Ayar Labs joins NVIDIA NVLink fusion
Collaboration aims to enable high-bandwidth, low-latency optical connectivity for next-generation rack-scale AI systems.
Ayar Labs has joined the NVIDIA NVLink Fusion ecosystem, expanding collaboration around co-packaged optics (CPO) technologies for large-scale AI infrastructure and heterogeneous computing architectures.
The company said its optical interconnect solutions will be made optically and electrically compatible with NVIDIA optical and SerDes technologies, enabling hyperscalers and system developers to build AI infrastructure around NVIDIA’s NVLink Fusion platform.
As AI factories scale to larger GPU counts and increasingly heterogeneous architectures, data movement, bandwidth density and power efficiency are emerging as critical system bottlenecks.
Ayar Labs said its CPO technology is designed to address these challenges by replacing traditional copper interconnect limitations with high-bandwidth, low-latency optical connectivity.
The collaboration supports the development of flexible rack-scale AI architectures capable of integrating custom CPUs and accelerators while maintaining compatibility with NVIDIA’s NVLink ecosystem.
According to the companies, combining NVLink Fusion with Ayar Labs’ optical connectivity technologies will help customers scale AI systems more efficiently while improving power efficiency and infrastructure utilisation across large AI deployments.
The announcement further highlights growing industry momentum behind co-packaged optics as AI workloads continue to drive demand for higher-performance optical interconnects and next-generation system integration technologies.
