"China, with over 3,000 fabless companies... Must counter with a smart minority" [The Reviving Semiconductor Division of Labor (Part 2)]

Experts have suggested that to boost South Korea's competitiveness in system semiconductors, the country should focus on selecting and intensively nurturing promising companies rather than obsessing over increasing the number of fabless (semiconductor design) firms. Given the difficulty of engaging in quantitative competition with nations like China, which possess thousands of fabless companies, the argument is that a strategy of "selection and concentration" is necessary to foster competitive fabless firms through mergers and acquisitions (M&A) and partnerships with large conglomerates. Cultivating practical talent and expanding semiconductor design and verification infrastructure were also cited as parallel tasks.
In an interview with the Financial News on the 7th, Kim Yong-seok, a distinguished professor at Gachon University's College of Semiconductors, diagnosed that strengthening competitiveness is urgent, stating, "The global market share of South Korean system semiconductors is merely at the 1–3% level." He emphasized that a strategy of 'selection and concentration' is particularly necessary in the domestic fabless market.
With over 3,000 fabless companies in China compared to only about 120 in South Korea, there are limitations to competing by simply increasing the number of firms. "Increasing quantity may not be a winning strategy. Ultimately, what matters is how we effectively nurture capable fabless companies," said Professor Kim. "Even if the number of fabless firms shrinks to 40 or 50, we must find ways to develop them into competent companies through M&A and other means."
It was also emphasized that in order to foster the system semiconductor industry, demand companies such as large corporations must be connected with fabless companies from the early stages of semiconductor development. This is because even if domestic system semiconductors are developed, it is difficult for fabless companies to grow if they are not adopted in actual products.
The government is also promoting the 'K-on-Device Artificial Intelligence (AI) Semiconductor Technology Development' project with a total budget of 800 billion won by 2030. In this project, large demand companies and fabless companies are also scheduled to work together to develop commercially viable prototypes.
"Professor Kim stated, 'The reason system semiconductor localization is not progressing well is ultimately because large corporations do not use fabless products,' adding, 'It is important to create an environment where large corporations and fabless companies can collaborate and develop together.'"
Cultivating talent suited to the field was also cited as a challenge to be addressed in strengthening the competitiveness of the system semiconductor industry. In particular, it was pointed out that personnel for system software, which is required to run semiconductors in actual products, must be trained alongside semiconductor design experts.
"Professor Kim stated, 'Software accounts for 60 to 70 percent of AI semiconductors, which is larger than the chip (hardware) itself. Therefore, it is crucial to cultivate system software personnel capable of effectively operating the chips.' He added, 'Education at universities must shift toward fostering practical talent that companies can immediately utilize.'"
Jonghwan Lee, Professor of System Semiconductor Engineering at Sangmyung University, also advised that university education and practical training environments should be transformed to be more field-oriented to meet industry demands. "To cultivate students who wish to work for fabless or foundry companies, it is necessary to secure a stable supply of semiconductor design tools starting from the educational field," said Professor Lee. "There should also be an increase in the operation of departments similar to contract departments, which lead to employment at those companies through industry-academia collaboration."
In addition to education, plans were also presented to expand infrastructure to support fabless development. The proposal is to establish a public foundry so that semiconductors designed by fabless companies, universities, and research institutes can be manufactured and verified at a low cost.
"Since it is costly to manufacture and verify actual semiconductors, we must create a virtuous cycle structure where manufacturing and verification proceed smoothly through the establishment of a public foundry," said Professor Lee. "If universities, research institutes, and fabless companies fully utilize the public foundry after its establishment, it will be of great help in revitalizing the system semiconductor ecosystem."
[email protected] Kang Gyeong-rae, Im Su-bin Reporter