半導體晶片設計廠商事業發展策略分析：以輝達為例

Abstract

在人工智慧、大數據成為風雲題材之際，2017年度《麻省理工科技評論》(MIT Technology Review) 公布了全球50大最聰明公司榜單，前五名登榜公司的第一名即為NVIDIA，由2015年位列榜上第28位，2016年進步到12位，2017則一舉搶下冠軍寶座。一家製造GPU、以解決圖像問題為主要業務的公司，如何在21世紀之初嗅到整個大數據潮流的發展，進而解放GPU平行運算功能，使GPU應用從傳統圖像處理轉移到運算與深度學習，是非常值得探討的問題。 本研究透過個案研究法，進行初級與次級資料蒐集，分析NVIDIA的產業環境、競爭者狀況，了解GPU產業的NVIDIA過去是如何透過創造獨特價值、追求高品質，在高端市場、高附加價值的目標設定下，創造開發端到使用者端的生態圈，來達到用戶的黏性，並進而開拓新的產品應用市場。NVIDIA商業模式主要有兩大重點，分別是平台網絡、槓桿與規模效應，進而創造和用戶間的良性循環。平台網絡部分藉由策略手段使用戶能留在自己的生態系統中，並由開源和授權手段去影響力不能及的市場，創造未來潛在商機；槓桿和規模效應則以一個架構為基礎，將其發揮到人工智慧、遊戲、資料中心、自動駕駛等不同的終端市場應用，使基礎架構發揮更大功能，且減少不必要資源浪費、以最低成本達到最大效益。 最後，透過分析輝達的發展策略與對應資源能力、重要里程碑與關鍵決策等，歸納出管理上的意涵，並對個案公司與其所要轉向切入的AI運算產業提出策略建議，包含產業發展重點、未來可能須注意的動態發展方向等。

While Artificial Intelligence(AI) and big data became hot topics, NVIDIA got first place on the list of 50 smartest companies on the world, which announced by 2017 MIT Technology Review. At 2015, it ranked 28th, then progressed to 12th in 2016, and finally hit the champion on 2017. How can a company which manufactures GPU and used to solve graphic problems got the future data trend and leveraged their key strength to parallel computing? How they make GPU application transfer from traditional graphic processing to computing and deep learning? The research aims to answer what kind of environment did they encountered, what’s the key resources and ablilities to success. By case study method, this research analyzes initial and secondary data in order to get the industry environment, also competitive condition. Then look into how NVIDIA create unique values, pursue high quality and construct an ecosystem between developer and users. The business model of NVIDIA has two major focus, one is create ecosystem for users through platform network, and the other is leverage the economic of scale through one framework. NVIDIA use some strategies to keep users stay in their ecosystem, and affect the market which they are unable to reach through open API and licensing. Moreover, through one framework strategy it can leverage the economic of scale to different final applications with lower cost. Through the study of development strategies and key resources, key decisions. Finally, we induct some strategy recommendations for NVIDIA and the AI industry which NVIDIA is intended to enter.