This is the way

[Zest28]

With the export ban of NVIDIA GPU's to China, some people in China are running local LLM models on .... Mac Mini's. No need for large cooling solutions and energy requirements when running something like this locally.

Simply connect many Mac Mini's to form a super cluster to run local LLM's.

 

[conmee]

That's probably the iCloud datacenter in Cupertino. 😉

 

[splifingate]

The switches connecting probably cost 32x the minis!

 

[Populus]

This… if this starts happening at a large scale, I’m worried the bigger demand of minis makes their price to rise…

The M5 mini cannot come soon enough. I wish they finally release it before this mini-farms for LLMs rise in popularity.

 

[VitoBotta]

I've seen that video before, but I read it was from a bot farm - not related to LLMs.