In early summer 2026, the industrial landscape of Asia experienced significant upheaval as Jensen Huang, founder and CEO of NVIDIA, visited Taiwan and South Korea. His trip was not merely a business engagement or customer relations effort; it served as a field assessment of the newly emerging global industrial order and power dynamics following the AI revolution. It also represented a strategic move to gauge the direction of supply chain restructuring and industrial alliances for the next decade.
In Taiwan, Huang was treated like a hero. Crowds gathered at airports and event venues, and the media closely followed his every move. This phenomenon goes beyond the typical star CEO narrative; Taiwan is the physical heart of NVIDIA's AI empire. NVIDIA-designed GPUs are produced through TSMC, and numerous Taiwanese partners supply servers, circuit boards, packaging, cooling systems, and networking equipment. A significant portion of the vast industrial ecosystem that constitutes AI data centers is concentrated in Taiwan.
Currently, a top-tier AI server rack contains hundreds of thousands of components, while large-scale data centers utilize millions of parts. Taiwan is at the center of this complex value chain. Although NVIDIA's chips are changing the world, the manufacturing capabilities that bring these products to life are largely based in Taiwan. Thus, Taiwan is NVIDIA's most crucial strategic partner, and Huang's visit was a means of inspecting the core production base of his industrial empire.
However, paradoxically, Taiwan is both NVIDIA's greatest strength and its most significant risk. The concentration of cutting-edge semiconductor production capabilities in one region is advantageous in terms of efficiency but poses substantial geopolitical risks. As U.S.-China strategic competition intensifies and military tensions in the Taiwan Strait rise, global supply chains are becoming increasingly unstable. This is why the U.S. government and global corporations are discussing supply chain diversification.
At this juncture, South Korea's strategic value becomes apparent. Huang's visit to South Korea immediately after Taiwan was not solely to encourage HBM suppliers. He is already envisioning a world beyond generative AI. His focus extends beyond GPU sales to how AI will transform industries and society. At the center of this future is physical AI.
South Korea possesses advantages that differ significantly from Taiwan. While Taiwan boasts the world's leading semiconductor manufacturing ecosystem, South Korea is a comprehensive industrial nation with capabilities in semiconductors, automobiles, batteries, shipbuilding, steel, telecommunications, platforms, and content. Samsung Electronics and SK Hynix dominate the memory semiconductor market, while Hyundai Motor Group has established global competitiveness in electric vehicles, autonomous driving, and robotics. LG is expanding its battery and automotive electronics businesses, and Naver operates its own AI models and cloud platforms.
This industrial structure holds significant implications for the era of physical AI. While generative AI has transformed the realms of text, images, and videos, physical AI will revolutionize factories, logistics warehouses, hospitals, farms, construction sites, and entire cities. As AI integrates with robotics, manufacturing, and mobility, the entire real world is transitioning into a massive AI platform. What Huang sought to observe in South Korea was not merely the production capacity of memory semiconductors but the national capability to implement this industrial transformation.
Thus, the most symbolic moment of Huang's visit was the sharing of samgyeopsal and soju. While the media portrayed this as a symbol of human connection and friendly management style, from an industrial perspective, it carried another meaning. It represented a meeting between a company holding the reins of the global AI industry and a nation with the world's best manufacturing capabilities, discussing the future industrial order.
However, in the realm of international business, hospitality and friendship are not everlasting. What remains are interests and structures. NVIDIA needs South Korea, but South Korea also needs NVIDIA. The critical question is who is more desperate. So far, South Korean companies have gained growth opportunities through NVIDIA. However, as the AI era deepens, NVIDIA will also require South Korea's manufacturing capabilities, industrial data, and physical AI testing environments. This is where South Korea can leverage its negotiating power.
While the samgyeopsal and soju were undoubtedly warm, the bill on the table was cold. It was not merely a meal cost but a calculation determining how power and value in the AI industry will be distributed over the next decade.
◆ The Essence of the AI Semiconductor War: The One Who Dominates the Ecosystem Wins, Not Just the Chips
Many consider NVIDIA the winner of the AI era. However, more accurately, NVIDIA is not merely an AI semiconductor company but an AI ecosystem company. Today, NVIDIA's competitiveness lies not in the GPU itself but in the vast platform built around it.
Looking back at the history of the semiconductor industry, the true winners have always been the companies that controlled the standards. IBM opened the computer era, but Microsoft dominated the operating system world. Nokia ruled mobile phones, yet Apple and Google seized control of the smartphone ecosystem. Hardware may change, but platforms endure. Platforms attract developers, retain users, accumulate data, and create new markets.
NVIDIA's true weapon is not the GPU but CUDA. Millions of developers worldwide are creating AI models within the CUDA environment, and numerous universities, research institutions, and companies are conducting education and research centered around CUDA. As the number of developers increases, so does the software, leading to a network effect where more developers attract more software, and vice versa. Once this structure is established, it becomes difficult for competitors to disrupt the ecosystem, even with superior chips.
Ultimately, the AI semiconductor market is not merely a chip competition; it is a platform competition, an operating system competition, and a data competition. The true power in the AI era lies not in making chips but in setting the rules by which AI operates.
Currently, the global AI industry operates under a structure where the U.S. dominates platforms and software, Taiwan handles production, and South Korea supplies memory. American tech giants control AI models and cloud platforms, with NVIDIA providing the core operating system of that ecosystem. Taiwan is responsible for cutting-edge production capabilities, while South Korea is tasked with memory supply, particularly HBM.
The critical question is where the highest added value lies. The answer is clear: it resides with those who control the platform. Platforms create markets, set prices, and accumulate data. In contrast, manufacturing can always face new competitors. While semiconductor manufacturing has a very high entry barrier, it is relatively more replaceable than platforms.
South Korea's strengths are evident. Samsung Electronics and SK Hynix are the world's leading memory companies. HBM is emerging as a key component in the AI era, and the expansion of AI data centers presents significant opportunities for South Korean semiconductor firms. However, to be frank, HBM is not yet a platform. It is an essential component but does not occupy a position that sets market rules.
A more significant issue lies in the era of physical AI. In the future, AI will expand into factories, automobiles, robotics, and urban infrastructure. In this process, industrial data will become the core competitive advantage. The know-how and operational data from manufacturing sites will be the most critical assets for future AI. If this data flows into foreign platforms, South Korea risks becoming a data colony despite being a manufacturing powerhouse.
Thus, future competition will extend beyond semiconductor competition to data sovereignty competition. In the AI era, data is the new oil, and in the physical AI era, manufacturing data is the gold mine. Who controls this data will determine the future of nations.
◆ South Korea's Choice: Evolving from NVIDIA's Partner to a Leader in Physical AI
South Korea stands at a historically significant crossroads. We are no longer a developing country. We are an industrial nation with top-tier competitiveness in semiconductors, automobiles, batteries, shipbuilding, steel, and digital industries. However, we still have a long way to go regarding platform sovereignty.
South Korea's greatest strength lies in its industrial diversity. The U.S. has strong platforms but a weak manufacturing base, Taiwan specializes in semiconductor manufacturing, and Japan is a precision manufacturing powerhouse but slow in digital platform transition. In contrast, South Korea possesses semiconductors, automobiles, batteries, robotics, platforms, and content simultaneously. This industrial portfolio holds tremendous potential in the era of physical AI.
However, potential only becomes reality with a strategic approach. What is needed now is a national-level AI industrial strategy. First, we must not become complacent with the success of HBM. We need to maintain our memory lead while expanding into AI system semiconductors, advanced packaging, and data center infrastructure. Second, we must build a South Korean physical AI platform. An integrated ecosystem connecting semiconductors, automobiles, robotics, and cloud services must be created. Third, we need to establish a sovereign AI system that protects manufacturing data. Data sovereignty will become a strategic asset on par with national security.
In particular, Hyundai Motor Group, Samsung Electronics, SK Group, LG Group, and Naver must evolve from mere manufacturing companies to platform companies. The past success formula of the fast follower strategy can no longer guarantee the future. Now, we must create our own standards and design ecosystems.
The era of physical AI presents both a crisis and an opportunity for South Korea. Collaborating with NVIDIA is necessary, but such collaboration should not lead to dependency. We must leverage NVIDIA's technology while simultaneously building our own platforms and data sovereignty. This is true strategic independence.
What is truth? It is recognizing that power in the AI era lies not in chips but in platforms and data. What is justice? It is ensuring that the fruits of industrial data and technology created by South Korea rightfully return to South Korea. What is freedom? It is having the ability to choose our future without being dependent on others' platforms.
The Jensen Huang phenomenon is not merely a CEO craze; it is a mirror reflecting South Korea's current position in the AI era. In that mirror, we see two images: one of pride as a world-class manufacturing powerhouse and another of anxiety over not yet securing platform sovereignty.
Now, South Korea must make a choice. We must choose the path of leveraging NVIDIA's success without becoming dependent on it, becoming a key partner in the global supply chain while building our own ecosystem, and advancing from an AI semiconductor powerhouse to a leader in physical AI.
Jensen Huang's visit to South Korea has concluded, but the questions he posed to the nation are just beginning. Will we view the world from the shoulders of giants, or will we become giants ourselves? The next 20 years for South Korea will be determined by the answer to that question.
※ This article was generated using generative AI and has been reviewed by an editor.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.