I wonder why Apple haven’t looked at something like a branded and seamless p2p inference option for customers that have other Apple gear? Pretty sure there are a few open source options that do this now. vLLM and LocalAI.io
Where do you read they haven’t looked? Maybe they did but found out it didn’t work well enough, or they still are looking and haven’t found a way to do it well enough, or they fear the political backlash it would create for restricting such a feature to other Apple gear.