NVIDIA, South Korean chip designer Rebellions and telecom operator SK Telecom announced a collaboration on November 9 to build sovereign AI infrastructure using NVIDIA's NVLink Fusion technology NVIDIA Newsroom1.
Rebellions plans to use NVLink Fusion to connect its next-generation Rebel R200 neural processing units (NPUs) to the NVIDIA MGX rack architecture NVIDIA Newsroom1. NVLink Fusion allows third-party accelerators, such as custom ASICs, to communicate with NVIDIA GPUs within the same rack NVIDIA Newsroom1. This enables a "disaggregated inference" approach where different stages of AI model processing are assigned to the best-suited processor NVIDIA Newsroom1. The companies stated they intend to use NVIDIA GPUs for the "prefill" phase of large language model generation and Rebellions NPUs for the "decode" phase NVIDIA Newsroom1.
SK Telecom intends to deploy the hybrid systems in its AI data centers to support its "full-stack sovereign AI" strategy NVIDIA Newsroom1. According to secondary reporting, SK Telecom plans to launch a gigawatt-scale AI factory in Korea built on NVIDIA's DSX AI Factory architecture, with an operational target of 2027 CryptoBriefing2. The collaboration is supported by South Korea's Ministry of Science and ICT NVIDIA Newsroom1. No independent performance benchmarks or pricing details for the resulting hybrid systems have been published.