Rebellions to Use NVIDIA NVLink Fusion for Korean Sovereign AI
What happened
On November 8, 2026, NVIDIA, Rebellions, and SK Telecom announced a partnership to integrate Rebellions' upcoming R200 neural processing units (NPUs) with NVIDIA's NVLink Fusion and MGX architecture. The collaboration aims to build new AI infrastructure for Korea, with deployment targeted for 2028. SK Telecom will use these systems in its AI data centers, connecting them to its proprietary AI models and services. The project is supported by Korea's Ministry of Science and ICT as part of a broader effort to strengthen domestic AI capabilities.
Key details
- Rebellions' R200 NPU is still in development and targets low-latency, energy-efficient inference for large AI models.
- NVLink Fusion will connect Rebellions' NPUs with NVIDIA GPUs in a disaggregated-inference setup, where each processor handles different stages of AI response generation.
- SK Telecom plans to deploy the new systems in its AI data centers and link them to its foundation models, including A.X K2.
- The Korean government, through the Ministry of Science and ICT, is backing the collaboration to help domestic NPUs enter the AI infrastructure market.
- Deployment of the new systems is planned for 2028, following completion of processor and rack system development.
Why it matters
Korean AI companies, including Rebellions and SK Telecom, will gain a pathway to use domestically developed NPUs alongside NVIDIA GPUs in large-scale AI data centers. This arrangement could give SK Telecom and its partners more control over their AI infrastructure and potentially improve efficiency and performance for Korean-language and industry-specific AI services. The government support also provides Korean chipmakers with an opportunity to compete in the AI hardware market, which could affect procurement and technology decisions for data center operators and service providers in Korea.