Nvidia Confirms October Launch for RTX Spark Windows PCs
Nvidia has confirmed a clear launch window for its RTX Spark Windows PCs. The first systems will reach buyers in October. Meanwhile, the company used IFA 2026 to outline both the hardware and the software around it.
These machines target a new class of compact PCs. They aim to run local AI work on the device itself. At the same time, they still support gaming and creative tasks.
The RTX Spark platform pairs a Blackwell RTX GPU with a Grace CPU. Nvidia rates peak AI performance at up to one petaflop. In addition, the top configuration offers up to 128GB of unified memory and a 20-core CPU.
That design lets the same machine host local AI agents. It can also handle games and creation apps. Furthermore, Nvidia’s Windows Agent framework can run background agents. Users get controls inside the operating system.
Lenovo is among the first partners to show finished designs. It announced the Yoga Pro 9n and the Yoga 9n 2-in-1. Acer, for its part, showed a compact desktop. Nvidia added that six other OEMs are preparing RTX Spark systems for the same October window.
Game publishers have also lined up behind the platform. Electronic Arts, Embark, and Ubisoft have announced support. As a result, the machines are not positioned as AI-only boxes.
Software updates arrived alongside the hardware news. Nvidia said Hermes Agent, OpenClaw, and Perplexity Portable Computer will get simpler local model setup on Windows. In parallel, the company updated llama.cpp and vLLM to raise local inference speed.
Nvidia claims llama.cpp can deliver up to 1.9 times higher throughput on a GeForce RTX 5090. Those gains are available directly. They also reach users through apps such as LM Studio and Ollama.
Nvidia also introduced Personal AI Router, or PAIR. The tool is free and open source. It can send AI inference requests across compatible PCs on the same local network.
The PAIR beta supports Windows, macOS, and Linux. Compatible hardware includes GeForce RTX 20 Series and newer GPUs, RTX PRO workstation GPUs, DGX Spark systems, and Apple M4 or newer machines. Therefore, a household can share idle compute without sending work to the cloud.
Taken together, the October PCs and the new software point in one direction. Nvidia wants more AI work to stay on the desk, not only in remote data centers. Pricing for most partner systems has not been announced yet.