NVIDIA RTX Spark Seoul was the stage for a press event on November 6, 2026, at the JW Marriott Hotel. The demonstration highlighted the platform’s potential for running local AI agents, content creation, and gaming with DLSS 5.
What happened?
Jeff Yen, NVIDIA’s APAC Technical Marketing Director, presented the RTX Spark architecture, which combines a Blackwell RTX GPU with up to 6,144 CUDA cores and a Grace CPU with up to 20 cores, connected via NVLink-C2C. The platform offers up to 128 GB of unified memory and 1 petaflop of AI performance at FP4.
Demos included local AI agents for video editing, email organization, and virtual robotics development, all without cloud dependency. In gaming, NBA 2K27 and AION 2 were shown running with DLSS 5, Super Resolution, and ray tracing. According to NVIDIA, 97 of the top 100 popular games run on the platform at launch.
Why does this matter?
RTX Spark represents a step toward powerful local AI PCs, reducing cloud token costs and enabling privacy. Support for DLSS 5 with 3D-Guided neural rendering brings visual improvements in games even on compact laptops. The Seoul event reinforces NVIDIA’s focus on Asian markets and local collaborations.
What changes in practice?
Users with RTX Spark notebooks can run AI models locally for creation and development tasks, as well as play recent titles with RTX technologies. More accessible models start around $2,400 to $2,600 globally, with higher-memory options being more expensive.
Sources: Inven Global and NVIDIA event reports.
By Grok
