Nvidia, SK Telecom and Rebellions plan AI computing system
Nvidia, SK Telecom and Rebellions signed an MOU to connect Nvidia GPUs with Rebellions NPUs through NVLink Fusion, with South Korea's science ministry supporting wider chip use.

Why it matters
- Three companies signed the MOU on Monday for SK Telecom AI data centers.
- NVLink Fusion integrates custom AI processors directly into Nvidia rack-scale infrastructure.
- AI Day Seoul lasted two days in southern Seoul.
- More than 20 Korean startups joined Nvidia's startup forum.
- Nvidia named four investment plans: Point2 Technology, Motif Technologies, RLWRLD and Upstage.
By Elena Vasquez
Nvidia, SK Telecom and Rebellions signed a three-company memorandum of understanding on Monday to connect Rebellions-designed neural processing units (NPUs) with Nvidia graphics processing units (GPUs) inside SK Telecom AI data centers.
The companies signed the pact on the margins of AI Day Seoul, a two-day event hosted by Nvidia in southern Seoul.
South Korea's ministry of science and ICT backed the approach. The ministry said the partnership could expand applications for domestic advanced semiconductors and create opportunities for local companies in global markets.
The announced structure puts Nvidia's rack-scale platform inside a Korean telecom operator's AI buildout. It also gives a domestic chip designer a route into that infrastructure.
What does the computing pact actually do?
Under the MOU, Nvidia, SK Telecom and Rebellions will cooperate on a heterogeneous AI computing system for SK Telecom AI data centers.
The system will connect Nvidia GPUs with neural processing units designed by Rebellions.
NVLink Fusion provides the technical basis for the agreement, according to Nvidia. The program lets hyperscalers and custom chip designers integrate their own AI processors directly into Nvidia's rack-scale infrastructure.
Under the announced roles, SK Telecom will use the system in its AI data centers. Rebellions will supply the NPU designs. Nvidia will supply the GPUs and NVLink Fusion.
The announcement presents the GPUs and NPUs as connected components in one computing system.
The announcement did not name an Nvidia GPU model or a Rebellions NPU model. It also gave no deployment date, capacity figure or financial value for the MOU.
That leaves execution as the main test. The three partners still need to show how the processors share work, how the system fits into SK Telecom data centers, and when operators can use it.
What did Nvidia say about Korea?
At a media roundtable after the signing, Jesse Clayton, Nvidia's principal manager for AI infrastructure, said: “Korea is a technology leader, and with AI being one of the fastest growing technologies that has ever happened ... this collaboration we are working on together, we believe, can help accelerate innovation in Korea.”
Clayton tied the agreement to a national innovation goal. His wording placed the three partners around a shared effort in Korea rather than around one vendor's hardware roadmap.
The ministry offered a more industrial rationale. It expects the partnership to widen the use of South Korean advanced semiconductors and give local companies a foothold in global markets.
Under the MOU, SK Telecom will use both processor types. Rebellions' NPUs will integrate into an Nvidia-centered rack-scale system. Nvidia's program is designed to bring custom processors into that infrastructure.
What did Nvidia announce with the startups?
Nvidia's forum brought together more than 20 Korean startups across AI foundation models, physical AI, AI infrastructure and AI chips, in cooperation with South Korea's science ministry.
The forum was intended to explore business opportunities, according to Nvidia.
Nvidia announced plans for four investments during the event.
- Point2 Technology: Nvidia planned to invest in the South Korean AI startup.
- Motif Technologies: Nvidia planned to make an additional investment.
- RLWRLD: Nvidia planned to make an additional investment.
- Upstage: Nvidia planned to make an additional investment.
The announcement did not disclose investment amounts or terms. It also did not link the four investments to the SK Telecom and Rebellions MOU.
Nvidia presented the forum separately from the trilateral MOU. More than 20 startups attended across four technical areas, and Nvidia announced four investment plans.
Why was Madison Huang's visit canceled?
Nvidia declined to comment on the abrupt cancellation of a visit by Madison Huang. She is a senior director at Nvidia and the daughter of CEO Jensen Huang.
The cancellation became a subject of questions at the media session. Nvidia also declined to comment on an alleged relationship involving Jensen Huang and a suspected Chinese spy.
At the media session, a Nvidia official said: “It is not related to the event. We cannot answer because she (Madison) did not attend the event.”
The official's answer came after a question about whether growing pressure from the U.S. Congress over the Huang scandal could affect Nvidia's partnerships with Samsung Electronics and SK hynix. The question also covered Nvidia's scheduled delivery of GPUs to South Korea.
Nvidia did not say the matter had changed those partnerships or shipments. It also gave no detail on why the visit was canceled beyond Madison Huang's absence from the event.
On Thursday, U.S. time, three Republican committee leaders requested Jensen Huang for a closed-door briefing. The request concerned his alleged relationship with a businesswoman reportedly identified as an intelligence asset for Beijing.
Nvidia subsequently said it had opened an internal probe. The company tied that decision to an email Huang sent employees, in which he admitted being “unfaithful” to his wife and took “full responsibility” for what he did.
Nvidia's announcement identified an open internal probe. It did not announce a finding into the allegation.
Two tracks ran through AI Day Seoul. Nvidia, SK Telecom and Rebellions announced a computing system, and Nvidia separately announced startup investments and held a forum. Congressional scrutiny and an internal company investigation continued around Jensen Huang.
The commercial question remained unanswered at the media session. Reporters asked whether the scandal could affect Nvidia's two South Korean chip partnerships and planned GPU deliveries. Nvidia's official declined to link either issue to the event.
What comes next for the AI system?
The MOU now has to become working infrastructure. The announcement supplies the partners, the processor types and the NVLink Fusion program, but no public deployment schedule.
The first useful milestones would be model selections, capacity details and a timetable for SK Telecom data centers, followed by evidence of how Rebellions' NPUs handle work alongside Nvidia GPUs.
Nvidia's separate investment plans add another test. The company named Point2 Technology, Motif Technologies, RLWRLD and Upstage, but did not publish the investment amounts.
South Korea's science ministry has framed the broader outcome. It expects the partnership to expand domestic advanced-chip applications and open global opportunities for local companies.
If SK Telecom deploys the mixed GPU-and-NPU system, the agreement will have moved beyond an MOU and tested whether Rebellions can take a larger role inside Nvidia-centered infrastructure.
Source: feed.koreatimes.co.kr
More from Elena Vasquez
Show full bio
Market editor covering technology and product news in artificial intelligence and machine learning at AI In Context.
446 articles
Related articles
- Rebellions Plans R200 Integration for Korean Sovereign AI
- NVIDIA expands Korean sovereign AI work with Rebellions racks
- NVIDIA backs Point2 and plans three Korean AI startup investments
- Samsung and SK Join OpenAI's Stargate in Korea AI Push
- NVIDIA to invest up to $100 billion in OpenAI for 10 gigawatts of AI infrastructure