NovoLINC Launches MaxLINC Thermal Interface Material for High-Power AI Chips
New TIM delivers thermal resistance as low as 0.7 mm²·K/W and supports heat flux above 100 W/cm², targeting multi-kilowatt GPUs, CPUs and ASICs in liquid-cooled AI datacentres.
NovoLINC, an innovator in advanced thermal interface material (TIM) solutions for high announced MaxLINC, a new generation of thermal interface material engineered for the rapidly increasing power and heat flux of AI computing. MaxLINC achieves thermal resistance as low as 0.7 mm²•K/W with excellent reliability, delivering over 20% cooling energy savings for AI servers compared to current industry phase-change TIM or thermal grease products.
As AI accelerators move toward multi-kilowatt power levels, the thermal interface between semiconductor devices and advanced cooling systems is becoming an increasingly critical performance bottleneck. MaxLINC is designed specifically to address this challenge.
Built on NovoLINC's proprietary nanostructured composite architecture, MaxLINC is engineered for high-heat-flux applications exceeding 100 W/cm², while helping mitigate localised hot spots. The product platform is designed for use across multiple thermal interfaces, from direct chip-to-cold plate interfaces and chip-to-package heat spreaders to interfaces between packaged devices and cold plates or other cooling hardware and accommodates device and package warpage as high as 350 µm.
This combination of ultra-low thermal resistance, high-heat-flux capability, mechanical compliance, application flexibility, and warpage accommodation addresses critical-to-quality requirements emerging on next-generation semiconductor thermal roadmaps. These capabilities are becoming increasingly important as the industry shifts from air cooling to direct liquid cooling and other advanced cooling architectures for multi-kilowatt GPUs, CPUs, ASICs, and other AI accelerators.
"Heat is quickly becoming the limiting factor in AI computing," said Dr Ning Li, Co-Founder and CEO of NovoLINC. "Every bit of thermal resistance we remove turns directly into headroom that can be redirected to higher chip performance, more compute, less wasted energy, and lower operating costs for the data centre operator."
Scaling Manufacturing in Pittsburgh
MaxLINC samples are available now for customer qualification. To support growing customer demand and commercialisation of the MaxLINC product line, NovoLINC is expanding its Pittsburgh-area operations into a new facility in Sharpsburg, Pennsylvania. The facility will provide increased capacity for manufacturing scale-up, product qualification and testing, and continued R&D.
"The Sharpsburg facility gives us the capacity and infrastructure to scale production alongside customer demand," said Darren Goetz, Vice President of Operations at NovoLINC. "As more customers move MaxLINC through qualification and toward production, this facility positions us to support their requirements at the volume, quality, and pace they expect."
NovoLINC has built commercial engagement across the AI infrastructure ecosystem, working with hyperscalers, advanced AI and networking semiconductor companies, AI server OEMs and ODMs, and cooling technology manufacturers.
The company is a member of the Open Compute Project (OCP) Startup Program and the NVIDIA Inception Program and has been recognised as a Pittsburgh Technology Council Tech 50 honoree.
