從後摩爾時代的產業特性, 探討高階晶圓代工未來的競爭策略-以T個案公司為例

Abstract

晶圓代工向為我國重點發展的科技產業。值此全球半導體產業成長趨緩，摩爾定律已達物理極限的挑戰下，一線業者莫不集中資源開發先進技術，其中以奈米微影技術的研發最為重要。而與微影技術息息相關的光罩製作，其複雜程度與製作成本亦急遽增加，漸非中小型IC業者所能獨力負擔。根據TRI分析，奈米製程的研發成本從 0.13µm的5億美元提高至32nm的14億美元。另外根據 IEK及 IC Insight的預估[1]，全球半導體產業的平均建廠成本從 2000年的16億美元暴增至2015年的54億美元，增幅約3倍。對整合元件製造商(IDM)而言，很少有如此經濟規模的產品能填滿產能，因為資金限制及高階技術研發的瓶頸，不得不放棄自有晶圓廠，並尋求策略聯盟的研發平台或委外代工，這也預告著晶圓代工產業未來龐大的商機。

據台積電董事長張忠謀先生預估，2011年晶圓代工總產值將佔全球IC整體銷售額的28%，這個比例將持續增加到2012年的40%，主要原因在於IDM公司加速委外代工所致。其中以：(1)2009年 IBM、AMD等IDM公司結合特許半導體(Chartered)成立格羅方德(Globalfoundries)[2]；(2)日本IDM廠在2011年311大地震後，加速進行委外代工，並開始走向輕晶圓廠策略以降低風險[3]；(3)龐大的建廠成本及IC設計業的競爭[4]，迫使歐美大部分的IDM公司轉型，採取輕晶圓廠策略。整個IC製造產業正面臨前所未有的改變，分別在CPU、DRAM及Logic產品呈現IST三強鼎立(Intel、Samsung、TSMC)的寡佔局面。其餘二線業者則致力於策略聯盟，發展共同製程平台以降低成本，期能在利基市場佔有一席之地。

在這樣的產業趨勢下產生所謂「More than Moore」的概念，就是要超越摩爾定律，在摩爾定律以外進行反向思考，尤其是成熟的主流製程能不能帶給IC設計業者更多附加價值，而非一味地追求先進製程的微縮技術。事實上許多成熟的製程包括電源管理IC、車用控制晶片、影像感測器(CIS，CMOS Image Sensor)、微機電系統晶片(MEMS，Micro Electro Mechanical Systems)的獲利率並不亞於先進製程技術。據此觀之，半導體產業正朝向兩極化發展：(1)講求效能、省電的高階製程繼續遵循「摩爾定律」，投入巨資研發突破物理、材料極限的新技術；(2)利用相對成熟技術，開發特殊應用晶片的利基市場區隔。

分析群雄並起晶圓代工廠商的營運模式，經由個案公司(全球晶圓代工的標竿企業)的成功因素，探討晶圓代工業者在後摩爾時代的競爭策略及可能面臨的挑戰，特別是高階製程的發展與供應鏈上可能遇上的瓶頸，是本研究探討的主要目的。本研究將從Albert S Humphrey 的SWOT分析、Michael E. Porter的五力分析、競爭策略理論，就後摩爾時代產業的新趨勢，包括整合元件製造商(IDM)委外代工商機、半導體技術發展的兩極化、產業的整併與聯盟、終端產品的質變、IC的異質整合及行動裝置的產業變革，研究晶圓代工服務業的新商機與未來挑戰，並提出因應的競爭策略建議，以供業界參考。

The foundry industry is the focus of high-tech. development in Taiwan. However, the global semiconductor business is facing the difficulties of growth slowdown. As Moore's Law is reaching its limit, the process development cost is going higher and higher. All the top companies have streamlined their resources in developing nanometer lithography and mask manufacturing. According to the analysis of IEK and IC Insights [1], the R&D cost of nano-fabrication rises from 0.13µm’s 0.5 billion to 32nm’s 1.4 billion. Besides, the average cost of new fab start-up also increases from 1.6 billion in 2000 to 5.4 billion in 2015. It is observed that less and less IC companies have the products volume to fill such huge capacity. Owing to financial difficulties of leading-edge technology development, IC manufacturing companies tend to give up their own fabs, and seek the R&D platform of alliance and manufacturing outsourcing. It also means that huge business opportunities for the foundry in the future.

As Moore’s Law is reaching its limits, the concept so-called "More than Moore", it is the IC industry goes for another solution rather than the pursuit of advanced nano-techniques. In fact, the profitability of many mature process including power management IC, automotive chips, image sensors(CIS，CMOS Image Sensor), and MEMS (MEMS，Micro Electro Mechanical Systems) are as much as the advanced process technology. Therefore, there are two paradigms shift in the industry: (1) high-end technology follows Moore’s Law to drive the transistor geometry down to 10nm node; (2) the use of mature technology in development of niche market segments, such as the application-specific chips.

In the analysis of industry characteristics of post Moore’s Law era, which include: (1) opportunity of foundry outsourcing; (2) paradigm shift of technology development; (3) merge & acquisition in the supply chains; (4) heterogeneous IC systems architecture; (5) industry revolution of mobile computing. The purpose of this study is to develop new foundry business model in post Moore’s Law era, especially in the surge demand of mobile devices of year 2010. This study will also analyze the competitive landscape of top-5 foundry companies: TSMC, UMC, Globalfoundries, Samsung LSI and Intel, via Albert S Humphrey SWOT, Michael E. Porter five force analysis, and to explore the competitive strategy suggestion for the advanced and matured IC companies.