Wall Street Soars, KOSPI Stays Trapped... U.S. Has Big Tech and Chip Design, Korea Only Memory in English

Lack of Sectors at the Upper End of the AI Value Chain
South Korean Market Has No Sectors to Rotate Into
Risk of Being Left Behind in the First Half of Next Year

Amid the recent high-interest-rate environment, U.S. stocks in New York have continued to climb, while the KOSPI has remained range-bound. Some analysts say this reflects the widening gap between the U.S. and South Korea in the structure of their AI value chains, as the AI theme shifts from infrastructure investment to monetization.

On the 12th, the KOSPI and KOSDAQ indexes are displayed on an electronic board in the dealing room at Hana Bank's headquarters in Jung-gu, Seoul. Yonhap News Agency.

According to iM Securities on October 12, the U.S. has leading publicly listed companies across five upper-tier segments of the AI value chain: ▲applications (Palantir and ServiceNow) ▲cybersecurity (CrowdStrike and Palo Alto Networks) ▲cloud platforms (M7) ▲semiconductor design (NVIDIA, Broadcom, and Marvell) and ▲electronic design automation (EDA) for semiconductors (Synopsys and Cadence). In contrast, South Korea's industry is concentrated in manufacturing, including contract manufacturing (foundries), memory, and equipment.

U.S. companies in these upper-tier segments have relatively little inventory and capital-equipment exposure, along with high customer-retention rates and subscription-based revenue models, so they are not directly affected even when memory prices fall. Marvell, for example, is a fabless semiconductor design company that outsources manufacturing to foundries, and data centers account for 79% of its revenue. In the EDA sector, three companies—Synopsys, Cadence, and Siemens—control 70-80% of the global market. Subscription-based recurring revenue accounts for 80-85% of these companies' sales, and their customer-retention rates are close to 100%.

By contrast, South Korea's industry is centered on memory integrated device manufacturers (IDMs), which bear the costs of capital investment and inventory themselves. As a result, when supply exceeds demand, falling average selling prices (ASPs) and inventory valuation losses directly erode profits. Even when benefiting from the same AI boom, upper-tier companies can expand their margins, while lower-tier companies face the structural limitation of exposure to price cycles. The KOSPI's heavy concentration in semiconductors is also a concern. Samsung Electronics and SK hynix together account for more than half of the KOSPI's market capitalization, while semiconductors make up only about 18% of the U.S. S&P 500.

With upper-tier sectors missing from South Korea's market, there are often no corresponding sectors or companies in the domestic market when sector rotation takes place in the U.S. Even when U.S. capital shifts from beneficiaries of infrastructure investment to AI applications and cloud companies, U.S. indexes can continue to rise simply by rotating between sectors within the index. South Korea, by contrast, lacks upper-tier sectors that capital can move into, causing funds to flow out of the stock market. In other words, U.S. sector rotation takes place within the index, while in South Korea, capital leaves the index.

Foreign investor flows also suggest that this sector rotation may already have begun. The continued absence of a return in foreign investor flows can be interpreted as a process of reallocating capital from AI infrastructure to other sectors. In this process, South Korea is inevitably used as a source of funds for sales, and foreign investors' decisions to reduce their exposure to AI infrastructure lead directly to lower allocations to the South Korean stock market.

The slowdown in investment growth is another concern. Infrastructure companies' share prices respond not to the absolute amount of capital investment but to its growth rate. Even if big tech companies (large information technology companies) continue to increase their investment, share prices of companies lower in the value chain are the first to peak once the growth rate starts to decline. Quarterly capital-investment growth among hyperscalers is expected to peak in the third quarter of this year.

Ultimately, South Korea remains an intermediate-goods supplier in the AI cycle, so its relative strength is bound to weaken as the AI market moves into the monetization and application stages. Some sectors, including power equipment and substrates, are benefiting from investment in physical AI infrastructure. However, their weight in the overall market is small compared with that of Samsung Electronics and SK hynix, making it difficult for them to offset concerns that the broader stock market will be left behind.

Brokerages say that the earnings resilience of South Korean semiconductor companies will be key to a stock market rebound as AI moves into the monetization phase and high interest rates become the new normal. "As AI moves into the application stage in the first half of next year, the risk that the South Korean stock market will be left behind may be greatest while application-layer sectors such as software and security benefit, because these sectors are underrepresented here," said Kim Junyoung, a researcher at iM Securities. "However, if surging token consumption driven by wider AI adoption leads to renewed shortages of inference capacity, memory and AI infrastructure could benefit again in the second half of the year, creating a virtuous cycle." He added, "A strategy of reducing concentration in memory, taking a defensive position centered on high-bandwidth memory (HBM) and power infrastructure, and preparing for a rebound in the second half would be effective in anticipation of weak relative performance in the first half."

"Because the high-interest-rate environment is unlikely to ease soon, investors should look for opportunities within AI infrastructure, where growth is outpacing costs," said Hwang Suuk, a researcher at Meritz Securities. "The fundamental reason South Korean semiconductor stocks have underperformed despite the global big-tech rally is the lack of earnings momentum. For the index to rebound, it is crucial for market forecasts to take shape that earnings can grow for three consecutive years through 2028."

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