Nvidia unveiled PAIR (Personal AI Router) at IFA 2026 in Berlin, a free open-source tool that transforms compatible devices on your home network into a private local AI cluster for running inference tasks without relying on cloud services.

PAIR automatically discovers compatible PCs—including systems with Nvidia RTX GPUs, DGX Spark, and Mac devices—on the same local network and connects them into a unified inference cluster. The tool routes AI inference requests across available nodes, helping users maximize local compute while keeping prompts, files, and agent context private on their home network.

The system supports popular local inference backends like Ollama and LM Studio, giving apps a consistent single endpoint while intelligently proxying requests to available local compute across Windows, Linux, and macOS devices. This means developers and AI enthusiasts can now leverage multiple machines on their network to handle complex AI workloads that might overwhelm a single device.

Nvidia emphasized that PAIR requires no special hardware, cables, or complex cluster setup—users simply download the free software, install it on supported systems, and pair devices on the same local network. The tool works with GeForce RTX GPUs from the 20 Series and newer, DGX Spark/GB10 systems, and Mac M4 or newer devices.

The announcement positions Nvidia as a leader in the growing 'personal AI' movement, where users increasingly want to run AI models locally for privacy, cost savings, and reduced latency. By making the tool free and open-source, Nvidia is also building its ecosystem and encouraging adoption of its hardware platforms for local AI inference.

Sources