The a16z Show · Thursday, August 6, 2026
Recounting early AI developments, Simon Mo identifies 'Burt' as a model that necessitated GPU usage for efficiency, preceding even the widespread adoption of GPUs. He also mentions 'ResNet' for quantification in computer vision, which initially required Nvidia K80 GPUs.
“Probably Burt. And before that it was ResNet for quantification, like computer vision, classification.”
“So ResNet already needed to run on Nvidia K80, which is one of the first GPU. Other places. But way over, but even at this point, ResNet you can still kind of run on a commodity even CPU devices, it's just very slow.”
“But for Burt, where running it at like, wow, you have to run on GPU to make it anything faster or efficient for any translation or any task.”