Rebellions Plans R200 Servers With NVIDIA and SK Telecom for Korean AI Infrastructure
The AMW Read
The report adds implementation details to an already covered partnership, with potential implications for accelerator deployment that remain untested.
Rebellions Plans R200 Servers With NVIDIA and SK Telecom for Korean AI Infrastructure
Rebellions, NVIDIA, and SK Telecom have signed an agreement to develop Korean AI infrastructure using Rebellions' planned R200 neural processing unit. The companies intend to connect R200 chips to NVIDIA's platform through NVLink Fusion and use NVIDIA's MGX server designs. SK Telecom plans to test the combined system in its AI data centers. The companies expect R200-based systems to become available from 2028.
The agreement shows how a local chip designer can pursue sovereign AI infrastructure while relying on an established foreign hardware ecosystem. Rebellions says using NVIDIA's interconnects, rack designs, and supply chain could reduce the time and capital needed to build its own systems. The arrangement also extends NVIDIA's infrastructure ecosystem to another company's accelerator. Rebellions and SK Telecom aim to build on earlier work with Rebellions' Atom Max servers, which already support several SK Telecom services.
For infrastructure buyers and investors, the test is whether the proposed division of inference work delivers practical gains. The companies plan to use NVIDIA GPUs for prefill, which processes input prompts, and R200 NPUs for decode, which generates output tokens. They have yet to validate the resulting throughput and power efficiency in service. The investment amount, data-center scale, and final customers remain undisclosed. This adds technical detail to AI Market Watch's earlier coverage of the partnership, rather than establishing that an R200 system is already deployed.
