#TEF26: AI Ethernet at 448G per Lane: Measurement Science Advances

John Calvin, Senior Strategic Planner at Keysight Technologies, discusses the critical measurement challenges facing the advancement of high-speed Ethernet in AI infrastructure. Calvin examines the technical decisions surrounding the transition to 400 Gbps Ethernet, including modulation format choices and the limitations imposed by current channel and connector technology. He highlights a significant milestone in measurement science: breaking through the 100 gigahertz barrier that has defined the industry for decades, opening up the 100-250 gigahertz spectrum for next-generation network development. Calvin also addresses the importance of looking beyond immediate speed targets to 800 Gbps signaling and the role of advanced test and measurement tools in validating interoperability for scale-up and scale-out AI networks. The key technical challenges in advancing from current speeds to 400 Gbps Ethernet How channel performance and connector limitations influence time-to-market decisions The significance of crossing the 100 gigahertz measurement threshold The relationship between measurement science and modulation format selection Why the industry must simultaneously focus on 400 Gbps and 800 Gbps development Keysight's approach to developing test solutions for AI network interconnects 📚 CHAPTERS: 0:00:00 - Introduction and Role at Keysight Technologies 0:00:27 - Challenges in Advancing to 400 Gbps Ethernet 0:01:12 - Breaking the 100 Gigahertz Barrier in Measurement Technology 0:01:47 - Measurement Science and Modulation Format Decisions 0:02:18 - Looking Beyond 400 Gig to 800 Gig Signaling