18h ago

Elon Musk highlights diffusion models for video generation

Elon Musk stated that diffusion models are especially well-suited for video understanding and generation. He projected that video will constitute the bulk of future AI workloads. The remarks were amplified by Luca Ambrogioni, Assistant Professor of Machine Learning at Radboud University and principal investigator of the Generative Memory Lab.

0
Original post

Nice to see that Elon is a bitstream diffusion believer too!

12:05 AM · May 19, 2026 View on X

Seems a lot of autoregressive models will be converted to diffusion models

8:56 PM · May 19, 2026 · 4.2K Views