Acacia’s Tom Williams joins Vipul Deokar from Meta and Rakesh Chopra from Cisco to discuss the need for scale across networks and why silicon, systems and optics must be co-optimized. Tom’s presentation focuses on the importance of building an efficient optical backbone to maximize GPU utilization. This requires coherent optical innovations such as 800ZR/ZR+ and 1600ZR/ZR+ because they enable operators to scale cost-effectively while also being extremely reliable, power efficient and compact.
Tom explains how optical innovations such as coherent pluggables and multi-rail open line systems are at the heart of AI infrastructure because they deliver the performance, efficiency, and reliability for AI workloads that maximize uptime and return on AI investments. At each inflection point – from 400G to 800G to now 1.6T – the industry has had to deal with growing bandwidth demand and coherent optics has risen to the challenge every time.
