全球利基型動態記憶體之產業與競爭分析

隨著5G、AI與物聯網等科技發展，帶動全球利基型DRAM的需求上升。DRAM產業具有資本密集、高進入與退出障礙、價格彈性低、投資遞延等特性，使廠商易於陷入價格競爭中。而利基型DRAM 廠商不僅面對產業價格波動及景氣循環的風險，還須面對產品製程演進緩慢與市場規模小的雙重挑戰。本研究試圖從目前台灣DRAM產業現況歸納出產業供需波動的關鍵因素，並藉由產品生命週期、五力分析、價值網、競爭策略、商業模式等分析，探討全球利基型DRAM的產業與競爭分析，提供廠商未來發展的可行策略，主要研究發現如下。第一，本研究透過長期供需波動，描繪出DRAM產業的蛛網模型，並提出DRAM產業蛛網現象的主要原因，在於投資遞延所導致的產品生產周期較長，當新應用或既有應用升級時，或當廠商一窩蜂擴廠增加產能時，都會加劇價格與產量波動的蛛網現象。第二，以五力分析檢視DRAM產業競爭狀況，發現現有廠商之競爭、供應商議價能力為二大威脅來源，買方議價能力次之，而潛在競爭者進入門檻高，以至於短期間威脅性低。第三，以價值網模型分析DRAM廠商之間的合作關係，由於DRAM產品需仰賴系統主晶片才得以發揮功能，故最主要的互補者為主晶片商。DRAM 廠商應加入相同目標市場的主晶片商所建立之平台，擴大在生態系統中的影響力。第四，以產品生命週期分析DRAM產品不同製程、不同類別的技術推進，發現相較於過去，現今DRAM產業在新製程推進的難度以及所需投資金額更高，而單一世代在市場存在的時間也更長。第五，探討目前市場前五大自有晶圓廠DRAM廠商所採取的競爭策略，結果發現DRAM廠商，例如華邦電子，採取集中化策略，鎖定利基型DRAM產品市場，提供符合市場所需產品及技術，即使產品已進入衰退期，但整體需求量仍遠大於案例公司所能供給的量。並以華邦電子為例，討論利基型DRAM廠商如何建構可行的商業模式。最後，建議利基型DRAM廠商未來應：(1)透過與主晶片進行SIP(System In Package)／MCP(Multi-Chip Package)或是提供客製化規格產品的方式，持續深化與主晶片廠商、客戶的關係；(2)提供長生命週期的產品，減低主晶片廠商與客戶需重新進行產品設計的困擾; (3)掌握上游原物料供貨情況，確保供應鏈來源無虞穩定，以維持DRAM市場的穩定，滿足下游客戶的需求；(4)關注各類應用市場的發展，了解市場未來可能需要的記憶體類型、製程及各種先進封裝技術，持續投入研發資源，以提升效能、降低成本，使利基型DRAM市場能穩定長久的發展。

關鍵字

利基型記憶體；蛛網理論；產品生命週期；五力分析；價值網；競合；競爭策略；商業模式

並列摘要

With the development of technologies such as 5G, AI and Internet of Things, the global demand for specialty DRAM has been rising. The DRAM industry is characterized by capital-intensity, high entry and exit barriers, low product differentiation, low price elasticity, and deferred investment, making price competition prevalent among competitors. Specialty DRAM firms thus face not only risks of price fluctuations and industry cycles, but also challenges of slower product process evolution and smaller market size. This research aims to examine the key factors underlying the supply and demand fluctuations of the DRAM industry, and analyzes its industrial characteristics such as product life cycle, five forces structure, and value net. In addition, this research examines the competitive strategy and business model of Taiwanese DRAM firms. The findings of the study provide suggestions of future development to DRAM firms. The main research findings are as follows. 1. This study describes the Cobweb Model of the DRAM industry through long-term supply and demand fluctuations, and proposes that the main reason for cobweb phenomenon in the industry is the longer production cycle caused by deferred investment. When new applications or existing applications are upgraded, or when manufacturers rush to increase production capacity, it will exacerbate the cobweb phenomenon of price and quantity fluctuations. 2. By using the five force analysis to examine the competition in the DRAM industry, this research identifies the major threats within the industry: intensive competition among existing firms, strong bargaining power of suppliers, and bargaining power of buyers. The threat of potential competitors is low in short-term due to high entry barriers. 3. The value network model is used to analyze coopetition among DRAM firms. Because DRAM products rely on the main chip of the system to function, the most important complementor is the main chip manufacturer. Therefore, DRAM firms should join the platforms established by the main chip manufacturers in the same target market in order to expand their influence in the ecosystem. 4. Based on theory of product life cycle, this research then analyzes the advancement of different processes and different types of technologies of DRAM products. It is found that the DRAM industry is increasingly difficult to advance in new processes due to the requirement of high investment. Thus a single generation of DRAM technology can sustain in the market for a longer period of time. 5. By examining the competitive strategies adopted by the worldwide top five DRAM firms with own-fab, it is found that DRAM firm, such as Winbond Electronics Corp., adopts a focus strategy to concentrate on the specialty DRAM market in order to provide products that meet the market needs. Historical data shows that even during the recession period, the overall demand of specialty DRAM is still far larger than the supply provided by the case company. It also discusses the feasible business models for specialty DRAM firms by taking Winbond Electronics Corp. as an example. Several suggestions stand out: (a) DRAM firms should continue to strengthen the relationship with main chip manufacturers and customers by SIP(System In Package)／MCP(Multi-Chip Package) with main chip or developing customized product; (b) DRAM firms should provide long life products to reduce the re-design effort of main chip manufacturers and customers; (c) DRAM firms should have better control over the supply of raw materials in order to maintain the stability of the DRAM market and to meet the needs of customers; (d) DRAM firms should be aware of the development of markets of various applications, and understand various types of memory, advanced process and advanced package technologies that the market may need in the future, and continue to invest in research and development for enhancing performance and reducing costs of DRAM to make specialty DRAM market grow stably for a long time.