Can I connect several computers to run one AI model?
Yes. Distributed inference software such as exo splits a model across machines so their memory adds up. Jeff Geerling connected four Mac Studios with 1.5 TB of combined memory and ran the 671B-parameter DeepSeek V3.1 with exo.
The machines must be connected with a fast link. Thunderbolt 5 cables directly between Macs work best. Ethernet works, but slower links limit speed.
Pooling memory is the main reason to cluster: it lets you run models that do not fit on one machine. With the right setup, it can also make models run faster.
Other tools that can split models across machines include llama.cpp’s RPC mode and vLLM with Ray, mostly on NVIDIA GPUs. exo is built for clusters of Apple silicon Macs and finds devices automatically.
Sources
Updated