Nvidia has announced Personal AI Router, or PAIR, a free software tool meant to make multiple computers on a local network work together on artificial intelligence inference tasks, according to The Verge. The beta is available now for Windows, Linux, and macOS, the report says. Despite the name, PAIR is not a physical router. The Verge reports that it is open-source Nvidia software that finds compatible machines on a network, links them, and prepares them to run local AI workloads used by tools such as Ollama and LM Studio. The core idea is to use machines when they are idle. According to The Verge, PAIR is designed to avoid interrupting normal activity and can adjust when devices join or leave the network, including when someone starts using a PC for something else, such as gaming. Nvidia’s pitch is that agentic workflows can break larger tasks into smaller jobs, making a pool of local machines more useful than one computer shouldering every request. Compatibility is still specific. The Verge says PAIR works mostly with Nvidia hardware: GeForce RTX 20-series cards and newer, RTX Pro GPUs, and DGX Spark systems. Apple M4 chips or newer are also supported, according to the report. Nvidia also gave a security outline. The Verge reports that devices are paired with a six-digit code and that the connection is protected with mutual Transport Layer Security, or mTLS, creating an encrypted channel that authenticates both sides of the connection. The company’s example of available household compute was, as The Verge put it, an “incredibly extreme case.” In a media briefing cited by The Verge, Nvidia product manager Seth Schneider described a home with multiple powerful systems, including RTX laptops, a DGX Spark desktop, a gaming desktop, and a MacBook Pro, estimating about 165 teraflops of underused compute in that scenario. Asked what a more realistic setup might look like, Schneider said Nvidia expects many PAIR users to have something closer to one MacBook or Windows laptop plus one gaming PC, according to The Verge. Nvidia paired the PAIR announcement with another local-AI setup push. The Verge reports that Perplexity Portable Computer, Hermes Agent, and OpenClaw will offer simplified local setup experiences for Nvidia GPUs on Windows, aiming to reduce manual configuration to a few clicks. Who benefits: Users with compatible RTX systems, Apple M4 chips or newer, or multiple local machines stand to benefit most. Developers using tools such as Ollama, LM Studio, or the listed agent apps may get an easier path to running work locally. Who's exposed: The value is likely limited for users without compatible hardware or without multiple machines to pool. Because the report does not include benchmarks, real-world performance and reliability remain open questions.