Nvidia is targeting a new market with RTX Spark, a system-on-chip (SoC) designed for artificial intelligence (AI) PCs.

The company is targeting the on-device AI market, enabling Windows PCs to run large AI models and AI agents locally.

Asus, Dell, HP, Lenovo, Microsoft and MSI began selling RTX Spark-powered laptops on Oct. 16.

"Nvidia has combined its expertise in gaming PCs, laptops, workstations and supercomputers to create a single superchip, RTX Spark," Nvidia Vice President Ida Keita said at the RTX Spark launch event held at COEX in Seoul on Oct. 16. "We are approaching an era when everyone will have a personal AI agent, and RTX Spark will enable secure on-device AI without token usage limits."

The least expensive RTX Spark laptop currently available in Korea is the MSI Prestige N16, priced at 5.799 million won. Nvidia attributed the high price to rising memory costs. "When we were developing the product, 128GB of memory was reasonably affordable, and we considered it the optimal configuration," an Nvidia official said at the event. "We didn't expect memory prices to rise this much."

RTX Spark is an SoC for Windows PCs that Nvidia officially announced in June. It combines an Arm-based 20-core Grace central processing unit (CPU) and a Blackwell-architecture graphics processing unit (GPU) in a single package, connecting them through NVLink-C2C chip-to-chip interconnect technology.

The CPU was designed by Taiwan's MediaTek, while the GPU features 6,144 CUDA cores and fifth-generation Tensor Cores. AI computing performance reaches 1 petaflop (PFLOP) at FP4 precision, with total thermal design power (TDP) of approximately 80 watts. Nvidia said the integrated design consumes less power than configurations using separate CPUs and discrete GPUs, allowing it to fit into thin laptops.

RTX Spark supports up to 128GB of LPDDR5X memory. Conventional discrete graphics cards struggle to run large AI models independently because of graphics memory (GDDR7) capacity limitations. By allowing the CPU and GPU to share a large memory pool, RTX Spark can run open AI models with approximately 70 billion parameters and AI agents locally on a PC.

Beyond AI, the chip supports content creation and gaming. Nvidia said its TensorRT inference acceleration engine reduces post-production time in creative applications such as Adobe Photoshop, Premiere and Substance.

The platform also supports 8K video editing and rendering of large 3D scenes of up to 90GB without an external power connection. It supports Nvidia gaming technologies, including DLSS 5's 3D-guided neural rendering, and can run the latest AAA games at more than 100 frames per second (FPS) at a resolution of 2560×1440.