NVIDIA RTX Spark: The Superchip That Reinvents Windows PCs for the Age of Personal AI

NVIDIA just unveiled RTX Spark at GTC Taipei — a superchip that combines a Blackwell RTX GPU with a 20-core Grace CPU to deliver 1 petaflop of AI performance in a PC that fits in a backpack. The implications for on-device AI agents are massive.

What Makes RTX Spark Different

This is not a GPU refresh. RTX Spark is an entirely new category: a unified superchip with 6,144 CUDA cores, 5th-gen Tensor Cores with FP4 precision, and up to 128GB of unified memory connected via NVLink-C2C. All of this fits into laptops as slim as 14mm and as light as 3 pounds — with all-day battery life.

For context: a 1-petaflop AI PC was science fiction a year ago. Today it is shipping in consumer hardware this fall.

The Agent-Ready PC

NVIDIA and Microsoft collaborated on new Windows security primitives specifically designed for AI agents — identity, containment, policy, and end-to-end security. On top of that, NVIDIA built OpenShell, a runtime that lets users define which models run locally vs. in the cloud based on privacy policies.

Both the OpenClaw Foundation and Nous Research have already endorsed the platform. Hermes Agent, one of the most popular open-source agents, will run natively on RTX Spark hardware.

What This Means for Developers

RTX Spark can run 120-billion-parameter LLMs locally with up to 1 million tokens of context. That means entire codebases can fit in model memory on a laptop. Combined with the Surface RTX Spark Dev Box announced at Microsoft Build 2026, developers now have a dedicated AI development machine with 128GB unified memory, preconfigured with VS Code and GitHub Copilot.

Adobe is also rearchitecting Photoshop and Premiere Pro from the ground up for RTX Spark, promising 2x faster AI performance on creative workloads.

"The PC is being reinvented. For forty years, you launched apps. Click. Type. With RTX Spark and Microsoft Windows, you ask — and the PC does the work." — Jensen Huang, NVIDIA CEO

Availability and Bottom Line

Consumer RTX Spark-powered devices are scheduled for fall 2026. The Surface RTX Spark Dev Box launches in the US later this year with pricing yet to be announced.

The bigger picture: on-device AI agents have been bottlenecked by hardware. RTX Spark removes that bottleneck. When every developer can run a 120B parameter model locally with full privacy, the way we build software changes. The agent era is no longer coming — it is being delivered in a 14mm chassis.

Source: NVIDIA Newsroom (nvidianews.nvidia.com) — "NVIDIA and Microsoft Reinvent Windows PCs for the Age of Personal AI", May 31, 2026