NVIDIA Announces PAIR Beta to Spread AI Inference Across Idle Home PCs
PAIR treats the household's existing PCs as a pooled local AI resource rather than demanding one powerful machine, and it lands alongside the RTX Spark Windows PC push that NVIDIA is using to court game publishers.
Reporting from 1 source: GameBusiness.jp.
NVIDIA announced the beta release of PAIR (Personal AI Router), open-source software that distributes AI inference across multiple PCs on a home network. Announced at IFA 2026, PAIR detects compatible PCs on the local network and routes AI work to machines with spare capacity, supporting Ollama and LM Studio on Windows, macOS, and Linux. Supported hardware includes GeForce RTX 20 series and later, RTX PRO workstation GPUs, DGX Spark, and Macs with Apple M4 or later.
NVIDIA positions PAIR as a response to a simple fact: more than half of US households own two or more PCs, and most of that processing power sits idle through the day. The software detects compatible machines on the local network and shifts AI inference to whatever has spare capacity, adapting as devices join or leave.
The company demonstrated the approach by spreading five sub-agents of the Hermes Agent across three PCs, reporting double the throughput of a single node. PAIR pairs with the RTX Spark Windows PCs arriving in October, a line that publishers including Electronic Arts, Ubisoft, and Riot Games have signed on to.
NVIDIA also cited work with the llama.cpp and vLLM communities, claiming up to 1.9x throughput gains on the GeForce RTX 5090 and up to 1.4x on DGX Spark clusters.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.
Sources
- GameBusiness.jp NVIDIA、複数PCにAI処理を分散する「PAIR」ベータ版提供を発表―アイドルGPUを有効活用