Sparse Analog AI Chips Cut Data Center Energy 100x | Guillaume Verdon
Guillaume Verdon joins MTS to discuss the launch of the open-source Z1T model family, the co-evolution of sparse transformer algorithms with thermodynamic hardware, and how sparse probabilistic processing achieves a 100x efficiency gain for data center inference workloads. Check out our sponsor Lovable: https://lovable.dev. Turn ideas into software people love.
MTS is a live news and interview show covering technology, business, politics, and culture as it happens.
Follow MTS:
X | https://twitter.com/MTSlive
YouTube | https://www.youtube.com/@mtsituation
Spotify |…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.