#549: The Apple of Networking? Cisco’s 100T Full Stack (Connects 128,000 GPUs)

The bottleneck in AI isn't compute anymore, it's the network. In this video, I sit down with Martin, the architect behind Cisco’s Silicon One, to discuss the massive leap to 100 Terabits per second. We go deep into the silicon level to understand how "intelligent agents" embedded in the hardware are solving the packet loss problem for massive AI training clusters.We cover the new 1.6T Linear Pluggable Optics (LPO), why Cisco is becoming the "Apple of Networking" by building their own full stack, and why they believe Ethernet has officially won the data center war.Topics Covered:• Cisco Silicon One: The architecture behind the 100Tbps & 51.2Tbps chips.• AI Scale: How to interconnect 128,000 GPUs without stalling.• Hardware Agents: Real-time traffic rerouting at the silicon level.• 1.6Tbps Optics: Moving DSPs out of the module to save power (LPO).• Ethernet vs. InfiniBand: Why standard Ethernet is winning in AI.Big thank you to ‪@Cisco‬ for sponsoring my trip to Cisco Live Amsterdam!// Martin Lund SOCIALS // LinkedIn: / martinlundca // Website REFERENCE // https://blogs.cisco.com/sp/cisco-sili... // David's SOCIAL //Discord: discord.com/invite/usKSyzbTwitter: www.twitter.com/davidbombalInstagram: www.instagram.com/davidbombalLinkedIn: www.linkedin.com/in/davidbombalFacebook: www.facebook.com/davidbombal.coTikTok: tiktok.com/@davidbombalYouTube: / @davidbombalSpotify: open.spotify.com/show/3f6k6gE...SoundCloud: / davidbombalApple Podcast: podcasts.apple.com/us/podcast...// MY STUFF // https://www.amazon.com/shop/davidbombal // SPONSORS // Interested in sponsoring my videos? Reach out to my team here: [email protected] Please note that links listed may be affiliate links and provide me with a small percentage/kickback should you use them to purchase any of the items listed or recommended. Thank you for supporting me and this channel! Disclaimer: This video is for educational purposes only. #cisco #ciscolive #ciscoemea