Speculation Over Customer Behind SK hynix's 23 Trillion Won in Sales

Uncertainty for Semiconductor Production and Investment

Factories and Robots Also Run on Nvidia's Platform

Need to Diversify Customers and Reduce Dependence

D

As Nvidia CEO Jensen Huang faces the threat of being ousted, calls are growing to assess the risks of South Korea's accumulated dependence on Nvidia across its industries. Cooperation has deepened to encompass semiconductor sales, large-scale GPU investments, and factory and robot development, increasing the potential impact of uncertainty surrounding a single company's management on the performance and future businesses of Korean companies.


Yonhap News Agency

Yonhap News Agency

View original image

KRW 23.26 Trillion in Semiconductor Sales, 260,000 GPUs... Interconnected Sales, Investment, and Technology Cooperation

Samsung-Hynix and Hyundai Motor All Tied In... 'Kkanbu Chairman' Faces Ouster: Risks for Korean Companies? [Why&Next] View original image

According to the Financial Supervisory Service's electronic disclosure system and industry sources on November 12, SK hynix generated KRW 23.2601 trillion in revenue from a single customer last year, accounting for 23.9% of its total revenue. The company did not disclose the customer's name, but industry sources identify it as Nvidia. Samsung Electronics has also expanded its business with Nvidia by supplying sixth-generation high-bandwidth memory (HBM4) for its next-generation Vera Rubin platform and manufacturing its inference-only chip, Groq 3, on a contract basis.


Samsung Electronics Vice Chairman Jun Younghyun said in June 2026, "We will do our utmost to help Nvidia succeed and become its best partner." As cooperation deepens, changes to Nvidia's orders or product launch schedules can also affect Korean companies' production and investment plans.


Korean companies that had been component suppliers have also become major Nvidia customers. In October 2025, Nvidia agreed to supply more than 260,000 graphics processing units (GPUs), including 50,000 each to the government, Samsung Electronics, SK Group, and Hyundai Motor Group, and 60,000 to NAVER Cloud.


Hyundai Motor Group will use 50,000 GPUs to develop AI for autonomous driving and robotics, and will invest about $3 billion with Nvidia to establish domestic AI technology centers and other facilities. Samsung Electronics plans to apply AI across all stages of semiconductor manufacturing, while SK Group plans to complete an AI factory with more than 50,000 GPUs by the end of 2027.


NAVER is also linked to Nvidia through an equity stake. On November 6, Nvidia paid KRW 1.3412 trillion for newly issued shares, securing an approximately 4.13% stake in NAVER and becoming its third-largest shareholder. NAVER will invest the funds in an AI factory anchored by its Sejong data center.


Factories and Robots Also Depend on the Platform... Need to Diversify Customers and Develop Proprietary Technology


With sales, purchases, and investment all intertwined, the burden of changes in Nvidia's business strategy has also grown. If the growth in AI investment slows, Korean companies could face both declining semiconductor sales and delays in recouping their GPU investments.


Technology dependence poses a longer-term problem. Nvidia's platform is also being used to develop LG's humanoid robots and HD Hyundai's welding and painting robots. Doosan Corporation supplies copper-clad laminates (CCLs) for AI accelerators, while LS Electric and Samsung Electro-Mechanics are developing products to fit Nvidia's power systems. The broader the range of applications, the more time and money it will take to switch to other technologies.


"Korea has a great deal of the manufacturing data and semiconductor production capacity that Nvidia needs," said Heo Junyoung, a professor of economics at Sogang University. "Nvidia is likely to see that by bringing Korea into its fold, it can bring many components into its fold at once."


Korean companies need to prepare by diversifying their semiconductor customers and reducing their dependence on Nvidia for their production and development infrastructure. Even if they broaden their suppliers to include Google's and Amazon's in-house AI chips and AMD, there are limits to how much they can reduce risk if their factories and robots operate solely on Nvidia technology.


Using domestic neural processing units (NPUs) alongside GPUs is another option under consideration. SK Telecom and Rebellions are pursuing a system that combines Nvidia GPUs with domestically developed NPUs. "The goal is not simply to reduce dependence, but to select the right chips and operate them efficiently as AI services become more diverse," an SK Telecom official said.


However, the benefits of cooperation must also be considered. "Rather than spending a lot of money on trials when the market has yet to be proven, it is not too late to first test things using Nvidia's established platform and then become independent," said Lee Kyungjeon, a professor of business administration at Kyung Hee University. "What matters more is whether Korean companies have developed the capabilities to be selected by Nvidia."



Professor Heo said, "Even if our own platform is somewhat less advanced, moving forward with it is a way to bypass Nvidia and protect profit margins over the medium to long term. Otherwise, Nvidia is likely to reap the rewards."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing