Nearly half of South Korea's top 30 conglomerates have established direct or indirect business relationships with NVIDIA. As NVIDIA expands its business territory beyond graphics processing units (GPUs) to include artificial intelligence (AI) infrastructure and physical AI, the connections with domestic companies are growing. However, there are rising concerns about increasing dependency.
According to a report by Aju Economy on August 25, an investigation into the cooperation between the top 30 conglomerates, as defined by the Fair Trade Commission, revealed that seven of the top 10 groups, including Samsung, SK, Hyundai Motor, LG, Lotte, POSCO, and HD Hyundai, are involved. Expanding the scope to the top 20 groups adds KT, LS, CJ, Kakao, and Doosan, bringing the total to 12. Including Naver and Coupang, the number of groups connected to NVIDIA rises to 14.
Samsung Electronics collaborates with NVIDIA in the AI semiconductor sector, particularly in high-bandwidth memory (HBM). SK Group is working on building AI factories utilizing NVIDIA GPUs. Hyundai Motor Group is expanding its cooperation in physical AI areas such as autonomous driving and robotics, while LG is enhancing its collaboration in robotics and AI infrastructure. Lotte is pursuing partnerships in AI data centers, and POSCO Group is applying NVIDIA's AI technology in manufacturing.
Additionally, KT is involved with NVIDIA in AI infrastructure and the development of large-scale AI models, while LS engages with NVIDIA in the power infrastructure sector. CJ is expanding its business areas in logistics and manufacturing, Kakao is growing its AI cloud business, Doosan is exploring opportunities in energy, robotics, and electronic materials, Naver is focusing on AI factories and cloud services, and Coupang is looking into AI factories and AI inference software.
The connections between NVIDIA and domestic companies are expanding beyond the semiconductor supply chain to include data centers, power infrastructure, automotive, shipbuilding, construction machinery, robotics, and logistics. Analysts suggest that the areas where NVIDIA's AI platform aligns with domestic business models are broader than expected.
It is essential to monitor the impact of changes in NVIDIA's investment cycle on the overall domestic industry. A slowdown in investments in AI data centers and AI factories could affect not only GPU demand but also investments in related industries such as power, cooling, networking, and equipment.
NVIDIA's structure of 'circular finance,' where it directly invests in AI companies or supports funding for AI infrastructure through financial institutions, is also seen as a variable. If the funds supplied to AI companies increasingly lead to purchases of NVIDIA GPUs, it may become challenging to distinguish between actual AI demand and investment scale.
While NVIDIA's second-quarter results are expected to exceed consensus estimates, clear signs of a slowdown in AI demand have yet to emerge. However, experts argue that alongside the benefits of the AI supercycle, the risks associated with increased dependency on specific corporate platforms should be assessed.
Lee Jong-hwan, a professor at Sangmyung University’s Department of System Semiconductor, stated, "NVIDIA is strategically managing customer relationships by providing various benefits to some companies that are trying to reduce their dependence on its GPUs. South Korean companies should strengthen their monitoring, but it does not appear to be an immediate cause for concern."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.