"Semiconductor Supercycle, Just Beginning"... How Long Will the Soaring K-Industry Boom Last? [Why&Next]

Semiconductor Market Outlook: Expert Analysis

Samsung HBM4 Sales Surpass $1 Billion

SK hynix Becomes No. 1 in KOSPI Market Cap

"Demand to Continue Beyond AI, Reaching Drones and Defense"

"Peak Is Coming... Possibly by the End o

Samsung Electronics’ next-generation high bandwidth memory (HBM), the HBM4, has surpassed $1 billion (approximately 1.538 trillion won) in sales just about 130 days after mass production and shipment began. Cumulative sales are expected to exceed $10 billion by the end of the year. As the memory semiconductor boom led by HBM accelerates, SK hynix has overtaken Samsung Electronics, a comprehensive electronics company, to become the largest company by market capitalization on the KOSPI. Experts predict that this boom could extend beyond the existing cycle and continue even after 2028, and note the possibility that AI demand may expand across a wide range of industries, including defense and automotive.

SK hynix Semiconductor Cluster Phase 1 Fab Construction Site, Cheoin District, Yongin City, Gyeonggi Province. Photo by Kang Jin-hyung

According to the Korea Exchange on June 23, as of 10:18 a.m. that day, SK hynix’s market capitalization stood at 2,034.7653 trillion won, surpassing Samsung Electronics (2,006.7351 trillion won). The previous day, at 12:51 p.m., SK hynix (2,084.6544 trillion won) had also overtaken Samsung Electronics (2,084.1983 trillion won), indicating a continuing trend of reversal. When including Samsung Electronics’ preferred shares (177.3240 trillion won), the combined market capitalization comes to 2,184.0591 trillion won, outpacing SK hynix, and both companies are boosting their corporate value on the back of the AI semiconductor boom. The fact that both companies have demonstrated strong competitiveness in the HBM market, which is essential for AI accelerators, is seen as the reason investor expectations are being reflected in their share prices.

Exploding Demand, Insufficient Supply... From Cycle to ‘Supercycle’

The semiconductor industry points to a “supply-demand imbalance”—caused by expanded global AI infrastructure investment and supply shortages—as the background for the recent supercycle. As big tech companies competing in AI development ramp up investment in data centers and accelerate in-house AI chip development, demand for high-performance memory semiconductors such as HBM is increasing rapidly. However, supply has not kept up, resulting in ongoing price increases. An industry insider explained, “Right now, supply is insufficient and demand is overflowing, leading to significant price hikes and even talk of a supercycle.”

Global semiconductor market forecasts support this trend. According to the World Semiconductor Trade Statistics (WSTS), the global semiconductor market is expected to reach $1.511 trillion (2,329 trillion won) in 2026 and $1.914 trillion (2,950 trillion won) in 2027. The year-on-year growth rate is also projected to climb to as high as 89.9% in 2026.

In particular, expanded investment in AI data centers and rising HBM demand are simultaneously driving up both memory semiconductor prices and shipment volumes, resulting in Korea’s two leading chipmakers achieving results at levels unlike previous cycles. The performance outlook for Samsung Electronics and SK hynix is rising so steeply that it is becoming impossible to predict. In the market, the operating profit forecasts for both companies for the second quarter of this year have nearly doubled in just the past three months. In the first quarter, securities firms repeatedly raised their earnings forecasts, but actual results still exceeded market consensus. According to a recent report by Goldman Sachs, Samsung Electronics’ 2028 operating profit is projected at 610 trillion won, while SK hynix is forecast at 454 trillion won. Combined, this exceeds 1,000 trillion won.

Experts: “Boom to Continue for Now”... But Cycle Peak Will Arrive

Semiconductor experts believe that the current boom is not part of the traditional memory cycle, but is instead a new demand cycle created by the spread of the AI industry. As applications such as AI data centers, custom AI chips, autonomous vehicles, and humanoids expand rapidly, demand for high-performance memory, including HBM, is expected to continue for the foreseeable future. However, opinions are divided on when the supercycle will peak. There are both cautious views that see a potential demand adjustment in the second half of next year, and optimistic projections that the boom could last beyond 2028.

Yoo Hoejun, professor of Electrical and Electronic Engineering at KAIST, said, “I believe the outlook for Korea’s semiconductor industry is closely tied to NVIDIA. As long as AI keeps advancing, NVIDIA will continue to grow, and as long as NVIDIA grows, so too will Samsung Electronics and SK hynix.” He added, “Since Samsung Electronics and SK hynix are both expanding their capacity, the semiconductor boom is likely to persist.”

An Ki-hyun, Executive Director of the Korea Semiconductor Industry Association, also said, “The supercycle will continue for the time being. I believe growth will persist at least through 2028.” He explained, “Unlike the replacement cycles for PCs and smartphones, the AI-driven market is just beginning. People are continually using AI, and demand is continuously increasing.”

Furthermore, the view that AI semiconductor demand could extend beyond existing data centers to much broader fields adds weight to the argument for a prolonged boom. There is analysis that AI semiconductor demand could expand beyond applications such as autonomous vehicles and drones to humanoid robots, defense, and other industries. Lee Jonghwan, professor of System Semiconductor Engineering at Sangmyung University, stated, “The areas of AI application are expanding rapidly, and accordingly, demand for AI semiconductors is surging. That demand will inevitably continue to grow, and HBM will be essential in that expansion.”

However, there is considerable debate over when the current boom will reach its peak. The industry cautiously analyzes that this cycle cannot last forever, and the peak could come as early as the second half of next year. Explosive semiconductor demand could be adjusted, supply would be realigned accordingly, and price stabilization could bring performance back to a normal trajectory. Professor Lee commented, “By the second half of next year, there is a high possibility that the previously rising demand will be somewhat adjusted. However, if new applications with additional demand emerge, demand could still increase; if not, it may simply be maintained.”

Investment trends among AI companies are also cited as a key variable determining the direction of the cycle. Kim Yangpaeng, Senior Research Fellow at the Korea Institute for Industrial Economics and Trade, said, “A variable would be if AI companies such as OpenAI and Anthropic reduce their investments. If OpenAI or Anthropic fail to go public and are unable to raise further investment, the current trend could subside.” He added, “It is difficult to pinpoint the exact timing of the supercycle peak, but it could come as soon as next year.”

There are also suggestions that structural changes in the AI market after 2030 could become another turning point. If the market shifts from AI training to AI inference, demand for general-purpose memory could rise above HBM, and combined with the effects of memory makers expanding production, it may become difficult to maintain current levels of profitability. Kim Yongseok, Distinguished Professor at Gachon University’s College of Semiconductors, also commented, “By around 2030, the AI inference market is expected to surpass the AI training market, and it is uncertain whether the current supercycle can continue beyond that. In inference, general-purpose memory is used more than HBM, and as memory makers continue to expand production, HBM supply will increase, making it difficult to sustain the current level of extraordinary profits.”

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