Nvidia Unveils RTX Spark With 1 Petaflop for Local AI on Windows PCs
Updated
Updated · InfoWorld · Jun 16
Nvidia Unveils RTX Spark With 1 Petaflop for Local AI on Windows PCs
3 articles · Updated · InfoWorld · Jun 16
Summary
Computex 2026 brought Nvidia’s RTX Spark, a new platform for slim Windows laptops and compact desktops built to run AI models and agents directly on-device rather than in the cloud.
Up to 1 petaflop of FP4 AI performance, 6,144 Blackwell GPU cores, a 20-core Arm-based CPU and 128GB of unified memory are meant to reduce CPU-GPU bottlenecks for local inference and fine-tuning.
Nvidia is pitching Spark as more than an AI box, targeting gaming, creative work, AI development and persistent agentic workflows to broaden appeal beyond niche developer systems.
Pricing is still unclear across manufacturers, but the related DGX Spark desktop is listed around $4,699, suggesting an early premium market aimed at developers, creators and technical professionals.
The launch points to a hybrid AI model in which local systems handle privacy-sensitive, low-latency or disconnected workloads while centralized cloud services remain dominant for shared enterprise knowledge and governance.
Is Nvidia's $4,000 'AI PC' the future for everyone, or just an expensive new gadget for developers?
As AI agents move from the cloud to our desks, who will truly own and control them?
Nvidia RTX Spark Unveiled: 1 Petaflop AI PC Platform Set to Transform Windows Laptops in Fall 2026
Overview
At Computex 2026, Nvidia made a bold move by unveiling its RTX Spark superchip platform, marking its entry into the AI PC market. This strategic launch aims to redefine high-performance computing for Windows users, especially creators, AI developers, and gamers who need powerful local processing. RTX Spark features a 20-core Arm-based CPU and advanced GPU, delivering a transformative experience and pushing the Windows on Arm ecosystem into the agentic AI era. By targeting demanding users and promising significant performance gains, Nvidia positions itself to set new standards for AI-driven PCs.