Red De Suministro

Brian Krzanich, Intel general manager of Assembly Test (AT), looked through his deck of slides one more time. It was March 2005, and in a few days, he would present the AT team’s proposal for the siting of its next AT factory to Intel’s board. The new facility would be Intel’s largest AT plant to date, doubling the size of any existing AT plant and providing the company with more efficient capacity. In 2005, industry average costs to build a new AT factory ran about $80 million with annual operating costs of between $150 and $300 million. He thought back to the fall of 2001, when Intel’s global site selection team had first started gathering data on possible sites for a new AT plant. There were a host of considerations implicated in this proposal, with operational and strategic dynamics as well as national and international relationships at stake.

In their preliminary study of possible sites, Krzanich and his team had focused primarily on Asian and South East Asian locations, given that between 2002 and 2005, the total cost of operations in these countries were still the lowest in the world, and these markets represented important and growing opportunities for Intel. While U.S. regulations had prohibited the construction of a semiconductor fab in China, given intellectual property (IP) concerns, Intel had operated an AT plant in Shanghai for almost a decade with a second AT plant opened in Chengdu in 2003; the firm also had several university research and innovation labs in China. Would China be the best place for a third AT plant? India had also surfaced to the top of the short list. Intel’s chairman was knowledgeable about the country and its leadership, and some felt the time might be right to build a plant in India. Several other Asian countries also presented viable sites, and locations in Latin America and the Middle East were also being considered.

Several years of data collection, analysis, site visits, and on-the-ground negotiations lay behind the short list. In just a few days, Krzanich was scheduled to travel to Chennai, India, one of the potential sites, with members from Intel’s Finance and Enterprise group to pursue additional negotiations with local officials. Krzanich wanted to be sure he had the board’s approval before investing time in the next stage of negotiations. What would the board decide?

________________________________________________________________________________________________________________

Professor Juan Alcacer and Associate Director Kerry Herman, Global Research Group, prepared this case. Some data in this case have been disguised. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management.

Copyright © 2012, 2013 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-545- 7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu/educators. This publication may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.

This document is authorized for use only by Carmen Ortiz Zaldivar at Universidad Adolfo Ibanez (UAI) until February 2014. Copying or posting is an infringement of copyright. Permissions@hbsp.harvard.edu or 617.783.7860.

9-713-406

REV: DECEMBER 2, 2013

Do Not Copy or Post

713-406 Intel: Strategic Decisions in Locating a New Assembly and Test Plant (A)

The Semiconductor and Microprocessor Industry in 2005

In 2005, the semiconductor industry generated approximately $226 billion in revenue worldwide, about $17.5 billion of which was generated in the U.S.1 The industry was expected to hit over $300 billion by 2008, driven primarily by demand in consumer products such as personal computers, digital cameras, televisions and cell phones.2 The semiconductor industry was fragmented, with the top four companies (Intel, Renesas Technology, Samsung and Texas Instruments or TI) controlling about 30% of production.3 The industry produced silicon-based electronic circuits for use in computers and other electronic devices.4 One of the major uses of these semiconductors was in microprocessors. In 2005, the top two companies, Intel and Advanced Micro Devices (AMD), accounted for over 90% of global sales of microprocessor production.5 (Exhibit 1 provides additional information on the industry in 2005.)

The microprocessor market had been historically dominated by Intel, which held a market share in excess of 80%. The firm’s continuously high-paced innovation and regular introduction of new (faster, better, etc.) microchips quickly rendered competitors’ offerings obsolete. Intel’s manufacturing prowess meant it had the ability to slash prices while maintaining healthy margins (as unit sales increased and production costs declined due to economies of scale). One analyst noted, “Typically, by the time competition can react, Intel has already moved on to the next generation of chips.”6 Penetration of top-tier PC vendors—key accounts in the industry—was also a factor, as large corporate clients favored high-speed processors, a product with higher-margins that made this segment highly attractive. Dell’s November 2004 announcement that it would consider using AMD’s new chips, for example, was considered a significant break for AMD.

Chip Production and Manufacturing

Fabrication plants, or “fabs” as they were called, produced semiconductors (i.e., chips) on wafers made of pure silicon.a Much of the production was automated, relying on robots working in cleanroom environments. Semiconductor manufacturers followed one of two models in their chip production. Some designed and manufactured their chips; these firms were known as integrated device manufacturers (IDMs). Others outsourced manufacturing to foundries, and were called “fabless.” Typical costs to build a 300 millimeter (mm) wafer fab ran about $3.5 billion to $4 billion in the U.S., and about $2.5 billion to $3 billion for an international location. Fabs cost about $1 billion a year to operate, and required highly skilled engineers to run. In 2005, several manufacturers were having difficulty keeping up with the investment required to maintain fab competitiveness, and were shifting to a “fab-lite” model in which they outsourced an increasing share of their production.

Production of a chip typically consisted of 300 to 350 individual steps that took several hours each, and throughput time for an average wafer could be 45

...