NovoLINC has launched MaxLINC, a thermal interface material for AI computing, the company said. The Pittsburgh-based company said the product has a thermal resistance as low as 0.7 mm²·K/W and reduces the cooling energy used by AI servers by more than 20% compared with existing phase-change TIM and thermal grease products.
NovoLINC said MaxLINC is built on its nanostructured composite architecture and is designed for heat flux above 100 W/cm², while helping to reduce localized hot spots. The material is intended for use at several thermal interfaces, the company said, including direct chip-to-cold plate contacts, chip-to-package heat spreaders, and the interface between packaged devices and cold plates or other cooling hardware, and it accommodates device and package warpage as high as 350 µm.
Dr. Ning Li, co-founder and chief executive, said heat is becoming the limiting factor in AI computing, and that each reduction in thermal resistance becomes headroom that can go toward higher chip performance, more compute, less wasted energy and lower operating costs for data center operators.
Samples of MaxLINC are available now for customer qualification, the company said. NovoLINC said it is expanding its Pittsburgh-area operations into a new facility in Sharpsburg, Pennsylvania, which will add capacity for manufacturing scale-up, product qualification and testing, and research and development.
Darren Goetz, vice president of operations, said the Sharpsburg facility gives the company the capacity and infrastructure to scale production alongside customer demand. NovoLINC said it works with hyperscalers, AI and networking semiconductor companies, server OEMs and ODMs, and cooling technology manufacturers, and that it is a member of the Open Compute Project Startup Program and the NVIDIA Inception Program. More information about the company is available at its website.
