NVIDIA, Rebellions and SK Telecom Plan Sovereign AI Infrastructure for Korea

NVIDIA, Rebellions and SK Telecom announced a collaboration to advance Korea's sovereign AI infrastructure, according to a joint announcement dated November 8, 2026. Under the agreement, Rebellions will use NVIDIA NVLink Fusion to connect its Rebel R200 neural processing units to the NVIDIA platform and use the NVIDIA MGX rack architecture for rack-scale system development. SK Telecom plans to deploy these systems. Rebellions' R200 systems are expected to be ready for deployment in 2028. NVIDIA Newsroom report.

A Planned Three-Party Collaboration Backed by Korea's MSIT

The announcement describes a collaboration among NVIDIA, Korean AI chip developer Rebellions and telecom operator SK Telecom to advance what the companies call sovereign AI infrastructure.

The participants frame the work around AI services tailored to Korea's language, industries and priorities. The companies describe the arrangement as a collaboration to advance sovereign AI infrastructure, with MSIT supporting work between NVIDIA and Korean NPU companies beginning with Rebellions and SK Telecom. These are vendor characterizations of the collaboration's aims rather than reported outcomes.

Rebellions to Connect R200 NPUs Through NVLink Fusion

As part of the agreement, Rebellions will use NVLink Fusion to connect its next-generation R200 NPUs to the NVIDIA platform. The company says the R200 is being optimized for low-latency and energy-efficient inference using frontier models, including multitrillion-parameter mixture-of-experts models. Rebellions will connect R200 NPUs within a rack through NVIDIA NVLink scale-up networking.

NVIDIA describes NVLink Fusion as extending its vertically integrated, horizontally open AI platform to custom XPUs, letting chipmakers combine specialized silicon with the company's high-bandwidth, low-latency scale-up networking and infrastructure ecosystem. NVIDIA asserts this can help improve performance, reduce deployment risk and accelerate time to market. These benefits are presented as company claims.

SK Telecom Plans Deployment in Its AI Data Center

SK Telecom plans to deploy these systems in its AI data center, drawing on its AI data center operations experience, proprietary foundation models and public-facing services. Those services include work under Korea's AI for All project.

The companies plan to develop an approach to disaggregated inference, assigning different stages of AI response generation to the architecture best suited to each task. Their planned work includes using NVIDIA GPUs for prefill and Rebellions NPUs for decode, which the companies say would improve inference throughput and power efficiency. These are planned arrangements, not delivered deployments.

Planned Timelines and Forward-Looking Caveats

Rebellions' R200 systems are expected to be ready for deployment in 2028. SK Telecom's forward-looking statements cover the planned development, integration and deployment of AI infrastructure and systems, expected performance and benefits, and future collaboration. The company cautions that actual results may differ materially due to risks including technological and operational challenges and the availability of necessary resources.

NVIDIA's forward-looking statements likewise address expectations regarding its products and third-party arrangements. The company cautions that these statements are not guarantees of future performance. Organizations assessing the collaboration should distinguish the planned readiness date from a completed deployment and evaluate the stated benefits as vendor claims.

What This Means for Practitioners

For infrastructure planners, the announcement illustrates a stated model in which domestically designed processors are connected to a commercial accelerator platform through networking and a shared rack architecture. Practitioners tracking the collaboration should monitor the planned R200 readiness, expected in 2028, and the proposed GPU-plus-NPU inference split. Both are described in the context of future development and deployment.

Buyers and partners should separate plans from verified deployment and seek technical validation of vendor performance claims. The companies assert performance, efficiency and time-to-market benefits, while SK Telecom and NVIDIA forward-looking statements acknowledge technological, operational and resource risks that could change these plans.