In the first weekend of June 2026, all eyes in South Korea's semiconductor industry were on NVIDIA CEO Jensen Huang. After presenting his vision for the future of the global AI industry at the Computex event in Taiwan, he quickly traveled to South Korea. The local media covered his every move in real time, including meetings with executives from LG Electronics, Naver, and SK Hynix, as well as a dinner featuring samgyeopsal and soju. However, the global semiconductor industry is a place where cold calculations often lie behind smiling faces. While friendships exist in international business, the direction of the industry is ultimately determined by interests, supply chains, and market logic. In this context, Huang's visit to South Korea was more of a strategic assessment of the global supply chain for the AI era than a friendly visit.
During his stay, Huang reportedly focused on discussions with major partners, including Samsung Electronics and SK Hynix, regarding next-generation HBM supply plans, the expansion of AI data centers, and forecasts for future AI server market demand. While the media's attention was drawn to the dinner, the truly significant discussions took place behind closed doors, where the future AI industry's supply chain worth tens of trillions of won was at stake.
Many people today understand the AI revolution as a GPU revolution, which is not entirely incorrect. However, within the AI industry, there are already discussions about something even more critical than GPUs: HBM, or high-bandwidth memory. No matter how advanced a GPU is, if it cannot supply data in a timely manner, the overall performance of the AI system will inevitably decline. If the AI industry is a vast highway, the GPU is the engine, and HBM is the essential system supplying fuel to that engine. In this market, South Korean companies have secured world-class competitiveness. SK Hynix currently leads the global HBM market, while Samsung Electronics is launching a major counterattack centered on next-generation HBM4.
However, South Korea must be wary of a trap: becoming complacent with its current position as the world's number one in HBM. History shows that the winners of technological revolutions are not necessarily the countries that produce specific components best, but those that dominate the entire ecosystem. The UK invented the steam engine, but the U.S. emerged as the ultimate victor of the Industrial Revolution. Japan dominated the DRAM market, but South Korea took the lead in the memory semiconductor era. Similarly, while South Korea is a powerhouse in HBM, it cannot yet be called an AI powerhouse. Currently, in the global AI ecosystem, design is handled by NVIDIA, cutting-edge production is managed by TSMC, and cloud and platform services are dominated by U.S. tech giants. South Korea stands at a crucial point in the supply chain but does not yet control or design the entire system.
Thus, South Korea's goal should not be to remain an HBM powerhouse but to leap to an AI infrastructure powerhouse. The AI industry is no longer just about semiconductors. It is evolving into a massive national infrastructure industry that combines data centers, power grids, cooling systems, communication networks, cloud services, and networks. In the future, global AI competitiveness is likely to be determined not only by semiconductor production capacity but also by how much power can be reliably supplied and how efficient data centers can be built. In fact, the U.S. and China have already entered a competition to build super-sized AI data centers, and some experts predict that the biggest bottleneck in the future AI industry may not be a shortage of semiconductors but a lack of power.
South Korea possesses significant potential amid these changes. It has a world-class semiconductor industry, as well as expertise in communication infrastructure, nuclear power technology, battery production, and power grid management. The challenge is that these industries have yet to be integrated into a cohesive national strategy. A comprehensive strategy that connects semiconductors, energy, communications, data centers, manufacturing, and AI is now necessary. Only then can South Korea transcend being a mere semiconductor supplier and become a leader in AI infrastructure.
An even more critical concept emerges here: Physical AI. Many still perceive AI as conversational services like ChatGPT, but Huang has recently emphasized areas such as robotics, autonomous driving, smart factories, and humanoid industries. This indicates that the era of AI moving beyond computer screens into the real world has begun. At this juncture, South Korea can position itself more advantageously than any other country. With a world-class manufacturing base in shipbuilding, automobiles, steel, batteries, machinery, and semiconductors, the combination of AI and manufacturing could elevate South Korea from a mere IT powerhouse to a central player in the future industrial revolution. The era of Physical AI, where factories make autonomous decisions, robots move independently, and logistics systems optimize in real time, could provide a stage for South Korea's manufacturing sector to showcase its greatest competitiveness.
Another reality we must confront is that South Korea's true competitor in the semiconductor industry is not China but Taiwan. While South Korean society often focuses on U.S.-China competition, the strongest competitor in the semiconductor industry is Taiwan. The semiconductor ecosystem built around TSMC is not just a production facility but a vast industrial platform where design, production, packaging, and supply chains are interconnected. This is why Huang effectively utilizes Taiwan as his second base. Today, the AI supply chain connecting the U.S. and Taiwan is arguably the central axis of the global AI industry.
Ultimately, the message Huang's visit sends to South Korea is straightforward: being number one in HBM is just the starting point, not the destination. South Korea must now grow beyond being a memory powerhouse to become an AI infrastructure powerhouse and evolve from a semiconductor supplier to a key player in the AI ecosystem. Furthermore, it needs to establish a national strategy that connects semiconductors, data centers, power grids, robotics, manufacturing, and Physical AI. Only then can South Korea become a co-designer of a new industrial civilization rather than a subcontractor in the AI empire.
While the memories of samgyeopsal and soju will remain for a day, the questions raised by this visit could determine the future of South Korea's economy for decades to come. Will South Korea be satisfied with its current success, or will it leap to become a central nation in the AI era? History always offers opportunities to only a few countries at decisive moments. Now is the time for South Korea to seize that opportunity.
During his stay, Huang reportedly focused on discussions with major partners, including Samsung Electronics and SK Hynix, regarding next-generation HBM supply plans, the expansion of AI data centers, and forecasts for future AI server market demand. While the media's attention was drawn to the dinner, the truly significant discussions took place behind closed doors, where the future AI industry's supply chain worth tens of trillions of won was at stake.
Many people today understand the AI revolution as a GPU revolution, which is not entirely incorrect. However, within the AI industry, there are already discussions about something even more critical than GPUs: HBM, or high-bandwidth memory. No matter how advanced a GPU is, if it cannot supply data in a timely manner, the overall performance of the AI system will inevitably decline. If the AI industry is a vast highway, the GPU is the engine, and HBM is the essential system supplying fuel to that engine. In this market, South Korean companies have secured world-class competitiveness. SK Hynix currently leads the global HBM market, while Samsung Electronics is launching a major counterattack centered on next-generation HBM4.
However, South Korea must be wary of a trap: becoming complacent with its current position as the world's number one in HBM. History shows that the winners of technological revolutions are not necessarily the countries that produce specific components best, but those that dominate the entire ecosystem. The UK invented the steam engine, but the U.S. emerged as the ultimate victor of the Industrial Revolution. Japan dominated the DRAM market, but South Korea took the lead in the memory semiconductor era. Similarly, while South Korea is a powerhouse in HBM, it cannot yet be called an AI powerhouse. Currently, in the global AI ecosystem, design is handled by NVIDIA, cutting-edge production is managed by TSMC, and cloud and platform services are dominated by U.S. tech giants. South Korea stands at a crucial point in the supply chain but does not yet control or design the entire system.
Thus, South Korea's goal should not be to remain an HBM powerhouse but to leap to an AI infrastructure powerhouse. The AI industry is no longer just about semiconductors. It is evolving into a massive national infrastructure industry that combines data centers, power grids, cooling systems, communication networks, cloud services, and networks. In the future, global AI competitiveness is likely to be determined not only by semiconductor production capacity but also by how much power can be reliably supplied and how efficient data centers can be built. In fact, the U.S. and China have already entered a competition to build super-sized AI data centers, and some experts predict that the biggest bottleneck in the future AI industry may not be a shortage of semiconductors but a lack of power.
South Korea possesses significant potential amid these changes. It has a world-class semiconductor industry, as well as expertise in communication infrastructure, nuclear power technology, battery production, and power grid management. The challenge is that these industries have yet to be integrated into a cohesive national strategy. A comprehensive strategy that connects semiconductors, energy, communications, data centers, manufacturing, and AI is now necessary. Only then can South Korea transcend being a mere semiconductor supplier and become a leader in AI infrastructure.
An even more critical concept emerges here: Physical AI. Many still perceive AI as conversational services like ChatGPT, but Huang has recently emphasized areas such as robotics, autonomous driving, smart factories, and humanoid industries. This indicates that the era of AI moving beyond computer screens into the real world has begun. At this juncture, South Korea can position itself more advantageously than any other country. With a world-class manufacturing base in shipbuilding, automobiles, steel, batteries, machinery, and semiconductors, the combination of AI and manufacturing could elevate South Korea from a mere IT powerhouse to a central player in the future industrial revolution. The era of Physical AI, where factories make autonomous decisions, robots move independently, and logistics systems optimize in real time, could provide a stage for South Korea's manufacturing sector to showcase its greatest competitiveness.
Another reality we must confront is that South Korea's true competitor in the semiconductor industry is not China but Taiwan. While South Korean society often focuses on U.S.-China competition, the strongest competitor in the semiconductor industry is Taiwan. The semiconductor ecosystem built around TSMC is not just a production facility but a vast industrial platform where design, production, packaging, and supply chains are interconnected. This is why Huang effectively utilizes Taiwan as his second base. Today, the AI supply chain connecting the U.S. and Taiwan is arguably the central axis of the global AI industry.
Ultimately, the message Huang's visit sends to South Korea is straightforward: being number one in HBM is just the starting point, not the destination. South Korea must now grow beyond being a memory powerhouse to become an AI infrastructure powerhouse and evolve from a semiconductor supplier to a key player in the AI ecosystem. Furthermore, it needs to establish a national strategy that connects semiconductors, data centers, power grids, robotics, manufacturing, and Physical AI. Only then can South Korea become a co-designer of a new industrial civilization rather than a subcontractor in the AI empire.
While the memories of samgyeopsal and soju will remain for a day, the questions raised by this visit could determine the future of South Korea's economy for decades to come. Will South Korea be satisfied with its current success, or will it leap to become a central nation in the AI era? History always offers opportunities to only a few countries at decisive moments. Now is the time for South Korea to seize that opportunity.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.