[Editorial] Nvidia-Driven Uncertainty Grows... Urgently Review Reliance on HBM and GPUs

Nvidia CEO Jensen Huang. Photo by Kang Jinhyung.

Uncertainty triggered by the controversy surrounding the personal life of Jensen Huang, CEO of Nvidia, is becoming a real risk. U.S. media have reported that Nvidia's board is preparing to remove Huang. His future has not yet been decided, and no evidence has been confirmed that company secrets or U.S. national security have been compromised. Still, the uncertainty surrounding the future of a pivotal figure and a central company in the global semiconductor and artificial intelligence (AI) industries is cause for concern. Nvidia is a key customer for Korean companies' high-bandwidth memory (HBM), and a major supplier of graphics processing units (GPUs) and development software to Korea's AI industry. Both companies that sell semiconductors and those that build AI infrastructure are affected by Nvidia's business strategy. We must carefully consider whether our cooperation with Nvidia, which has created important opportunities for Korea's AI industry, could instead become a source of risk. Above all, it is important to build an industrial foundation that is not dependent on the fate of any particular company or executive.

Ensure Lasting Cooperation Beyond 'Kkanbu' Ties

In October 2025, Jensen Huang met with Samsung Electronics Chairman Lee Jae-yong and Hyundai Motor Group Executive Chair Euisun Chung in a gathering of close friends, or "kkanbu." In July 2026, he also met with President Lee Jae-myung. But personal ties do not guarantee the continuity of business cooperation. Companies should check in advance whether business initiatives planned with Nvidia, such as GPU adoption and joint development, are being properly carried out, and whether a change in leadership could become a variable. At the same time, they should prepare for changes in the business environment by maintaining cooperation with Nvidia while expanding their HBM customer base to include AMD and Google. Nvidia's plan to establish a research and development (R&D) center in Korea should also be monitored. In preparation for a major change in Nvidia's leadership, the Ministry of Science and ICT should communicate with Nvidia about the status of GPU adoption and whether the center's establishment is proceeding as planned.

China's AI Development Continues Despite Tighter Export Controls

Separate from Huang's future, another factor to watch is the U.S. move to tighten export controls on China. According to procurement records from Northwestern Polytechnical University in China, the university, which is under U.S. sanctions, purchased Nvidia H200 chips through a domestic company in October 2026. It also signed a contract to rent computing capacity from servers equipped with H200, A100, and A800 chips. Although Nvidia's direct involvement has not been confirmed, the case shows how difficult it is to control end users and remote computing. We must prepare for the possibility that the United States will use this as a reason to further tighten technology controls. China, too, is easing constraints by developing its own AI chips through companies such as Huawei and optimizing software. Even if regulations are tightened, China's technological development will not stop. The Ministry of Trade, Industry and Energy and the Ministry of Foreign Affairs should consult with the United States on the scope of regulations governing chip tracking and remote computing, and assess the impact on Korea's data center and cloud industries.

Tighter U.S. Sanctions on China Will Not Automatically Benefit Korea

If the United States tightens sanctions on China, it may become harder for China to secure advanced chips. But that will not automatically benefit Korea. If sales of Nvidia GPUs to the Chinese market decline, demand for the HBM used in those products will also be affected. China's efforts to develop its own semiconductor and memory supply chains are also a long-term concern. If companies cannot make up for lost orders in other markets, tighter regulations could instead result in losses. Companies should assess orders, delivery schedules, and end users under different export-control scenarios.

That does not mean the opportunities have disappeared. AMD, Nvidia, and Intel have recently held a series of AI events in Seoul. These events show their interest in the Korean market, but they do not necessarily mean that supply contracts will follow. Korea must turn that interest into actual customer acquisition and technological cooperation. The government should also support efforts to expand demonstrations of domestically developed AI semiconductors and software, broadening the range of technological options.

Ryu Je-myeong, Second Vice Minister of Science and ICT, described the uncertainty in the U.S. AI investment environment as an opportunity for Korea to catch up with leading countries and announced plans to invest 9.4 trillion won in the AI sector next year, in 2027. The opportunity should be seized now, while Nvidia is facing uncertainty, by broadening the range of customers and technological options. It is at least fortunate that the lesson that Korea must not become overly dependent on a single company or business leader for its semiconductor and AI strategies has come to us before it is too late.

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