晶圓代工廠與IC設計廠商互動關係演進之研究

Abstract

1980年代半導體產業產生第二次結構性變革後，誕生了專司IC規格開定與產品設計的無晶圓廠IC設計廠商，隨著末端產品的快速變化、低廉成本要求與製程不斷精進推陳出新，IC設計廠商與晶圓代工廠進一步衍生出與兩造廠商虛擬垂直整合息息相關的各項互動過程。面對不斷引發的製程難題與設計困境，兩造廠商必須相互提供解決方案共同解決愈來愈多的製程挑戰，雙方的合作關係更形緊密。因此，本研究企圖運用全面且系統性的觀點探討晶圓代工廠與IC設計廠商在互動關係上的演進歷程，進一步透過理論進行預測產業未來發展。 藉著質性研究方法，以目前身居晶圓代工廠龍頭角色的前三大技術與產能水準相互競爭的晶圓代工廠台積電、聯電與IBM陣營與其客戶作為研究標的，透過大範圍次級資料之蒐集，佐以個案公司之深度訪談，歸納出兩造廠商在互動關係的演變歷程，並且以協同合作思維針對互動關係之變化作深入剖析，輔以交易成本理論分析，推導本研究之準命題。最後根據互動關係的歷史脈絡，提出進一步的預測，提供半導體產業從業者、晶圓代工廠與IC設計廠商些許的洞見與建議。 本研究發現技術的演進與各晶圓廠的製程能力導致晶圓代工廠間產生不同的競爭型態，因而引發其對於互補性資產與資本需求而協同其它夥伴進行合作。因應如此的改變也使得晶圓代工廠與其顧客之間由「單純交易關係」、「策略合作關係」進而走向「全面整合關係」。在單純交易時期，晶圓廠與IC設計廠商間互動較少。0.25微米製程後，為能與其他競爭者有所差異並搶佔市場，晶圓代工廠開始向外尋求IC設計廠商、設計服務、IP服務、EDA工具等第三方公司提供互補性資產或服務以爭取客戶訂單，晶圓廠與IC設計廠商開始進行雙向的資訊流動，形成策略合作關係。進入90奈米製程，晶圓廠面對钜額的資本需求門檻壓力，積極尋求第三方公司與IC設計廠商的協同合作，擴大合作範圍，補足技術能力以追求技術發展的突破，企圖成為市場標準。此時降低投資金額、分散風險可謂全面整合關係的關鍵。 面對雙方日益攀升的交易成本，本研究建議晶圓代工廠與IC設計廠商應發展一套適當的控制機制，以降低合作的交易成本。除此之外，晶圓代工廠與IC設計廠商應慎選聯盟合作夥伴，強化長期的合作關係，使雙方合作更為順暢，為彼此創造雙贏的局面。

In the last decade, the semiconductor industry has evolved from vertically integrated to vertical disintegration structure. The Foundry/Fabless model has become the core of the vertical disintegration structure. As design and manufacturing complexities continue to increase, the interdependence between process technology and product design also increases significantly. It is becoming less practical for a Foundry to internally develop advanced process technology without collaborating with a IC design house. A newly collaborative mechanism for design-for-manufacturing is required. The goals of this study are to explore the evolutionary business models between the Foundry and Fabless. By conducting field interviews and empirical study, this research will summarize different phases of business models, describe the transaction targets, and provide some managerial implications. By three-layered framework to describe the evolutionary business models in which the strategic goal, value proposition and engineering processes are defined. There are three phases in the evolutionary business models. Before 0.25um technology, the Foundry/Fabless relationship can be described as a transaction-based relationship. In this phase, Foundry focus their core competency on manufacturing capability and provide logistics services to many Fabless design houses because of their economies of scale. They also provide better manufacturing services such as turnkey service, on-line order tracking and yield data access. The critical factor for this relationship is to have reliable and flexible manufacturing resources. For 0.18um and 0.13um technology, the relationship has evolved to be partnership-based and advanced process capability becomes another important value proposition. As a result, engineering processes emphasize technology development such as yield enhancement, AEC/APC, and IP/Library access. The critical factor for this relationship is to have quick yield ramp-up and verified advanced processes. As technology enters into 90nm and below, design complexity and manufacturability become the key issues. The relationship is now collaboration-based. For DFY and DFM purposes, engineering processes such as design service, reference flow, Litho/CMP friendly design, and engineering chain collaboration become important tasks. The critical factor for this relationship is to have lower risks in capital investment and technology development to facilitate rapid time-to-market.