[Lee Seong-yeop, Professor at the Graduate School of Technology Management, Korea University] The sudden resignation of Jensen Huang, founder of NVIDIA, is sending shockwaves beyond the artificial intelligence (AI) industry into the realms of national security and the competition for technological supremacy. As investigations into allegations of ties to Chinese intelligence agencies continue, the executives and core technological information of AI semiconductor companies have emerged as targets in the competition for national security among nations.

This situation also serves as a warning to South Korea, which is deeply involved in the global AI supply chain. It underscores the need to go beyond competitiveness in high-bandwidth memory (HBM) and develop independent AI development and operation capabilities, as well as supply chain resilience.

Security Controversy Extends to AI Company CEOs… Technology Equals National Competitiveness

On the 12th (local time), NVIDIA announced that founder Jensen Huang had stepped down from his roles as president, CEO, and board member. This came just four days after the board launched an independent investigation amid allegations that the company had been targeted by foreign intelligence activities. Chief Financial Officer (CFO) Colette Kress has assumed the role of interim CEO.

U.S. authorities are investigating allegations that Alice Huang, a former executive at Megaspede—a NVIDIA client—had ties to Chinese intelligence agencies. China has denied these allegations, and no actual intelligence activities or leaks of confidential information have been confirmed. NVIDIA also maintains that there is no basis to conclude that national security or company secrets have been compromised. The U.S. Congress is also examining the allegations and NVIDIA’s security systems.

This incident demonstrates that security issues involving the executives of AI companies have expanded into the realm of national security. Whereas diplomatic and military secrets and nuclear weapons technology were once the primary targets of international espionage, the scope has now broadened to include the executives and core technical information of AI semiconductor companies.

NVIDIA’s GPUs and the CUDA ecosystem form the core foundation for global AI development and operations. The performance and development schedules of next-generation chips, supply chain bottlenecks, and plans for the allocation of computational resources constitute strategic intelligence that can gauge the technological level and response capabilities of rival nations. This is why ensuring the security of AI companies has become a critical task for safeguarding a nation’s industrial competitiveness and defense capabilities.

We must also be wary of conflicts between corporate interests and national policy. While NVIDIA seeks to expand its presence in the Chinese market, the U.S. government aims to restrict the military use of advanced semiconductors and prevent technology leaks. It is essential to ensure that national security policy is not swayed by the interests of specific companies, while simultaneously strengthening the security and compliance frameworks of key companies.

In the global AI industry, debates continue over the profitability of massive investments in GPUs and data centers. However, a nation’s AI capabilities cannot be evaluated solely based on short-term profitability. Stable computing power is essential for the military to analyze sensitive information within its own environment and for public institutions to maintain essential services even during crises.

A lack of AI capabilities could lead to falling behind not only in the development of cutting-edge models but also in productivity competition across manufacturing, finance, and biotechnology, as well as in cyber defense. This is the backdrop against which the United States and China are investing heavily in AI despite the enormous costs involved.

U

.S

.-China Semiconductor Control Conflict… A Double-Edged Sword for South Korea’s HBM Industry

Alongside the Jensen Huang incident, another key factor to watch is the U.S. export controls on AI semiconductors to China. While it is uncertain whether this incident will directly lead to tighter regulations, the U.S. policy stance aimed at restricting the flow of advanced semiconductors into China is already affecting South Korea.

While tighter U.S. regulations could limit China’s access to cutting-edge AI computing power, they would also make it more difficult for U.S. companies to access the Chinese market. China is likely to accelerate the development of its own AI chips and software. If the U.S. and Chinese technology ecosystems diverge, the impact could extend beyond semiconductors to include cloud services and AI development tools.

For South Korea, this presents both an opportunity and a risk. HBM from SamsungElectronics(005930)and SK hynix(000660)is a core component of the U.S.-centered AI supply chain. As collaboration with GPU design firms expands during the development of next-generation HBM, South Korean companies have an opportunity to increase their influence not only in memory supply but also in co-design.

On the other hand, if the U.S. tightens export controls on AI accelerators bound for China, demand for related HBM could decline. However, increased investment in data centers by the U.S. and its allies could offset this decline. In the medium to long term, China’s localization of AI semiconductors and memory could threaten the market position of Korean companies.

Uncertainty surrounding local production facilities in China is also growing. The U.S. has revoked the Verified End-User (VEU) status of SamsungElectronics’ and SK hynix’s Chinese factories and switched to an annual licensing system for equipment imports starting in 2026. If licensing conditions are tightened, this could impose constraints on capital investment and process advancement.

South Korea must extend its technological edge in HBM to joint design of next-generation memory, system semiconductors, and competitiveness in AI infrastructure.

Beyond Securing GPUs, We Must Develop Our Own AI Chips, Models, and Cloud Infrastructure

The AI sovereignty that South Korea should pursue does not mean localizing all technologies. Rather, it involves participating in the global technology ecosystem while securing independent operational capabilities and autonomy to ensure that core functions are not disrupted by decisions made by specific countries or companies.

To this end, three tasks are critical.

First, South Korea must secure the AI computing capacity required for defense, public sector, and core research domestically. This goes beyond simply purchasing GPUs; it requires establishing data centers, power infrastructure, communication networks, and a skilled workforce. A minimum sovereign AI foundation is necessary to maintain essential national functions even if external supplies are cut off.

Second, we must develop our own AI chips, models, and cloud capabilities. Even when utilizing foreign GPUs, domestic technical teams must be able to verify and improve AI models. Technical flexibility—the ability to migrate data and operations to a different environment should the service terms of a specific cloud provider change—is also crucial.

A heterogeneous computing platform (XPU) that combines domestically produced neural processing units (NPUs) with foreign GPUs is another alternative. Representative examples include the NVLink Fusion collaboration between NVIDIA,SKTelecom(017670), and Rebellion; the AI software ecosystem being built by AMD, More, and LabUp; and the heterogeneous AI semiconductor collaboration between Intel and DeepX.

These collaborations represent Korea’s efforts to secure systemic competitiveness by connecting its own semiconductors to global AI infrastructure, moving beyond mere memory supply. The effectiveness of this technological integration and the results of actual commercialization must be verified in the future.

Third, we must strengthen industrial security systems to protect core technologies and talent. Executives at AI and semiconductor companies have access to sensitive information regarding technology development schedules, supply chains, and government policies. A counterintelligence and early-warning system is needed to proactively identify and respond to potential access by foreign intelligence agencies.

The government must secure an AI foundation for national security and essential public services, while private companies must develop technologies and business models capable of competing in the global market. It is also important to reduce redundant investment while maintaining a balance between international cooperation and technological self-reliance.

From a Semiconductor Supplier to an AI Powerhouse

The truth behind the allegations surrounding Jensen Huang and the question of responsibility must be determined based on the results of the investigation. This incident demonstrates that the global AI industry relies heavily on specific companies, their executives, and national policy decisions.

South Korea must be able to continue its research and development, corporate activities, and the operation of core systems by public institutions and the military even if the supply of next-generation AI chips from overseas is delayed or the service terms of global cloud companies change.

HBM competitiveness is a key bargaining chip that South Korea has secured within the global AI supply chain. To develop this into national competitiveness, the country must simultaneously cultivate its own AI chips, models, cloud services, and computing infrastructure.

In the competition for AI supremacy, South Korea’s goal must not be limited to securing a semiconductor supply. The country must be able to maintain core functions even if external supplies are disrupted and independently select the necessary technologies and partners. It must become a nation that develops and operates AI directly and translates those achievements into industrial and security competitiveness.