Nvidia releases free tool to pool home computers for local AI processing

Nvidia has introduced Personal AI Router (PAIR), an open-source software tool that connects multiple computers in a home network to share AI inference workloads without relying on cloud services.

How it works#

PAIR automatically discovers compatible machines on your network and coordinates them to handle local AI tasks—particularly those powered by tools like Ollama and LM Studio. Rather than sending all processing to a single device, the system distributes computational demands across multiple computers working in parallel, which can help manage complex AI workflows that break tasks into many smaller jobs. The software activates idle computing resources without interfering with other applications, and adapts dynamically if a device leaves the network (such as when someone starts playing a game on their desktop).

Device compatibility#

The tool supports recent Nvidia graphics cards—RTX 20-series and newer, plus RTX Pro and DGX Spark systems—alongside Apple's M4 chips and later. Nvidia estimates that a typical household with a gaming PC, work laptop, and other modern computers contains significant untapped processing potential, even when accounting for electricity costs.

Security and availability#

Devices pair through a six-digit code and establish encrypted connections via mutual TLS (mTLS), creating a two-way trusted channel between machines. The beta launched today for Windows, Linux, and macOS. Nvidia indicated that realistic deployments typically involve a laptop paired with a gaming PC rather than the multi-device powerhouse scenarios sometimes described.

Simplified agent setup#

Concurrently, Nvidia announced streamlined configuration for three AI agent applications—Perplexity Portable Computer, Hermes Agent, and OpenClaw—enabling users to deploy local agents on Windows with just a few clicks.