NVIDIA has unveiled the NVIDIA RTX Spark, a new superchip that reinvents the Windows PC for the era of personal AI agents, introducing what the company describes as a new class of computer that evolves from tool to teammate. Designed with equal emphasis on AI, content creation, and gaming, RTX Spark brings together three decades of NVIDIA innovation, including CUDA, RTX, DLSS, FP4, TensorRT, OptiX, Reflex, and G-SYNC, packaging these technologies into slim Windows laptops with all-day battery life as well as small, ultraefficient desktop PCs. This ambitious launch represents a fundamental rethinking of what a personal computer can be in an age increasingly defined by intelligent, autonomous software.
The vision behind this launch was articulated forcefully by Jensen Huang, founder and CEO of NVIDIA, who declared that “The PC is being reinvented.” He elaborated on the transformative nature of the platform, noting that for forty years users have launched apps through clicking and typing, whereas with RTX Spark and Microsoft Windows, users simply ask and the PC does the work. Huang emphasized that RTX Spark consolidates everything NVIDIA has built, from CUDA and RTX to its broader AI platform, into a single superchip capable of running local agents, frontier models, creative workflows, and RTX games all on a laptop, characterizing it as the new personal AI computer.
At the heart of this innovation lies a sophisticated hardware architecture. The RTX Spark superchip features an NVIDIA Blackwell RTX GPU with 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision, connected via the NVIDIA NVLink-C2C chip-to-chip interconnect to a high-performance, 20-core NVIDIA Grace CPU. MediaTek, a market leader in Arm-based system-on-a-chip designs, collaborated with NVIDIA on the custom CPU design, contributing to its best-in-class power efficiency, performance, and connectivity. This combination of cutting-edge components establishes a foundation capable of meeting the demanding requirements of modern AI workloads while maintaining the portability and efficiency that users expect.
A central motivation behind RTX Spark is its purpose-built design for personal AI agents, which have reached a critical inflection point. Open source projects such as OpenClaw and Hermes Agent have achieved record-breaking numbers on developer networks like GitHub and OpenRouter, yet broad adoption has been constrained by the inability to run agents securely and privately on users’ primary PCs. To address this challenge, NVIDIA and Microsoft have partnered to deliver a robust, secure Windows platform for on-device agents. The collaboration begins with a strong foundation comprising new Windows security primitives and the NVIDIA OpenShell runtime, ensuring that agents run safely and remain fully under user control.
The security and privacy architecture underpinning the platform is notably comprehensive. The new Windows primitives deliver identity, containment, policy, and end-to-end security capabilities for building and running agents natively, while NVIDIA OpenShell provides additional policy capabilities that allow users to define precisely what agents can and cannot do. OpenShell can intelligently route queries to local models based on the user’s privacy preferences and disguise personal information in queries sent to cloud models. This layer is already being adopted by leading agent developers such as Hermes Agent and OpenClaw in their new Windows apps, which will enable users to securely access powerful on-device agents capable of executing tasks across Windows applications, reasoning through cross-app workflows, generating images and video, coding plug-ins and apps, and semantically searching local files.
Industry leaders have expressed strong support for this vision. Vincent Koc, chief architect at the OpenClaw Foundation, described his organization as strong supporters of deploying agents securely into the Windows ecosystem, noting that running solutions like OpenShell alongside Microsoft security primitives on RTX Spark will enable a fully integrated stack for private, personal agents. The hardware demands of such agents are substantial, and RTX Spark rises to meet them with up to 1 petaflop of AI compute and 128GB of unified memory. Dillon Rolnick, CEO of Nous Research, captured the significance of this capability, suggesting that users will realize they are buying a full-fledged assistant rather than a typical laptop. Satya Nadella, chairman and CEO of Microsoft, framed the partnership within an even broader ambition, stating that the goal is to deliver unmetered intelligence to every home and every desk with Windows, and describing RTX Spark as a real breakthrough toward that vision.
Beyond its agentic capabilities, RTX Spark delivers the full NVIDIA AI and graphics technology stack to creators, AI developers, and gamers. Users can render ultralarge 90GB 3D scenes with OptiX and DLSS, edit 12K 4:2:2 video with the NVIDIA Blackwell decoder, run 120-billion-parameter large language models with a one million token context, and play AAA games at 1440p resolution and over 100 frames per second with ray tracing, DLSS, and Reflex. The platform will also power new RTX capabilities, including DLSS 4.5 Ray Reconstruction featuring a second-generation transformer model, coming to Blender and dozens of games, as well as RTX Video with 4x Frame Generation arriving in ComfyUI. With RTX technology already boosting performance across more than 1,000 games and applications, over 100 Windows software providers, including Adobe, Blackmagic Design, Blender, CapCut, ComfyUI, and OTOY, alongside game developers such as KRAFTON, NetEase, Remedy Entertainment, Riot Games, and XBOX, are embracing the new platform.
The creative possibilities are further amplified through a deepened partnership with Adobe, which is working with NVIDIA to rearchitect Adobe Premiere and Photoshop specifically for RTX Spark. Firefly-powered Generative Fill in Photoshop and Generative Extend in Premiere are among hundreds of accelerated tools that deliver enhanced creative power, precision, and control, with RTX Spark delivering up to 2x faster AI, editing, coloring, and effects across creative workflows. Shantanu Narayen, chair and CEO of Adobe, emphasized that the expanded partnership with NVIDIA and Microsoft will make creative experiences faster and more powerful than ever, building AI-native experiences that deliver the performance, intelligence, and responsiveness creators need. Adobe Premiere will feature a new video pipeline tapping into RTX Spark’s unified memory, Blackwell GPU, and TensorRT software, while the next-generation Photoshop engine will be optimized for GPU-accelerated compositing, and tools such as Substance 3D Painter and Stager will run natively for smoother 3D workflows.
In terms of design and form factor, RTX Spark laptops are engineered to be as slim as 14 millimeters and as light as three pounds, available in 14- to 16-inch sizes with precision-machined aluminum chassis that blend durability with a clean, modern aesthetic. Color-accurate tandem OLED displays with NVIDIA G-SYNC technology provide stunning visuals for both creative work and immersive gaming, while small, ultraefficient RTX Spark desktops cater to agents, creative workloads, gaming, and everyday productivity. Major hardware makers have rallied around the platform, with leaders from ASUS, Dell, HP, Lenovo, MSI, and Microsoft Surface offering enthusiastic endorsements and committing to designs that span the full spectrum of premium personal computing.
The collaboration between NVIDIA and Microsoft extends beyond personal agents to frontier agents through the NVIDIA DGX Station for Windows, which scales the Blackwell architecture to enterprise developers by placing an AI supercomputer deskside for running agents. RTX Spark laptops and compact desktops will be available this fall from leading manufacturers including ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with models from Acer and GIGABYTE to follow. As the industry gathers for additional detail at the Microsoft Build keynote, the introduction of RTX Spark stands as a bold declaration of NVIDIA and Microsoft’s shared ambition to usher in a new era of personal computing, one in which the PC transforms from a passive tool into an intelligent, secure, and deeply capable teammate working alongside its user.
