NVIDIA, Rebellions and SK Telecom Team Up to Build Korea's Sovereign AI Infrastructure

NVIDIA, South Korean AI chipmaker Rebellions and telecom giant SK Telecom announced on November 9, 2026, a collaboration to build sovereign AI infrastructure for South Korea, pairing NVIDIA's open computing platform with Korean-designed processors and SK Telecom's data centers. The project, unveiled at NVIDIA AI Day in Seoul, is backed by Korea's Ministry of Science and ICT and aims to give the country AI systems built around its own language, industries and priorities.

Modern AI data center in Seoul with rows of server racks and engineers

What the NVIDIA, Rebellions and SK Telecom Deal Means

Direct answer: NVIDIA, Rebellions and SK Telecom will jointly develop rack-scale AI inference systems for South Korea, using NVIDIA NVLink Fusion to connect Rebellions' next-generation R200 neural processing units (NPUs) to NVIDIA's platform. SK Telecom plans to deploy the systems in its AI data centers, and the first R200-based systems are expected to be ready for deployment in 2028.

The announcement, made at NVIDIA AI Day at the COEX convention center in Seoul, centers on a simple idea: combine the best of both worlds. Rebellions' R200 chips are designed for fast, energy-efficient AI inference, while NVIDIA supplies the high-bandwidth NVLink Fusion interconnect and the MGX rack architecture that lets different processors work together at scale. SK Telecom brings the data centers, its own AI models and the public-facing services that will run on the hardware.

NVLink Fusion and the R200: How the Technology Fits Together

NVLink Fusion is NVIDIA's technology for letting chipmakers connect their own custom processors to NVIDIA's infrastructure, sharing its high-speed scale-up networking and established ecosystem of server builders and component suppliers. For Rebellions, that means its R200 NPUs can sit in the same rack-scale systems as NVIDIA GPUs, splitting inference work between the two kinds of silicon.

Rebellions describes the R200 as optimized for low-latency, energy-efficient inference on frontier models, including multitrillion-parameter mixture-of-experts models — the giant AI models that power today's most capable chatbots and agents. The systems will be built on NVIDIA's MGX rack design, a modular server standard that lets manufacturers assemble AI servers more quickly.

The deal is not a one-off hardware purchase. The collaboration is supported by Korea's Ministry of Science and ICT (MSIT), and it forms part of a broader effort under Korea's "AI for All" project to connect domestic semiconductor innovation with the infrastructure needed to run AI at national scale. For readers tracking NVIDIA's busy year, the company has also been in the headlines for a separate governance story — see our report on NVIDIA's board opening an independent investigation into CEO Jensen Huang.

SK Telecom's Role: Data Centers, Models and Services

SK Telecom plans to deploy the R200-based systems in its AI data centers, drawing on its operations expertise and its proprietary foundation models, including A.X K2. The goal is full-stack sovereign AI: infrastructure, models and services that are developed in Korea, for Korean users and businesses.

That full-stack vision matters because sovereign AI is about more than owning servers. Countries pursuing it want AI that understands their language, reflects their culture and keeps sensitive data and workloads under domestic control. Korea's government has made that a national priority, and the MSIT's backing signals that this collaboration is part of a wider state-supported push rather than a purely commercial deal.

According to reporting on the announcement, NVIDIA is also working with three selected consortia and four Korean foundation-model builders under the AI for All initiative, with deployment recipes covering SK Telecom's A.X K2, LG AI Research's K-EXAONE 2.0, Upstage's Solar Open 2 and Motif Technologies' Motif-3. Explore related technology offers and tools here while following the AI infrastructure race.

Close-up of an advanced AI computer chip with glowing green and gold circuitry

When Will the R200 Systems Be Ready?

Rebellions' R200 systems are expected to be ready for deployment in 2028, according to the joint announcement. That timeline reflects the work still ahead: the R200 is a next-generation chip, and integrating it into rack-scale systems with NVIDIA's platform requires processor development, system engineering and testing before SK Telecom can put the hardware into live service.

No financial terms were disclosed, and the announcement does not say how many systems SK Telecom will install or give a firm date for live service. What it does establish is the direction of travel: Korea wants a domestic AI hardware ecosystem, and NVIDIA wants its interconnect and rack standards at the center of it — a pattern the company is repeating in sovereign AI deals around the world.

Why Sovereign AI Matters Beyond Korea

Sovereign AI has become one of the biggest themes in the global AI race. Governments from Europe to the Middle East and Asia are investing in domestic compute capacity, local AI models and national data infrastructure, driven by concerns about data control, economic competitiveness and access to cutting-edge chips. Korea, home to world-leading semiconductor manufacturing, is well placed to pursue that goal — and this deal gives its domestic chip designers a direct route into NVIDIA's ecosystem.

For the US tech audience, the deal is also a reminder of how the AI supply chain is globalizing. The next wave of AI infrastructure may not be built from a single vendor's chips, but from mixed systems that combine GPUs, custom NPUs and specialized accelerators — all linked by common interconnect standards. If the R200 systems hit their 2028 target, Korea could become a showcase for that heterogeneous model.

Frequently Asked Questions

What did NVIDIA, Rebellions and SK Telecom announce on November 9, 2026?

The three companies announced a collaboration to advance Korea's sovereign AI infrastructure, using NVIDIA NVLink Fusion to connect Rebellions' next-generation R200 NPUs to NVIDIA's open infrastructure platform, with SK Telecom deploying the systems in its AI data centers. The project is supported by Korea's Ministry of Science and ICT.

What is NVIDIA NVLink Fusion?

NVLink Fusion is NVIDIA's technology that lets chipmakers connect their own custom processors to NVIDIA's computing platform, using its high-bandwidth, low-latency scale-up networking and established server ecosystem. It allows mixed GPU-NPU systems to be built at rack scale.

What is Rebellions' R200 chip?

The Rebel R200 is Rebellions' next-generation neural processing unit (NPU), designed for low-latency and energy-efficient AI inference, including on multitrillion-parameter mixture-of-experts models. It is a Korean-designed AI processor.

When will the R200 systems be ready?

Rebellions' R200 systems are expected to be ready for deployment in 2028. No firm date for live service has been announced, and no purchase volumes or financial terms were disclosed.

What is sovereign AI?

Sovereign AI refers to computing infrastructure, AI models and services that a country develops around its own language, industries and priorities, keeping data and workloads under domestic control. Korea's effort runs under its "AI for All" initiative.

What is SK Telecom's role in the collaboration?

SK Telecom plans to deploy the R200-based systems in its AI data centers, using its data-center operations expertise, proprietary models such as A.X K2, and public-facing services to deliver AI to Korean users and businesses.

Search This Blog

TrendWire Daily