NVIDIA has unveiled RTX Spark, a new AI focused superchip developed in partnership with Microsoft, marking what the company describes as the beginning of a new era of personal AI computing.
Announced on May 31, RTX Spark is designed to transform Windows PCs from traditional productivity tools into AI powered assistants capable of executing tasks, reasoning across applications, generating content and running advanced AI models directly on the device.
“The PC is being reinvented,” said Jensen Huang, founder and CEO of NVIDIA. “For forty years, you launched apps. Click. Type. With RTX Spark and Microsoft Windows, you ask and the PC does the work.”
RTX Spark combines 30 years of NVIDIA technologies, including CUDA, RTX, DLSS, TensorRT, OptiX, Reflex and G-SYNC, into a single platform aimed at AI, content creation and gaming workloads. The superchip integrates an NVIDIA Blackwell RTX GPU featuring 6,144 CUDA cores and fifth generation Tensor Cores with FP4 precision, paired with a 20 core NVIDIA Grace CPU through the company’s NVLink C2C interconnect.
MediaTek collaborated with NVIDIA on the custom CPU design, contributing expertise in Arm based system on chip architectures to deliver improved power efficiency and connectivity.
The new platform is built specifically for on device AI agents, which NVIDIA believes are reaching a turning point in adoption. Through a partnership with Microsoft, the company is introducing a secure Windows environment for AI agents using new Windows security primitives and NVIDIA OpenShell, a runtime designed to ensure users maintain control over agent activities and data privacy.
According to NVIDIA, OpenShell allows users to define what AI agents can access, route queries to local models based on privacy preferences and mask sensitive information before data is sent to cloud services.
Leading AI agent developers including OpenClaw and Hermes Agent are already adopting the platform. Their applications will enable users to run AI assistants capable of automating tasks across Windows applications, generating images and videos, writing code and searching local files using natural language.
RTX Spark delivers up to one petaflop of AI computing performance and supports up to 128GB of unified memory, allowing complex AI models and agents to run locally rather than relying on cloud infrastructure.
“Our goal is to deliver unmetered intelligence to every home and every desk with Windows,” said Satya Nadella, Chairman and CEO of Microsoft. “RTX Spark marks a real breakthrough towards that vision.”
The first RTX Spark powered Windows PCs from leading manufacturers are expected to arrive later this year, positioning NVIDIA to compete more directly in the rapidly growing AI PC market.
