Rebellions, Nvidia to Build Sovereign AI Infrastructure in Korea

What happened

Rebellions is partnering with Nvidia and SK Telecom (SKT) to plan a Korean-standard sovereign AI computing infrastructure based on next-generation neural processing units (NPUs). Rebellions CEO Park Sung-hyun unveiled the plan at Nvidia AI Day Seoul 2026 on Nov. 9, and the three companies signed a memorandum of understanding (MOU) the same day. The project will connect Rebellions' NPUs with Nvidia's open infrastructure platform through NVLink Fusion, with the resulting systems to be deployed and tested at SKT's AI data centers. Rebellions will develop a rack-scale AI inference system using its next-generation R200 NPU and Nvidia's MGX server rack architecture; the R200 succeeds the R100 NPU, which is equipped with fifth-generation high-bandwidth memory (HBM3E), and is being designed for low-latency, high-efficiency inference of frontier AI models, including mixture-of-experts (MoE) models. Rebellions aims to begin supplying R200 systems in 2028. SKT will deploy the system in its AI data centers for testing under real-world operating conditions. The three companies aim to establish a "Sovereign AI Token Factory" and plan to introduce standards applicable across public institutions and the private sector to accelerate market adoption.

Impact and signal

Affected components and suppliers

Source and time

Source: The Elec · 11/9/2026 · View original source