[Chip-In] "The Semiconductor Supercycle Could Become a Poison...Urgent Need for National Strategy and System Semiconductor Development"

Interview with System Semiconductor Expert Kim Yongseok
"Korea's Semiconductor Industry Lacks a National Strategy"
"Top Player Samsung Must Move Beyond Memory-Centric Focus"
Developing 'Customized Proprietary Chips' and Fostering Design Talent

"The current semiconductor supercycle could potentially become a poison for Samsung Electronics."

Kim Yongseok, Distinguished Professor at Gachon University Semiconductor College and Director of the Semiconductor Education Center, who spent 31 years at Samsung Electronics and has spent 12 years teaching after retirement, shared this candid assessment in an interview with The Asia Business Daily on December 3. He noted, "Samsung Electronics is on the verge of success with HBM4 (high-bandwidth memory) and is expecting unprecedented results due to the rising prices of legacy memory." However, he cautioned that the company should not be satisfied with results driven merely by favorable market conditions, but should strive to build fundamental competitiveness.

Professor Kim believes that domestic semiconductor companies need strategies to ensure continued growth even after the supercycle ends. He said, "While memory is booming, especially centered on HBM, system semiconductors remain weak. Now, in the era of artificial intelligence (AI), is the turning point to reorganize fabless (design) and foundry capabilities."

"Now Is the Time to Develop Fabless and Foundry"

Kim Yongseok, Distinguished Professor at Gachon University, is posing after an interview with The Asia Business Daily. Photo by Yoon Dongju

He explained that for Korea to maintain its status as a semiconductor powerhouse, a national-level grand strategy is needed, beyond the efforts of individual companies. Korea is already lagging somewhat behind China, which is rapidly catching up. He said, "The semiconductor industry has increasingly become a core strategic asset that determines a nation's economy and security. It is difficult to respond to today's complex crises through the efforts of individual companies alone."

China is ahead of Korea in both quantity and quality in the fabless sector. Its foundry companies, such as SMIC, Hua Hong Semiconductor, and Nexchip, provide strong support. Market research firm TrendForce forecasts that by 2027, China will surpass Taiwan to become the world's number one in foundry production capacity. Professor Kim criticized, "The essence of China's rise lies in its national innovation strategy, while Korea lacks a national-level strategy."

Professor Kim explained that Samsung Electronics, which has long held the top spot domestically, must now prepare for a new phase of growth. He believes that the key to future growth lies in how Samsung Electronics can coordinate and maximize synergy across its broad portfolio, which spans from memory strengths to system semiconductor design and foundry. The structure in which multiple businesses coexist within one company has great potential, and with strategic selection and focus, this could lead to even greater expansion.

He also emphasized that decades of memory-centered leadership have built global competitiveness in related fields. Professor Kim assessed that the recently announced 1,000 trillion won investment plans by Samsung Electronics and SK Hynix can be interpreted as a process of seeking new growth axes based on accumulated competitiveness. He also noted the potential for the technology and capital accumulated in memory to expand into other areas such as system semiconductors and foundry in the future.

As solutions, he suggested organizational restructuring and strengthening software (SW) capabilities. Professor Kim said, "Foundry is fundamentally a service business that manufactures chips based on designs, so preparing the necessary design libraries is essential. There must be many experts in AI models and compilers." He recommended that now is the time to consider spinning off foundry or semiconductor design divisions. He judged that preparing for a spin-off now, when there is ample financial capacity due to the supercycle, and executing it in one to two years, would be optimal timing.

"Even If It Is Late, We Must Compete with Our Own Chips"

Kim Yongseok, Distinguished Professor at Gachon University Semiconductor College, participated in the development of Samsung Electronics' Exynos modem chip lineup, which is the core of Exynos, during his tenure at Samsung Electronics Communication Research Institute. Photo by Park Juni.

Professor Kim emphasized that the government should take the lead in developing proprietary chips. He said, "While it is necessary to use external chips to quickly develop and commercialize products, there will inevitably come a time when we become dependent on chip suppliers and are at their mercy. Ultimately, we need to develop our own chips."

In particular, on-device AI semiconductors must operate within individual devices such as smartphones, automobiles, home appliances, and robots. With general-purpose chips, it is difficult to achieve results in terms of performance and price, and chip costs also rise. Therefore, he explained, it is essential to develop 'customized chips' tailored to the characteristics of each product.

Talent Development Requires Qualitative Change..."Securing Software Talent"

Professor Yongseok Kim, Chair Professor at Gachon University, is being interviewed by The Asia Business Daily. Photo by Dongju Yoon

Finally, Professor Kim believes that a strategy is needed for talent development as well. He said, "The biggest factor in China's success is the hard work of outstanding talent. We need a structure where engineers are respected in society." He added that it is not just about increasing the number of professionals, but cultivating high-quality talent. Professor Kim stated, "In the semiconductor field, we need to train master's and doctoral-level professionals, especially master's-level talent with practical training," and "through bachelor's and master's programs, we should nurture master's-level talent in five years, establish specialized graduate schools for design, devices and processes, packaging, and so on, and develop specialized professionals."

He especially emphasized the importance of talent capable of understanding AI models (algorithms), as design capability is essential for AI semiconductors. He said, "We need separate personnel who can design AI models into chips and those who can make chips work properly through software (compilers, runtime software, system software). Since the software portion is even more important than chip design in AI semiconductors, securing software talent is the key to success."

This content was produced with the assistance of AI translation services.