Wow! It was really quite the OFC! Hasan Siraj, VP of Product Management in the Core Switching Group at Broadcom, discusses the company’s approach to building AI infrastructure at massive scale. Siraj examines three critical vectors—scale-up, scale-out, and scale-across—that are shaping the future of AI data centers. He addresses the Ethernet for Scale-Up Networking Group’s recent specification release, the expansion of XPU rack sizes from current configurations to potentially 1,000 XPUs, and Broadcom’s participation in the OCI MSA for extended optical reach. Siraj also covers the ecosystem developing around 1.6 terabit optics and co-packaged optics, Broadcom’s Stratus 6 production deployment, and Jericho 4 technology for cross-data center clustering.
**What You Will Learn:**
– How the three vectors of scale (scale-up, scale-out, and scale-across) address AI infrastructure challenges
– The significance of Ethernet standardization for scale-up networking
– Why XPU rack sizes are expanding and what the OCI MSA partnership enables
– The role of 1.6 terabit optics and 200G SerDes in building large AI clusters
– How Jericho 4 technology enables secure, reliable cross-data center deployments
– Broadcom’s position in the open AI infrastructure ecosystem
Like our videos? Want to participate? Contact info@nextgeninfra.io