How NVIDIA Connects 72 GPUs in One Rack @davidbombal
How NVIDIA Connects 72 GPUs in One Rack  @davidbombal
Uploaded November 2026 | Updated November 2026, 20 hours ago
NVIDIA CTO Michael Kagan joins David Bombal to explain AI networking, NVLink, and why GPUs alone are not enough to scale AI data centers.

How do you connect 72 GPUs in a rack? What is the difference between scale up, scale out, and scale across? And why do latency and network jitter matter when GPUs need to work together?

Michael explains how NVIDIA approaches these challenges using NVLink, InfiniBand, and Spectrum-X Ethernet. He also shares the story of co-founding Mellanox, joining NVIDIA, and building the networking behind increasingly large AI systems.

We discuss the path toward million-GPU systems, connecting AI factories across sites, and how simulation and shared experience could help robots learn. Michael finishes with advice for the next generation of engineers and founders.

If you're learning networking, working with data centers, or exploring AI infrastructure, this conversation will help you understand how the pieces fit together. What would you like Michael to explain next? Let us know in the comments.

Big thank you to @NVIDIA for sponsoring my trip to GTC Berlin 2026 and to them for sponsoring this video.

// Michael Kagan SOCIAL //
LinkedIn: linkedin.com/in/mikagan
Blog: blogs.nvidia.com/blog/author/michaelkagan

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

// MY STUFF //
amazon.com/shop/davidbombal

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

// MENU //
0:00 - Intro
0:37 - Michael Kagen introduction and backstory
06:15 - Scaling AI with networking
10:59 - AI Factory
15:58 - Scale up, scale out & scale across explained
17:20 - How it is all connected // NVIDIA NVLink
20:07 - How big will AI get? // Millions of GPUs
22:26 - NVIDIA open-source projects
23:17 - The future of AI
26:42 - Advice for the younger generation
28:49 - 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.
#nvidia #nvidiagtc #networking
How NVIDIA Connects 72 GPUs in One RackWARNING: AI tracks your faceHow to Hack AI Models and Agents (with Python demos)1000x More Powerful Than an RTX 5090These 6 Sci-Fi Movies Predicted The AI Nightmare!Learn How to Hack for FreeWhy the FBI Wanted This Hacking ToolHow to hack IP Cameras (Ethically) and learn IoT hackingWiFi Hacking Now Fits in a PagerHow Audio Files Can Hack Your PhoneThe End of TOKENMAXXING?Which is Ethernet? Whats the difference?
David Bombal |

How NVIDIA Connects 72 GPUs in One Rack

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER