The High Performance Computing (HPC) and ArtificiaI Intelligence (AI) era requires unprecedented computing capabilities. With this in view, industry is heading towards zettascale supercomputers, based on >100k processing units. Chiplets-based disaggregated processors and memory-centric computing architectures are taking over and require an ever more intensive communication between dies. Thus, today interconnect bottleneck is the main problem: the physical nature of copper wires does not allow the scalability of processing architectures to meet computing demand.

At NcodiN, we enable ultra-fast and energy efficient die-to-die interconnects using light as the vehicle of information. We develop an optical interposer based on a proprietary and revolutionary technology, unblocking the full scalability of multi-chiplets based HPC processors.

Partners