Bandwidth - Feeding the AI Beast: Interconnect Scaling in the Data Center cover art

Bandwidth - Feeding the AI Beast: Interconnect Scaling in the Data Center

Bandwidth - Feeding the AI Beast: Interconnect Scaling in the Data Center

Listen for free

View show details

About this listen

Ryan Cole digs into the AI data center crisis nobody's talking about: interconnect bandwidth. Modern AI workloads demand three terabits per second per GPU, but connections between processors can't keep up. He explores why silicon photonics and optical interconnects might be the only way forward—and why your hundred-thousand-GPU cluster is currently an expensive space heater.

Loved this episode? Discover more original shows from the Quiet Please Network at QuietPlease.ai, explore our curated favorites here amzn.to/42YoQGI, and catch just a slice of our AI hosts in action on Instagram at instagram.com/claredelish and YouTube at youtube.com/@DIYHOMEGARDENTV

This content was created in partnership and with the help of Artificial Intelligence AI
No reviews yet