The global semiconductor industry is undergoing a significant power shift. In the past, large buyers dictated prices while suppliers waited for orders. However, as artificial intelligence (AI) emerges as a central force and production capacity concentrates on high bandwidth memory (HBM), market power is gradually shifting toward suppliers. This change is evident in the analysis that even Apple, one of the world's largest information technology companies, is finding it increasingly difficult to exert the same negotiating power in memory procurement as it once did.
Recent reports that Chinese memory semiconductor company Changxin Memory Technologies (CXMT) rejected Apple's supply request symbolize this transformation. Jeong Byeong-seo, director of the China Economic and Financial Research Institute, explained in an analysis released on October 27 that although Apple proposed price discounts and long-term contracts, CXMT declined due to increased orders from domestic customers and limited supply capacity.
However, this matter requires careful handling. The negotiation terms between Apple and CXMT, any equity investment proposals, and the actual reasons for the rejection need to be verified through official announcements and independent cross-checks. It would be premature to conclude that Apple has officially been rejected or that the global memory market's dominance has shifted based solely on some reports.
Nonetheless, the economic and industrial questions raised by this controversy are significant. If the era where Chinese semiconductor companies must secure orders from the world's largest customers is coming to an end, South Korea's semiconductor strategy must fundamentally change.
First, market power in semiconductors is shifting from buyers to suppliers.
The semiconductor industry is a typical cyclical industry. When demand increases, investment in production facilities expands, and when supply exceeds demand, prices plummet. Price declines lead to reduced investment, and when supply shortages occur, prices soar. This cycle, known as the silicon cycle, has dominated the memory industry for decades.
However, the AI revolution has introduced new variables into this long-standing cycle. Previously, the demand for memory was centered around personal computers and smartphones, but now AI data centers and high-performance servers are directing market investments.
Particularly, HBM, which is in high demand from AI accelerator companies like NVIDIA, requires high technical barriers, complex production processes, and stringent customer certifications. Memory manufacturers are prioritizing production capacity for HBM and high-performance DRAM for servers, which may constrain the supply of general-purpose memory used in smartphones and PCs.
This creates an important economic paradox: even if sales of PCs and smartphones slow down, memory prices can still rise. This is not due to an overall increase in demand but rather a change in the allocation structure of production capacity.
The basic principles of economics remain valid. Prices are determined at the intersection of supply and demand. However, the key variable in price determination has shifted to how much production capacity is allocated to which products rather than the total semiconductor production volume.
In this environment, even large buyers like Apple find it challenging to pressure suppliers unconditionally. In the past, bulk purchases were a weapon for price reductions, but in a supply-constrained environment, long-term purchase agreements and stable partnerships become prerequisites for securing volume.
However, it would be a mistake to believe that supplier dominance will last forever. The history of the memory industry repeatedly demonstrates how excessive confidence formed during boom periods can lead to significant losses during downturns. Misjudging the current supply shortage as a permanent structure could sow the seeds for the next supply surplus.
Second, the rise of CXMT is not merely a matter of low-cost competition.
What South Korea should pay more attention to is the growth strategy of CXMT. For a long time, the Chinese semiconductor industry relied on advanced countries for technology and equipment while attempting to enter the market through price competitiveness. However, U.S. export controls and restrictions on access to advanced semiconductor equipment have forced Chinese companies to change their strategies.
Before securing market share abroad, there is a growing emphasis on leveraging the vast domestic market in China as a foundation for technology development and production expansion.
China is home to electronics manufacturers like Huawei, Xiaomi, and Lenovo, as well as large server and cloud demand companies. As these companies expand their domestic semiconductor supply chains, firms like CXMT can secure a certain level of demand before fully entering the global market.
This is a significant change. In the past, latecomer semiconductor companies struggled to expand production scales due to a lack of customers, but Chinese companies can accumulate technology validation and production experience based on their domestic market. When combined with government industrial policies, financial support, and supply chain diversification strategies from purchasing companies, the pace of catch-up for latecomers can accelerate.
Of course, China's technological self-sufficiency does not mean it has surpassed the technological advantages of South Korea and the United States. There are still barriers to overcome in yield rates for advanced processes, production costs, equipment accessibility, product reliability, and customer certification for high-performance memory.
Particularly, claims that next-generation memory has been supplied to specific smartphones cannot be equated with achieving world-class mass production competitiveness. Demonstrating technology and supplying prototypes are distinctly different from large-scale commercial production.
However, it is also risky to underestimate China's pursuit solely based on technological disadvantages. China's competitiveness does not stem merely from the performance of a single semiconductor but from the vast industrial ecosystem that purchases and utilizes semiconductors.
What South Korea should be wary of is not just the growth of a single Chinese company but the formation of an independent semiconductor economic zone within China, where design, production, purchasing, and consumption are interconnected.
Third, Samsung and SK Hynix must not rest on their HBM success.
The South Korean semiconductor industry stands at a critical crossroads. Samsung Electronics and SK Hynix are core companies leading the global memory industry. In particular, SK Hynix's competitiveness in HBM has once again elevated the status of South Korean semiconductor technology. Samsung Electronics also possesses a broad business foundation encompassing memory, foundry, and system semiconductors.
However, current competitiveness does not guarantee future dominance.
The most dangerous moment in the semiconductor market is not when performance is poor but when performance is too good, leading to a failure to recognize changes in the competitive environment.
HBM is undoubtedly a key product in the AI era. However, not all memory can become HBM. AI servers require not only HBM but also high-capacity server DRAM and storage devices, while various types of memory are used in smartphones, automobiles, industrial robots, and home appliances.
Especially in the upcoming physical AI era, demand for memory capable of low power consumption, high reliability, and real-time data processing will increase in robots, autonomous vehicles, smart factories, medical devices, and more.
If South Korean companies focus solely on HBM while Chinese companies expand their customer base in the general DRAM and low-power mobile memory markets, their long-term market dominance may weaken.
Therefore, Samsung Electronics and SK Hynix must pursue two strategies simultaneously. One is to maintain a technological gap in HBM and next-generation high-performance memory, while the other is to secure competitiveness in cost, quality, and supply stability in general and specialty memory.
Samsung Electronics should enhance its technological competitiveness and customer trust in HBM while strengthening synergies between its foundry and memory businesses. SK Hynix should connect its competitive advantage in HBM to the next generation of products while maintaining a balanced business portfolio.
Both companies should prioritize investment returns, customer structure, and long-term profitability per product over the scale of capital investment. High sales and operating profits alone do not equate to good management. True competitiveness lies in the ability to generate sufficient returns even during economic downturns.
Fourth, the U.S.-China semiconductor war is dividing the market beyond technological competition.
The growth of CXMT should not be viewed merely as a rise of a Chinese company. Behind it lies the semiconductor hegemony competition between the U.S. and China.
The U.S. is restricting China's access to advanced semiconductors and manufacturing equipment. In response, China is investing heavily in building a semiconductor supply chain centered around its domestic companies. While the U.S. aims to maintain a strategic advantage in advanced technology, China's goal is to secure an industrial self-sufficiency foundation that remains resilient against external pressures.
In this competition, South Korea finds itself in a challenging position. The U.S. is a key security ally and an important market for semiconductor technology, equipment, and AI industries. China, on the other hand, is both a massive semiconductor demand source and a rapidly growing competitor.
South Korea cannot abandon technological cooperation with the U.S., nor can it ignore the importance of the Chinese market. However, it also cannot continue to engage in unprincipled balancing between the two countries.
South Korea's choices must be clear. In advanced technology and security areas, it should strengthen reliable international cooperation, while in trade and industrial sectors, it should maximize legitimate market access and supply chain stability for companies.
Above all, semiconductor policy should not be treated merely as a diplomatic rhetoric. Semiconductors are a core asset of national security and an industry that supports jobs, exports, and national finances. Diplomacy must protect the industry, and the industry must enhance the country's strategic autonomy.
Recent developments indicate a need for a more serious assessment of China's technological pursuit. Reuters reported on October 28 that CXMT is pushing for limited production of advanced HBM3E despite low production yields. According to the report, CXMT's yield is around 20-30%, still significantly lower than the approximately 70% cited for SK Hynix, indicating that while China's pursuit is real, there remains a gap in advanced mass production competitiveness.
While it is essential to maintain technological gaps, it is also crucial not to mistake competitors' limitations as permanent. Today's technological barriers do not guarantee tomorrow's market barriers.
Fifth, we must confront the paradox of booming semiconductor sales and falling stock prices.
Even as memory companies improve their operating performance, stock prices can decline. The stock market tends to prioritize future profits over current earnings.
Investors are wary not just of simple performance declines but also of how long high profits can be sustained, whether expansion competition will lead to oversupply again, and how much the pursuit by Chinese companies will weaken the pricing power of existing firms.
It is important to distinguish here. The 40-50% drop in stock prices for the three memory companies mentioned in Jeong Byeong-seo's analysis must be evaluated in the context of the reference peak, comparison point, and currency basis. One should not generalize the decline of individual stocks as a common phenomenon across the entire industry.
More importantly, changes in corporate value matter more than stock price declines. A stock price that has halved is not necessarily undervalued, nor is a company that has achieved record profits necessarily overvalued.
The investment principles emphasized by Warren Buffett ultimately focus on a company's intrinsic value and sustainable competitive advantage. It is essential to look beyond short-term market excitement or fear to assess long-term cash generation ability and management quality.
The same applies to evaluations of Samsung Electronics, SK Hynix, and Micron. One must consider how sustainable the high profitability of HBM will be, how much the rise in general DRAM prices will translate into actual cash flow, and whether massive capital investments will lead to future productivity and competitiveness.
Corporate management must apply stricter principles of capital allocation during boom periods. Balancing facility investment, research and development, financial soundness, and shareholder returns is crucial. Unconditional expansion and excessive shareholder returns solely for short-term stock price support are not desirable.
The best managers in the semiconductor industry are not those who invest the most but those who invest in the most productive areas at the most appropriate times.
Sixth, South Korea must become a leader in the AI industrial ecosystem beyond being a semiconductor powerhouse.
South Korea's semiconductor policy has overly focused on production facilities, export performance, and technology development. While these elements are important, in the future, national competitiveness will be determined not only by who produces semiconductors and how much is exported but also by what industries and services are created using those semiconductors.
The competition in the AI era does not end at semiconductor factories. It extends to data centers, cloud computing, AI models, industrial software, robotics, automotive, healthcare, defense, and energy.
In particular, the development of agentic AI and physical AI presents new opportunities for the memory industry. If agentic AI can autonomously perform complex tasks beyond human instructions, physical AI will implement that intelligence in real-world machines and robots.
As these technologies spread, semiconductors will become the foundation of intelligence needed not only in data centers but throughout industrial sites and everyday life.
South Korea must seize this opportunity. It already possesses an industrial foundation that can combine semiconductor manufacturing competitiveness with automotive, shipbuilding, machinery, battery, telecommunications, and medical technology.
The government should not treat semiconductor and AI policies as separate industrial strategies but should promote them as a national-level integrated strategy. An ecosystem should be created where semiconductor companies, AI developers, manufacturers, universities, and research institutions can jointly develop and commercialize technologies.
Not only the semiconductor production bases in Yongin, Pyeongtaek, and Icheon but also industrial complexes and research hubs across the country should become the main players in the new AI industrial revolution. The physical AI industry initiative in North Jeolla Province is also worth considering within this national strategy.
To achieve this, stable power and water supply, skilled workforce, transportation networks, and research and development infrastructure must be supported. Simply promising massive subsidies will not cultivate world-class semiconductor and AI companies.
Creating a predictable system and environment that allows companies to invest long-term is the government's most important responsibility.
Seventh, the ultimate competitiveness of the semiconductor industry lies in investment in people and the future.
Looking back at the history of the global semiconductor industry, there have been no eternal winners. The U.S. pioneered the memory industry, Japan once dominated the global market, and then South Korea secured leadership through technological and production competitiveness. Now, China is rising to new challenges based on its vast domestic market and national resources.
The lesson from history is clear: technological superiority can only be maintained through continuous innovation.
However, the starting point of innovation is not machines but people. Building advanced semiconductor factories requires massive capital, but developing new technologies and breaking through process limitations requires creative talent.
South Korea must fundamentally overhaul its semiconductor talent development system. It should not stop at merely increasing university department quotas but should strengthen convergence education that connects design, processes, materials, equipment, packaging, and AI software. A research environment should be created where young researchers can take long-term challenges, and a system should be established that rewards achievements while allowing for failures.
It is also crucial to create an environment where not only large corporations but also small and medium-sized enterprises and startups in semiconductor materials, components, and equipment can thrive. A strong industrial ecosystem cannot be completed by a few large companies alone. Sustainable competitiveness is formed when capable small and medium-sized enterprises, research institutions, and universities grow together.
Additionally, one principle must be added: the purpose of industrial development is to enrich the lives of the people.
The success of the semiconductor industry should connect corporate profits and stock price increases to quality jobs, regional economic revitalization, and expanded opportunities for future generations. Given that it is an industry that consumes massive amounts of power and water, environmental sustainability and coexistence with local communities should also be core values of management.
Technology exists for humanity. Industries must grow alongside society, and economic development must continue in harmony with nature.
Conclusion: More Worrisome than Apple's Rejection is South Korea's Self-Satisfaction.
The controversy over Apple's supply negotiations with CXMT is not merely a corporate transaction issue. It serves as a warning that the supply structure of the global semiconductor market is changing and that China's industrial ecosystem is emerging as a new competitive variable.
However, one negotiation breakdown report should not lead to the conclusion that power in the semiconductor market has completely shifted. It is also uncertain how long the current supply shortage will last. The sustainability of AI investments, the timing of new production facilities coming online, the production yields of Chinese companies, and global economic conditions and final demand are all variables.
What South Korea must do is not fear this uncertainty but secure new competitiveness within it.
Samsung Electronics and SK Hynix should connect their current high profitability to investments in future technologies. The government must prepare a long-term industrial strategy that encompasses semiconductors, AI, and physical AI, going beyond short-term support measures. The National Assembly should focus on building institutional foundations for industrial competitiveness and the nation's future rather than engaging in political strife.
Investors should also avoid fixating solely on short-term fluctuations in performance and stock prices during boom periods. They must objectively assess corporate technological competitiveness, cash flow, investment efficiency, and long-term market dominance.
The investment philosophy of John Templeton teaches us to find opportunities when pessimism is extreme and to guard against risks when optimism is excessive. Warren Buffett's investment philosophy also emphasizes the importance of intrinsic value and long-term competitive advantage over market noise.
The insights of these two giants apply equally to today's semiconductor industry. We must not be intoxicated by booms or succumb to downturns. We should not be swayed by market fears, nor should we underestimate the rise of competitors.
The history of semiconductors shows that yesterday's winners are not guaranteed to be tomorrow's victors. As South Korea, which grew while witnessing Japan's decline, now faces China's pursuit, the choices made today will determine whether it follows Japan's path or writes a new history of innovation.
Far more important than whether Apple was rejected by a Chinese company is the question of whether South Korea can create technologies and values in the future semiconductor market that are irreplaceable. Companies that determine prices are strong, but those that shape the future of the industry are even stronger.
The South Korean semiconductor industry must now leap beyond being a country that simply produces memory to become a nation that creates a new industrial order in the AI era. This is the challenge for Samsung Electronics and SK Hynix and the path to securing the future of the South Korean economy.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.