NVIDIA, Rebellions and SK Telecom announce NVLink Fusion sovereign AI collaboration

Background

Sovereign AI is AI infrastructure that a country or organization develops and operates for its own use. Rebellions is a South Korean AI chip company ^[1]. NVLink Fusion is the NVIDIA technology through which Rebellions will connect its chips to NVIDIA's platform ^[1].

What was announced

On 8 November 2026 NVIDIA, Rebellions and SK Telecom announced a collaboration on sovereign AI infrastructure built with NVIDIA NVLink Fusion ^[1]. Rebellions will connect its Rebel R200 neural processing units (NPUs) to NVIDIA's platform through NVLink Fusion, and will use the NVIDIA MGX rack architecture for rack-scale systems ^[1]. Inside a rack, the R200 NPUs will connect through NVIDIA NVLink scale-up networking ^[1]. An NPU is a processor designed to run neural network calculations, which sets it apart from a general-purpose CPU.

Technical plan

Rebellions says the R200 is optimized for low-latency, energy-efficient inference, including multitrillion-parameter mixture-of-experts models ^[1]. Inference is the stage in which a trained model produces answers, as opposed to the training stage that comes before it. A mixture-of-experts model splits its capacity among many specialist sub-networks and activates only some of them for each input.

The companies plan disaggregated inference, with NVIDIA GPUs handling prefill and Rebellions NPUs handling decode ^[1]. Prefill processes the full input a user provides, while decode generates the output one token at a time. Disaggregated inference means running these two stages on different chips.

Deployment and partners

SK Telecom plans to deploy the systems in its AI data center and connect them to its proprietary models, including A.X K2, and to public services under South Korea's AI for All project ^[1]. The R200 systems are expected to be ready for deployment in 2028 ^[1].

South Korea's Ministry of Science and ICT supports the collaboration, and Deputy Prime Minister Bae Kyung-hoon said the government "will actively support this collaboration" ^[1]. Rebellions CEO Sunghyun Park said the announcement "marks a turning point, opening a new chapter for sovereign AI" ^[1].

Significance

The collaboration brings together a chip company's processors, NVIDIA's networking and rack platform, and a telecom operator's data center and services ^[1]. It is a concrete example of the sovereign AI idea, with the hardware, the operator and government backing all named in the announcement ^[1].