Is this the most powerful network switch ever built? In this interview from Cisco Live, we look at the new generation of Cisco 8000 and Nexus switches capable of routing 100 Terabits per second. We break down why AI data centers are forced to move from air cooling to liquid cooling, and how a single switch chassis can now handle the equivalent mobile traffic of 100 million people.

AI models are growing faster than the infrastructure can keep up. To solve this, the network switch had to evolve. In this video, I talk to Will from Cisco about the engineering challenges of building the "G300" generation of switches, hardware so dense that air cooling is no longer enough.

We discuss the massive architectural shift occurring in data centers, where Liquid Cooled Switches are becoming the new standard to support 1.6T Ethernet ports and massive GPU clusters.

Key Hardware Topics:
• The 100 Terabit Chassis: How Cisco architecture handles massive throughput.
• Liquid Cooling: Why switches are adopting "direct-to-chip" cooling just like gaming rigs.
• Scale-Out Networking: How these switches manage congestion for AI training jobs (Job Completion Time).
• Career Insights: Will manages 5,000 engineers and explains why understanding the physical layer and hardware constraints is a superpower for modern developers.

Big thanks to Cisco for sponsoring my trip to Cisco Live EMEA and for changing my life and the lives of many other people. 

//  Will Eatherton SOCIAL //  
LinkedIn:   https://www.linkedin.com/in/willeatherton/ 
Newsroom:  https://newsroom.cisco.com/c/r/newsroom/en/us/executives/will-eatherton.html 

// David's SOCIAL //  
Discord: https://discord.com/invite/usKSyzb  
X: https://www.twitter.com/davidbombal  
Instagram: https://www.instagram.com/davidbombal  
LinkedIn: https://www.linkedin.com/in/davidbombal  
Facebook: https://www.facebook.com/davidbombal.co  
TikTok: http://tiktok.com/@davidbombal  
YouTube: https://www.youtube.com/@davidbombal 
Spotify:  https://open.spotify.com/show/3f6k6gERfuriI96efWWLQQ 
SoundCloud:  https://soundcloud.com/davidbombal 
Apple Podcast:  https://podcasts.apple.com/us/podcast/david-bombal/id1466865532 

// MY STUFF // 
https://www.amazon.com/shop/davidbombal  

// SPONSORS //  
Interested in sponsoring my videos? Reach out to my team here: sponsors@davidbombal.com 

// MENU // 
0:00 - Coming up 
0:49 - Will Eatherton introduction and projects // Hyperscale, neocloud & enterprises 
08:03 - New Cisco hardware // Silicon One G300 
13:18 - Data centers + AI + GPUs 
16:27 - G300 use case & Cisco Nexus 
21:56 - Liquid-cooled switches 
24:12 - Networking as a career path 
28:41 - Development and opportunities in networking 
30:14 - Conclusion 

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

David Bombal

Cisco's new G300 ASIC delivers 100 Tbps switching capacity—equivalent to the mobile traffic of 100 million people—doubling bandwidth while increasing power by only 20%. The chip introduces architectural features for unified AI fabrics, supporting both GPU-to-GPU backend traffic (with packet spraying and reordering) and traditional server-to-server flows. Liquid cooling is becoming standard for hyperscalers and neoclouds, while enterprises adopt more slowly. The interview covers three deployment segments (hyperscalers, neoclouds, enterprises), the evolution from separate backend/frontend/storage networks to converged fabrics, and career advice emphasizing networking fundamentals, open-source contributions (especially Sonic), and hands-on experimentation over syntax memorization.

The 100Tbps AI Switch: Inside the Beast