NVIDIA, Rebellions and SK Telecom Announce Collaboration to Advance Sovereign AI Infrastructure With NVIDIA NVLink Fusion
AI-assisted digest, checked against the original source before publishing. How we work
NVIDIA partners with Rebellions and SK Telecom to build Korea's sovereign AI infrastructure using NVLink Fusion and MGX rack architecture.
NVIDIA, Korean AI chip startup Rebellions, and SK Telecom announced a collaboration to advance Korea's sovereign AI infrastructure, anchored on NVIDIA's NVLink Fusion interconnect technology. Korea's Ministry of Science and ICT (MSIT) supports the partnership, which aims to link domestic semiconductor development with the systems and operational capabilities required to run AI at scale.
Under the agreement, Rebellions will use NVLink Fusion to connect its next-generation Rebel R200 neural processing units (NPUs) to NVIDIA's open infrastructure platform, and will leverage the NVIDIA MGX rack architecture for rack-scale system development. The goal, per Rebellions cofounder and CEO Sunghyun Park, is to deliver low-latency, energy-efficient inference for frontier models at scale by pairing Rebellions' purpose-built silicon with NVIDIA's high-speed interconnects and rack-scale design expertise.
SK Telecom plans to deploy these systems in its AI data center, drawing on its data center operations experience, proprietary models — including A.X K2 — and public-facing AI services developed under Korea's "AI for All" project. The carrier frames the work as full-stack sovereign AI spanning infrastructure, models and services, per CEO Jung Jai-hun, who called the collaboration validation of next-generation sovereign AI infrastructure combining global technology with Korean AI semiconductors.
NVIDIA founder and CEO Jensen Huang said the collaboration gives nations the power to turn their language, culture and expertise into intelligence, adding that NVLink Fusion lets Korea's chipmakers connect custom ASICs to NVIDIA's AI factory platform. The announcement also positions NVLink Fusion as a bridge for third-party silicon makers seeking access to NVIDIA's rack-scale ecosystem rather than competing against it wholesale.
What remains to watch: whether the R200 NPUs meet performance and efficiency targets once integrated into MGX racks, and whether SK Telecom's deployments deliver on the full-stack sovereign AI ambition across infrastructure, models and public services. Equally open is the scope of MSIT's ongoing support and how the partnership might scale beyond Korea's borders.
A parallel question is how NVLink Fusion's openness reshapes the competitive landscape for other AI silicon startups, and whether more nations pursue similar government-backed tie-ups between domestic chipmakers, telecom operators and NVIDIA's platform.
Key facts
- • Rebellions to connect R200 NPUs to NVIDIA platform via NVLink Fusion.
- • NVIDIA MGX rack architecture to support rack-scale system development.
- • SK Telecom to deploy systems in its AI data center.
- • Partnership backed by Korea's Ministry of Science and ICT.
- • SK Telecom models include A.X K2, under 'AI for All' project.
Why it matters
NVLink Fusion gives AI silicon startups a path to plug custom NPUs into NVIDIA's rack-scale ecosystem instead of building full interconnect stacks alone. For builders, expect more government-backed sovereign AI deals pairing domestic chips with NVIDIA platforms. The R200's real-world latency and efficiency numbers in MGX racks will be the first test of whether the model works.
Spotted an error? Send a correction.
Get the day's AI news in one email
The Daily Epoch · 60-second read · unsubscribe anytime
