Nvidia wants your home computers to work as a team. The company’s new Personal AI Router, known as PAIR, is free, open-source software that discovers compatible machines on a local network, links them together, and prepares them to split AI inference tasks across multiple GPUs without sending a single prompt to the cloud.
Think of it like a load balancer for your living room. Instead of one PC struggling to run a large language model alone, PAIR orchestrates the workload across every capable device on the network, routing requests to whichever machine has available headroom.
What PAIR actually does
PAIR is not a physical router. No new hardware required, no box to plug into the wall.
It is software that handles network discovery, device coordination, and task distribution for local AI inference. Compatible tools include Ollama and LM Studio, two of the more popular applications for running large language models on consumer hardware.
