NVIDIA RTX Spark brings Grace CPUs, Blackwell GPUs, and unified memory to Windows PCs this fall

Microsoft and NVIDIA announced the RTX Spark platform last week in San Francisco, targeting high-end Windows laptops and small desktops arriving this fall. The new system-on-a-chip aims to bring Windows PCs closer to the performance and efficiency of Apple Silicon, according to Engadget. This launch marks the first time powerful Arm CPUs are joined with NVIDIA's Blackwell GPU technology and up to 128GB of unified memory on a single chip.

What is the RTX Spark hardware architecture?

The RTX Spark SoC combines NVIDIA Grace Arm CPU cores with Blackwell GPU technology and unified LPDDR5x memory. Top configurations feature up to 20 CPU cores, 6,144 CUDA cores, and 128GB of memory running at 9,400 Mbps, according to Engadget. A lower-tier option offers 18 CPU cores and 5,120 GPU cores. The chip also includes fifth-gen tensor cores, fourth-gen ray tracing cores, and a Blackwell video engine with AV1 UHQ support.

How does Microsoft support local AI agents?

Microsoft is integrating a "Get Started" experience into Windows for RTX Spark devices to streamline AI agent deployment. Users can find, install, and launch agents in two clicks through the Windows Store, according to The Verge. Initially, three agents—Hermes, Perplexity, and OpenClaw—are officially supported. Microsoft plans to add more agents over time, aiming to continue developing features so agents can offer new experiences through Windows.

What are the operational risks for IT teams?

Early testing by The Verge revealed setup issues with OpenClaw, which required uninstallation and reinstallation to resolve gateway errors. While Microsoft promises simplified agent management via the Windows Store, real-world deployment may involve troubleshooting dependencies. The shift to Arm-based hardware means these laptops use a different architecture than traditional x64 systems from AMD and Intel, which could affect software compatibility for some users.

Questions to ask your suppliers

Security teams should ask vendors whether the new Arm-based hardware supports existing endpoint detection and response tools. Verify if the unified memory architecture changes how data is handled during sleep or hibernation states. Check whether the device manufacturers provide clear documentation for managing local AI agents and their network dependencies. Ensure that patch management processes can handle updates for both the OS and the specialized AI software components. IT leadership needs to understand the total cost of ownership for these high-memory devices.

Background: AI agents

AI agents are software systems that perceive their environment, make decisions, and execute actions to achieve specific goals. They move beyond generating text to performing tasks like coding, scheduling, or data analysis. You must treat them as active participants in your network, not passive tools.

Read the full guide: AI Agents Examples: How Autonomous Software Changes Security

What to do and how to stay safe: NVIDIA

• Audit current x64 application dependencies to identify potential compatibility issues with new Arm-based Windows devices.
• Monitor vendor announcements for driver and firmware updates specific to RTX Spark hardware during the initial deployment phase.
• Test local AI agents in a isolated environment before approving them for general employee use to assess resource consumption.
• Review security policies to ensure they account for new local data processing capabilities enabled by the unified memory architecture.

Step-by-step guide: Browser Security Settings for Small Business Teams

General security guidance from the TechSec Feed newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

When will RTX Spark laptops be available?

Microsoft and NVIDIA expect the first wave of RTX Spark laptops and small desktops to launch this fall, according to Engadget.

Which AI agents are supported at launch?

The first three officially supported agents in Windows at launch are Hermes, Perplexity, and OpenClaw, according to The Verge.

Does RTX Spark use Intel or AMD processors?

No, RTX Spark uses NVIDIA Grace Arm CPU cores, moving away from the traditional x64 architecture used by Intel and AMD, according to The Verge.