NVIDIA RTX Spark Compact PCs Launch in October Alongside Local AI Router PAIR
NVIDIA announced that RTX Spark Windows PCs from six OEM partners, including Acer and Lenovo, will begin shipping in October with 20-core Grace CPUs and Blackwell GPUs. Alongside the hardware, NVIDIA released Personal AI Router (PAIR), a free open-source tool that intelligently distributes local AI inference tasks across idle devices on a home network. This launch brings high-efficiency, always-on AI compute to compact desktop and laptop form factors with OS-level governance for personal background agents. Meanwhile, PAIR enables users to combine spare compute from multiple PCs and Macs into a private local AI cluster, reducing GPU bottlenecking without relying on cloud services. RTX Spark delivers up to 1 Petaflop of compute powered by up to 128GB of unified memory shared between the Grace CPU and Blackwell GPU. PAIR supports Windows, macOS, and Linux, integrating with frameworks like Ollama and LM Studio across RTX 20-series or newer GPUs, DGX Spark, and Apple M4 or newer processors.
## BACKGROUND
RTX Spark is an Arm-based computing platform co-developed by NVIDIA and Microsoft to power Windows on Arm laptops and small form factor desktops for local AI workloads. As multi-agent workflows split complex tasks into parallel LLM requests, running them on a single machine can saturate the GPU, highlighting the need for local distributed inference systems.