How will Nvidia, SK Telecom and Rebellions connect their AI chips?

The companies plan to connect Nvidia’s graphics processing units with neural processing units designed by Rebellions, using Nvidia’s NVLink Fusion program. The system is intended for use in building SK Telecom AI data centers.

How could the Nvidia, SK Telecom and Rebellions agreement affect Korea’s NPU developers?

The science ministry said connecting locally produced NPUs to Nvidia’s architecture could broaden the applications for domestic advanced semiconductors and give local companies an opportunity to verify their technologies in the global market. It said it would support extending collaboration beyond Rebellions to other local NPU developers.

Nvidia, SKT, Rebellions to jointly establish heterogeneous computing system for AI

Listen

Summary

Nvidia, SK Telecom and Rebellions agreed Monday to connect Nvidia GPUs with Rebellions NPUs in a heterogeneous AI computing system for SK Telecom data centers. The companies signed an MOU on the sidelines of Nvidia’s two-day AI Day Seoul event in southern Seoul. The collaboration will use Nvidia’s NVLink Fusion program and is intended to expand applications for domestic advanced semiconductors and help local companies verify their technologies in global markets. The science ministry said it would support extending such cooperation to other Korean NPU developers.

Key Facts

NVLink Fusion allows hyperscalers and custom chip designers to integrate their AI processors into Nvidia’s rack-scale infrastructure.
The science ministry said the collaboration could help develop Korea’s advanced-chip ecosystem.
The ministry said it would support local NPU developers beyond Rebellions in pursuing similar collaborations.

By Yonhap

Published Nov 9, 2026 2:25 pm KST

Participants visit booths at the AI Day Seoul event organized by the U.S. chip giant Nvidia in southern Seoul, Monday. Yonhap

Nvidia, SK Telecom and domestic artificial intelligence (AI) chip startup Rebellions agreed Monday to combine their respective strengths to establish a heterogeneous computing system, Seoul's science ministry said.

Under the trilateral memorandum of understanding (MOU), the companies will cooperate in the establishment of a heterogeneous AI computing system connecting Nvidia's graphic processing units (GPUs) with neural processing units (NPUs) designed by Rebellions that will be used in the building of SK Telecom AI data centers. The tie-up will be based on the U.S. chipmaker's NVLink Fusion program, the ministry added.

The NVLink Fusion program allows hyperscalers and custom chip designers to integrate their own AI processors directly into Nvidia's rack-scale infrastructure.

The MOU was signed on the margins of AI Day Seoul, a tech event hosted by Nvidia that kicked off its two-day run in southern Seoul.

The agreement to connect locally produced NPUs to Nvidia's architecture will help expand the scope of application for domestic advanced semiconductors. It will also provide an opportunity for local companies to verify their technologies in the global market, the science ministry explained.

The science ministry vowed to support local NPU developers to ensure such collaboration could extend beyond Rebellions to other local NPU developers, and eventually help develop Korea's advanced-chip ecosystem.

Explore More