to me the biggest event more than deepseek launch was when zAI served their latest model on all Chinese chips.
Model serving is trivial, and inference is just memory bandwidth. The cost of serving will be asymptotic to flash read energy.
Having trained on your own chips, that is the impressive part.
alt Hacker News
Model serving is trivial, and inference is just memory bandwidth. The cost of serving will be asymptotic to flash read energy.
Having trained on your own chips, that is the impressive part.