Jensen Huang, the Nvidia chief executive officer (CEO), left Korea on the 9th after a five-day visit. From a samgyeopsal gathering to throwing the first pitch at a baseball game and visiting university labs, the domestic business community was abuzz and the stock market jumped for days. But behind the heightened expectations lingers an uncomfortable question: It is worth examining whether what CEO Huang, a titan of global big tech and a figure at the apex of Silicon Valley, left in Korea is merely a gift.
Some see the closer ties with Nvidia as a chance to elevate Korea's technological strengths and international standing in semiconductors, IT infrastructure, and Robotics, while warning of a simultaneous risk: forming a subordinate relationship with big tech. They say Korea should heed the precedent of Japan's former electronics and manufacturing giants, including NEC, Toshiba, and Hitachi, whose partnerships with U.S. big tech slid into dependency and eroded their competitiveness.
◇ "Contracts with Nvidia are not the ultimate goal but a springboard for growth"
According to the industry on the 10th, questions are being raised that the partnerships released by Nvidia and numerous domestic corporations are loosely binding and unclear. Most of the discussions during the visit amounted to memorandums of understanding (MOUs) or verbal agreements. MOUs carry no legal force, performance obligations, or penalties for breach. The industry views this trip as largely about assigning Korea a role within Nvidia's global artificial intelligence (AI) infrastructure expansion strategy, and says the bigger task ahead is determining what status Korea can secure going forward.
In semiconductor cooperation, the visit did not bring structural change. Samsung Electronics and SK hynix currently supply about 70% of the high-bandwidth memory (HBM) needed for Nvidia's AI chips. That figure cuts both ways. Korea is an essential partner to Nvidia, but without Nvidia there are few customers to absorb that volume. The current division of labor is clear and aligns with the interests of Samsung Electronics and SK hynix, but as the prime–subcontractor structure of design (Nvidia) and manufacturing (Samsung Electronics, SK hynix) hardens, medium- to long-term risks emerge.
For example, the fact that partnerships between Japan's semiconductor industry and Silicon Valley heavyweights in the 1980s ultimately culminated in the collapse of Japan's DRAM industry is telling. Japan once controlled 80% of the global DRAM market, led by the "big five" including NEC, Toshiba, and Hitachi, but it settled into a division of labor—manufacturing in Japan, design in the United States—while focusing only on expanding production capacity. Later, a combination of factors, including the price floor mandated by the U.S.-Japan semiconductor agreement, a strong yen, and Samsung Electronics' aggressive investment and supply-chain entry, cost Japanese corporations their competitiveness.
Concerns about SK hynix, which has hinted at a long-term supply contract with Nvidia, arise in a similar vein. As the center of gravity in the AI Semiconductor market shifts from training to inference, memory demand is changing by the moment. A senior official in the domestic AI fabless companies sector said, "As Broadcom, Google, and OpenAI strengthen partnerships with memory companies, being tied to a single customer can hinder flexible responses to change," adding, "We should not make a contract with Nvidia the ultimate goal; we should approach it from the perspective of using Nvidia as a tool to raise our technological level."
◇ Prepare a "two-track" strategy while bracing for China's rise
There are also question marks over Korea's potential and growth prospects in Robotics. That is because China could become a more significant partner than Korea in Nvidia's physical AI strategy. China is the world's largest industrial robot market and consumer. Chinese humanoid corporations such as Unitree are already developing on Nvidia's Isaac platform. China's manufacturing sites are also the world's largest testbeds where Nvidia can validate physical AI.
Analysts say the reason CEO Huang emphasized Korea in Robotics has more to do with geopolitics than technical prowess. Under current U.S. export controls, Nvidia faces restrictions on supplying high-performance graphics processing units (GPUs) to China. Korea is one of the few Asian allies able to introduce Nvidia's latest platforms without restriction. If U.S.-China relations ease or the export-control framework shifts, Korea's geopolitical premium in Robotics could quickly fade.
An IT industry official said, "Big tech like Nvidia and Apple traditionally do not settle for a single vendor," adding, "Once the technology of a partner like Korea reaches a certain level of standardization, they will look to China or other rivals based on standardized manufacturing recipes and move to lower supply prices."