does not establish whether the described materials are prototypes, in testing or broadly deployed, and the company’s performance claims have not been independently verified.
