Nvidia’s RTX Spark chip is arriving in laptops and compact desktop PCs designed to run AI models directly on the device. The systems, unveiled by Nvidia partners at IFA in Berlin, target users who want to use AI agents and other demanding tools without sending sensitive data to cloud services.
Luke Larsen reports for WIRED that Lenovo, Dell, Asus, Microsoft, HP, and Acer are among the companies preparing RTX Spark products. The first laptops are expected this fall.
Lenovo showed the Yoga 9n 2-in-1, a 16-inch convertible laptop with an OLED touchscreen, a 120 Hz display, six speakers, and a 9.2-megapixel webcam. It uses an Arm-based system-on-a-chip that combines Nvidia’s Grace CPU, Blackwell RTX graphics, and shared memory in one package. This approach resembles Apple’s integrated chip design more than a traditional Windows laptop with a separate graphics card.
The Yoga 9n supports up to 64 GB of memory. Lenovo’s 15-inch Yoga Pro 9n raises that limit to 128 GB. Memory capacity matters for local AI because larger models need more memory to run efficiently.
Compact PCs for local models
Acer and Asus are also bringing RTX Spark to mini PCs roughly comparable in size to Apple’s Mac mini. Nvidia’s partners say these devices can reach up to 128 GB of memory and deliver a petaflop of AI performance. A petaflop represents a very high level of computing capability, though real-world results will depend on the AI model and workload.
The systems are aimed at running AI workflows locally, including tools that can work with documents, email, and other personal information. Local processing can reduce reliance on cloud platforms and give users more control over private data.
Pricing remains unclear for RTX Spark devices. Larsen notes that comparable high-memory systems are already expensive. Lenovo’s AMD-based ThinkCentre X Ultra, another desktop designed for local AI, is expected to start at $3,699.
Photo: Asus
Stay up to date
AI for content creation: the latest tools, tips and trends. Every two weeks in your inbox:
