You are on page 1of 34

CASE: SM-156

DATE: 01/04/07

INTEL CENTRINO IN 2007:
A NEW “PLATFORM” STRATEGY FOR GROWTH

INTRODUCTION

In March 2003, Intel launched its new Centrino® mobile platform for use in notebook
computers. By early 2007 the platform had achieved strong success for the company. Creating
Centrino, however, had required Intel to make major changes to its strategy and organization.
The development of Centrino was part of Intel’s “right hand turn” toward multiple performance
vectors beyond maximizing clock speed, including improvements coming from increased power
efficiencies, form factor and connectivity. This strategic shift, together with the introduction of
new multi-core architectures,1 fundamentally changed Intel’s definition of success for the future.
It was a dramatic move forced on the company, in part, by physics and changing industry and
competitive forces; but also made possible, in part, by a radically innovative microprocessor
architecture developed by its scrappy, geographically distant microprocessor design center in
Israel.

Intel had built its reputation developing and selling increasingly fast microprocessors for PCs.
In 2000, Intel was by far the industry leader in the most profitable microprocessor industry
segments. Desktop PC microprocessors were still the main source of Intel’s revenue and profits
and in many ways this segment dominated the company’s outlook. Yet there were signals in the
industry that things were changing that could impact Intel’s microprocessor leadership position.

The mobile computing segment in particular was growing rapidly. These computing devices,
such as ever thinner and lighter laptop PCs, relied on battery power and required microprocessor
architectures that were fast yet power-efficient. Precisely to serve this vital segment, Intel had
1
Multi-core refers to placing two or more computational engines within a single processor. For more information,
see: “Intel Multi-Core Processor Architecture Development Backgrounder,”
www.intel.com/cd/00/00/20/57/205707_205707.pdf
Professor Robert A. Burgelman and Philip E. Meza of the Stanford Graduate School of Business and Evan Berrett of
Intel Corporation prepared this as the basis for class discussion rather than to illustrate either effective or ineffective
handling of an administrative situation.

Copyright © 2007 by the Board of Trustees of the Leland Stanford Junior University. All rights reserved. To order
copies or request permission to reproduce materials, e-mail the Case Writing Office at: cwo@gsb.stanford.edu or
write: Case Writing Office, Stanford Graduate School of Business, 518 Memorial Way, Stanford University,
Stanford, CA 94305-5015. No part of this publication may be reproduced, stored in a retrieval system, used in a
spreadsheet, or transmitted in any form or by any means –– electronic, mechanical, photocopying, recording, or
otherwise –– without the permission of the Stanford Graduate School of Business.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 2

developed and launched Intel®2 Centrino®3 mobile technology as a “platform” of components
designed to work together. The Centrino platform included a microprocessor, chipset
(combining graphics and memory capabilities) and wireless communications products. Creating
the Centrino platform marked a new paradigm for Intel that would change the way the company
thought about design, architecture, manufacturing, and marketing.

Given Centrino’s success, Intel’s top management was interested in the lessons that could be
drawn from the experience of creating the Centrino platform to inform the company’s
“platformization strategy” going forward.

GOT GIGAHERTZ?

During the 1980s and 1990s, processor speed was seen as the primary measure of value for
microprocessors in the consumer PC market. Processor speed, also called frequency or “clock
speed,” is the speed at which the processor executes instructions and is often expressed in
Megahertz (MHz), which is 1 million cycles per second or Gigahertz (GHz) which is 1 billion
cycles per second. Over time, Intel optimized its design processes and manufacturing facilities
to produce faster and faster microprocessors while at the same time doubling the number of
transistors found on a given space on a microchip—the latter is the phenomenon first described
in 1965 by Intel cofounder Gordon Moore and popularly known as Moore’s Law.4

Intel’s first commercial microprocessor, the 4004 introduced in late 1971, contained 2,300
transistors and performed at 108 Kilohertz. During the 1990s, as Moore’s Law predicted, the
number of transistors on a chip doubled about every two years, and increased processor speed
followed. By August 2001, Intel’s Pentium® 45 processors ran at 2 GHz and contained over 42
million transistors.

Intel’s own highly successful marketing efforts influenced consumers to value faster
microprocessors. Software vendors and new technologies also helped fuel the desire for faster
processors by creating software that required faster processor speed. By the end of 2000,
processor frequency and design leadership, coupled with the ability to manufacture effectively,
helped Intel grow to over an 80 percent share of market in PC microprocessors, while achieving
margins nearing 60 percent in a maturing product-market segment. At this time, Intel’s internal
development plans (called product roadmaps) continued to rely on processor speed as a key
driver of microprocessor value. But as early as 2000, physical limitations, shifting consumer

2
Intel® is a registered trademark of Intel Corporation or its subsidiaries in the United States and other countries.
3
Centrino® is a registered trademark of Intel Corporation or its subsidiaries in the United States and other
countries.
4
Gordon E. Moore, “Cramming More Components Onto Integrated Circuits,” Electronics, April 19, 1965. Moore’s
Law describes the exponential growth in the number of transistors that could occur on an integrated circuit every
year or two, and predicts its continuation—which has held true for over 35 years. The practical impact of Moore’s
Law was predicted (presciently) in a cartoon in that 1965 article, which depicted “Handy Home Computers” being
sold next to notions and cosmetics in a department store.
5
Intel® Pentium® 4 is a registered trademark of Intel Corporation or its subsidiaries in the United States and other
countries.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 3

demand, increased competition, and a major microprocessor architecture design innovation
provided impetus for Intel to change the way it thought about its future.

THE VIEW FROM SANTA CLARA

The birth of Centrino followed the confluence of two separate streams of events: one playing out
at Intel’s headquarters in Santa Clara, California and the other emanating from the company’s
Israel Development Center (IDC) in Haifa, Israel. (See Appendix for a timeline of key events
leading to Centrino.) It took a couple of market shocks in Intel’s main businesses to bring
together the developments that led to Centrino.

Reorganizing in Good Times

In 2000, Intel was riding the crest of a wave of investment in technology. Microprocessors for
desktop PCs, Intel’s bread and butter, were selling at record numbers. Beyond desktops,
customers were buying increasing numbers of expensive laptop computers too. In fact, during
this time Intel was supplying its own workforce, numbering into scores of thousands of people,
with laptop computers.

While times were good in terms of sales and revenue, Intel suffered some uncharacteristic
missteps. The company experienced a series of production problems that left it short of
inventory. This opened the door for Intel’s main competitor, Advanced Micro Devices (AMD)
to increase its share of the PC processor market from 16.7 percent in 2000 to 20.2 percent in
2001.6 Further, AMD had actually surpassed Intel at its own game of increasing speed,
becoming the first to release a 1 GHz processor and beating Intel to market by two days in March
2000.

In the summer of 2000 Intel made a bet on its new Pentium 4 architecture. Its Microprocessor
Group (MPG) located in Santa Clara and a microprocessor design team located in Oregon had
driven this architecture. In backing the new Pentium 4 architecture, Intel stopped the
development of most of its Pentium III lines. Pentium III development teams in Folsom,
California and in Texas dropped their projects and started working on Pentium 4. One exception
was the IDC in Israel, which kept working on Pentium III derivations.

At the beginning of 2000 there was tension between Intel’s two largest groups with responsibility
for its core business: the Microprocessor Group (MPG), which led microprocessor technology
development (and took a longer R&D view), and the Intel Architecture Business Group (IABG),
responsible for selling microprocessors into various products (which had a shorter term business
focus). In April 2000, top management decided to combine IABG and MPG into one
organization called Intel Architecture group (IAG) under the leadership of Albert Yu and Paul

6
John G. Spooner, “AMD scores points against Intel in 2001,”CNet, January 24, 2002.

we conducted a market segment analysis (MSA) to further explore the market needs.”11 7 Intel sometimes assigns two executives to co-lead a group. 2.000 Intel employees and brought together three separate business groups: Enterprise Platforms Group (EPG).10 It turned out that the third task was to be particularly consequential. or the speed at which the processor executes instructions) was running out of steam. Provide an outside perspective about resource balancing.11. “Israel Inside: The Evolution of Centrino/Banias/Mobility. February 9. . and Mobile Platforms Group (MPG). MPG refers to Mobile Platforms Group and not Microprocessor Group. p.” 2003. 9 This paragraph is informed by David Perlmutter and Shlomo Maital.” Financial Times.8 with each of the three groups having responsibility for profit and loss (i.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Technology and Manufacturing (TMG) and Sales and Marketing (SMG) remained separate functional groups (Exhibit 1). Chandrasekher explained. for example. Otellini alone led IAG. Chandrasekher described the purpose of the centralized planning and marketing group he was assigned to run: 1.e.. clock speed. “Intel Inside Out: The Chip Industry Leader Adapts to Changing Consumer Demands. described the need for Intel to expand its focus beyond clock speed itself and take a more holistic approach. (After Yu transitioned to lead a different group.” Speed Had Become Less of a Differentiator By 1999 the slower clock speeds of lower end microprocessors used in inexpensive PCs were starting to become fast enough to run most available consumer software applications without users experiencing an appreciable difference in performance. 4 Otellini. This led Paul to consider how we could best optimize our organization to take advantage of the market. 11 Chris Nuttall. Centralize marketing so general managers within IAG would not be distracted by marketing responsibilities such as product ramps. unless otherwise indicated. 8 Note: after the 2000 reorganization. Help determine what the market really cared about. gigahertz. it’s not sufficient to drive the levels of growth and innovation that will allow our industry to prosper. Desktop Platforms Group (DPG). running the business) and R&D.” Albert Yu and Paul Otellini co-managed IAG for several months before Yu transitioned to lead Intel’s optoelectronics group.. These studies confirmed that the market was more interested in what a product could do rather than discrete measures of performance. 2005. 10 All quotes from Anand Chandrasekher are from the authors’ interview on November 10.7 IAG would have both P&L responsibility and responsibility for R&D. Speaking at the 2001 Intel Developer Forum. Otellini said. 2006. Paul Otellini. “Paul Otellini and others started to believe that the megahertz and gigahertz orientation (i. 3.9 In an effort to aid the transition to this new structure. then executive vice president of Intel Architecture Group.e.) The combination that created IAG affected some 11. leaving Otellini as sole head of IAG. As early as 2001. Otellini temporarily split out marketing and planning from IAG and centralized these functions under the direction of Intel executive Anand Chandrasekher. In 2000. “While this focus on raw processor speed [clock speed] is important. a relationship it calls “two-in-a-box.

5 Also. at Intel’s Developer’s Forum. 14 Intel® Pentium® III is a trademark or registered trademark of Intel Corporation or its subsidiaries in the United States and other countries. then vice president and chief technology officer of IAG warned of a looming threat to Intel’s long- established focus on processor frequency. distinguish itself from competitors.12 Only four months later. 12 It was the Pentium III chip that contained the controversial “Unique ID” which caused privacy concerns among consumers. which rivaled Intel’s CPUs as the fastest x86 processors in the world for several years. In contrast. Hotter Than the Sun Almost simultaneously. the new Pentium® 4 processor introduced in 2000 featured a new microprocessor architecture called Intel NetBurst®15 technology that gave off more heat than any previous Intel or AMD processor.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. especially at the low end. within 10 years Intel’s chips would theoretically generate energy densities equal to the heat of the sun (Exhibit 2). Colwell. Gelsinger pointed to data that suggested if Intel continued to drive megahertz at its current rate. Intel launched its Pentium III microprocessor. 2006.” Gelsinger asserted there were heat limitations that would eventually have to impact chip design. In a speech titled “Hotter Than the Sun. AMD’s newly gained frequency leadership put pressure on Intel in the desktop arena. 155-156. Put simply. Pat Gelsinger. for the first time in years Intel experienced production and execution problems internally that impacted its output. AMD produced ample supplies of Athlon chips. in June 1999. growing concerns about the physical limitations of a focus on gigahertz were becoming apparent. Intel’s microprocessor roadmap was moving forward along its traditional trajectory of increased clock speed. 13 Intel® Pentium® II is a trademark or registered trademark of Intel Corporation or its subsidiaries in the United States and other countries. Intel would need to find ways other than raw microprocessor speed to bring value to the market. There were only minor differences between this microprocessor and its predecessor Pentium II. winning new OEM relationships and executing well. In January 2001. and support the profit margins the company had long earned. Meanwhile. Although the transition from Pentium® II13 to Pentium® III14 microprocessors in 1999 did not vastly increase microprocessor heat output. AMD announced that its Athlon chip was the first to break the 1 GHz barrier. In February 1999. “The Pentium Chronicles. despite emerging potential technical barriers and the increased competition from AMD. pp. 15 Intel Netburst® technology is a trademark or registered trademark of Intel Corporation or its subsidiaries in the United States and other countries. the ever-faster microprocessors that Intel was developing were also becoming more power-hungry to operate and putting out increasing amounts of heat. Intel was forced to respond with price reductions and as a result its average selling price began to slide from the traditional low $200s into the high $100s directly impacting Intel profitability. For a brief description of this episode. AMD launched its Athlon microprocessor. by 1999 AMD began closing the gap in some market segments. If this were true. see Robert P. On March 6. despite Intel’s long history of product leadership. During the same period.” Wiley Interscience. . 2000.

who together led the desktop PC business. the maximum operating efficiency of the processor was viewed to be more dependent upon the number of instructions executed per cycle rather than the clock speed. Until this time. Paul Otellini. 17 Within Intel. which is measured by the formula: Performance = Clock Speed x Number of Instructions per Cycle. Pat Gelsinger. The 2001 SLRP. • “Power Wall” a discussion of the physical limitations of heat and power that were fast approaching microprocessor architectures. but in our guts we felt that it was poised to really take off.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.” The discussions in this meeting helped inform the agenda for the Product Line Business Plan (PLBP) meeting in September and October 2001. Craig Barrett. conducted in May. Chandrasekher recalled: “The 2001 SLRP identified mobile products such as notebook computers as a significant growth opportunity for Intel. there are many other effective methods to raise performance substantially other than by boosting clock speed. Going forward.” . Intel’s revenues dropped precipitously (Exhibit 3) as the PC market declined in absolute units for the first time ever (Exhibit 4). This was not obvious at the time because notebooks had not been growing much faster than desktops. measured by instructions per cycle. some engineers in Intel’s Israel Design Center (IDC) had already started to think differently about performance. preceding IDC designs had failed to progress to manufacturing. for ideas that emerged from SLRPs. SLRP is pronounced “slurp. Then the Market Dropped In 2001. Anand Chandrasekher led the research for the “Jazz the PC” SLRP theme and Pat Gelsinger led the research for the “Power Wall” SLRP theme. Louis Burns and Bill Siu. The purpose of the PLBP process was to develop business plans. Instead of focusing on “boosting megahertz” (clock speed) at all costs. then chairman. and I) met in a room for eight hours to discuss the SLRP findings.” (See Exhibit 5 for market share estimates for notebook PCs. David Perlmutter. 6 Anticipating the problems that heat generation would eventually pose. the IDC engineers began to focus on the overall performance of microprocessors. called road maps at Intel. had two themes determined by Intel’s CEO. targeted specifically for mobile computers.com/cpu/new_core_conroe/. first appeared on the Intel roadmap for mobile products. 16 Focusing on the microprocessor’s efficiency. Craig Barrett: • “Jazz the PC” looking for ways to make PCs more exciting to users and stimulate sales. See http://www.) Chandrasekher continued: “After the 2001 SLRP. It was in this environment that Intel conducted its 2001 Strategic Long Range Planning Process (SLRP)17 in which the company’s top executives reviewed and debated important issues facing the company.digital-daily. and balancing it with the right frequency goal became the basis of the innovative microprocessor architecture—codenamed Banias—developed by the IDC (see below). having been killed by 16 Within this new paradigm. the general manager of the IDC. eight senior executives (Andrew Grove. It was at the 2001 PLBP that the above-mentioned new Banias microprocessor design.

000 miles away from Intel’s main desktop centers. . These microprocessors were developed by the Microprocessor Group (MPG). had to make do with desktop R&D efforts. and by extension the IDC. This was about to change. the group felt they were in constant “survival mode. on the other hand. established in 1974. codenamed “Chopaka” and “Timna”18 had been cancelled before ever going into production. Not surprisingly. the Mobile business group inherited desktop-oriented microprocessor designs. at that time still a separate functional research group. another small design team in Israel. Chopaka is a lake in Washington state and Timna. manufacturing capacity priority. Each business used microprocessors aimed at the needs of their respective markets. which consisted of four businesses focusing on Servers. Indeed. At the time.” looking for the next project that would provide them with a temporary opportunity to exist. so the mobile segment was seen as a secondary opportunity⎯one that did not justify the rich funding. freedom to drive its own roadmaps. The Israel Design Center. well respected and the primary money maker for the company. and robust R&D efforts that were enjoyed by the Desktop and Server business groups. The Mobile business group. some at the IDC felt more like “internal contractors” to the company than a fully integrated Intel design group. Yet the IDC was at the periphery of the Intel universe. Source of Timna information: http://timna-park. Workstations. over 6. In fact Intel had dropped Timna in September 2000 just days before it 18 Until they are released as branded products. the mobile market was small relative to the desktop market. Feeding the IDC’s sense of insecurity was the fact that its two most significant microprocessor development projects. attracted some of the best graduates from Israel’s most prestigious universities. In addition. or what it would take to win in the mobility segment.il/. the Desktop Group at the company was well funded. was busy solving two important problems that would have a substantial impact on Intel’s future. had gained a toehold on a mainstream line of business for Intel.co. THE VIEW FROM THE ISRAEL DESIGN CENTER Several years before the reorganization of 2000. is the site of the world’s oldest copper mine. given that Intel had over 80 percent share of the PC market at the time. setting the company on a new path toward mobility.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. A New Focus on Mobility In 1998 all microprocessor designs were driven by IABG. a park in Israel. Banias. Intel projects are assigned code names usually taken from physical locations. As such. As result. Oregon were pursuing the trajectory toward increased clock speed. how they would best serve the mobile customer segment. Desktops and Mobile. and relied on leftover marketing budgets. California and Hillsboro. With the 2001 PLBP. the Mobile business group’s executives had not fully articulated to Intel’s leadership how their products should be differentiated from Intel’s desktop products. 7 Intel management in earlier stages for various reasons. was less well funded so it drafted from the work of the Desktop Group. while Intel’s microprocessor development teams in Santa Clara.

unless otherwise cited. Yet while Timna successfully met its primary goals. more dies could be put on a wafer for roughly the same manufacturing cost per wafer. it did not offer sufficient cost reductions compared with the Pentium III. but you’ll have to pay a price for it. These setbacks left the IDC scrambling for its next project and the group feeling even more marginalized. I can put everything you want [including the graphics and memory controller] into the Pentium III die size. Economies were also realized by adding more components to a single chip. A die that is completed and cut from the wafer becomes a chip. One of the engineers said. while maintaining the same die size.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. who thought he could build a cheaper. related a corridor conversation from the fall of 1997 with an Israeli engineer. but ultimately had impact on Intel’s strategic direction for the mobile segment.’”20 IDC senior managers listened as the engineer outlined how a small change in frequency (clock speed) could make a big difference in die area—that there were diminishing returns with increasing chip frequency that might not be worth the tradeoff. . RDRAM cost more than twice as much as alternative technologies. Perlmutter recalled: “There was a group of people congregating near the bulletin board. Timna’s original goal had been to reduce cost. ‘I have a solution for you for die reduction. but you’ll have to let me lower the frequency by 10 percent.19 Timna was never manufactured for sale. 8 was scheduled to hit the factory. Nelson and authors. but perhaps more importantly. then vice president and general manager. Timna was targeted at the “Value PC” category (under $1. March 20. David (Dadi) Perlmutter. 2006. Unknown to the IDC leadership at the time. 20 All quotes from David (Dadi) Perlmutter are from the authors’ interview on February 9. Subsequent quotes from this interview will not be cited. lower-power-consuming processor than the Pentium 4. The Timna Innovation The Timna project had sought to reduce costs by getting more dies on a wafer and cramming more components into each individual die. 2007.000 at the time). Israel Development Center. it kicked off a critical discussion about alternative value propositions beyond 19 Email correspondence between Donald W. IDC engineers had learned that by giving up a little bit of frequency. the group’s experience with Chopaka and Timna had prepared it to play a pivotal role in bringing about one of Intel’s most significant products of the last two decades: Centrino™ Mobile Technology. an unorthodox solution began to emerge. Additionally. While working along these lines. (Note: die size refers to the area used on a silicon wafer to make a chip.) Timna used the Pentium III core architecture while integrating other platform components such as graphics and memory controllers into the microprocessor. yet it yielded several critical discoveries and drove important strategic discussions that would be crucial for Intel’s mobile strategy going forward. which meant reducing die size. so ultimately it was not the best solution for the value PC market. This line of thinking was not only innovative for the successful design of the Timna chip. they could free up a lot of die space (Exhibit 6). which was pivotal not only for the Timna project. By decreasing the size of a die. Simcha Guchman. which Intel had been able to make more cheaply than it initially had planned. Timna relied on a memory technology called RDRAM from memory company Rambus.

the IDC engineers had discovered how to drive power optimization with little or no impact on performance. in the end. “We funded Banias as an experiment. This rule basically optimized power and performance versus optimizing for just performance. .”24 Four Vectors of Mobility Banias’ existence seemed temporary. 9 sheer microprocessor speed. At the time the Banias project was launched. 23 Pentium Pro was the first product based on the sixth generation of the Pentium architecture (P6). The IDC team was surprised to discover that the Timna design reductions not only reduced power use and extended battery life. Otellini recalled. which had been the rule before Banias. at the 21 Capacitance is a measure of the amount of electric charge stored (or separated) for a given electric potential. length of lines. Frequency reduction on the chip. Perlmutter said that overall die capacitance became a design target. and the IDC developed a tool to estimate and measure capacitance across the development process with measures to hit aggressive targets. just like speed and die area size. The intent of the Banias chip was to achieve even higher performance with lower power consumption. Furthermore. He also pointed out that before Banias. Banias had unique power optimization capabilities and efficient performance. the IDC team began work on the Banias chip design. Otellini are from the authors’ interview on December 5. 2005. Another Timna-driven discovery was also critical to Intel’s future mobile strategy.). had little to no impact on overall performance. etc. The IDC team began to think about trade-off options that would have been untenable at Intel in the past. combining the old Pentium Pro23 architecture with learning gained from the ill-fated Chopaka and Timna projects. It was truly a blue sky project to see whether we could combine the learning from Chopaka and Timna with the Pentium III architecture to create a low power [micro-architecture]. but due to second-order phenomena (heat. unless otherwise cited.” Perlmutter explained that power use goes up linearly with capacitance21 and frequency. Banias is Born After Timna. it did not appear on any of the Intel microprocessor roadmaps at the time. Banias was built on a Pentium III core. whereas in the Banias design most circuits would be off and only turned on when needed. 22 Voltage is the difference of electrical potential between two points of an electronic circuit. circuits were always on and only turned off when not needed. Banias was already being designed for lower frequency. Dadi Perlmutter discussed how low-power chip design works: “You have to start designing for low power from the outset. Subsequent quotes from this interview will not be cited. However. and they were rejected in cases where they added power consumption above a certain level. The team began work on Banias in Q4 1999. You can’t put it in later. It was introduced in 1995. The IDC tried to convince others at Intel that Banias and its low power/high performance vector was worth pursuing and worth investing in. Intel was unsure where the chip would fit.22 As noted earlier. and goes up exponentially with the operating voltage.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. performance features were measured against power increase. but had no clear home within Intel. in the Banias design. 24 All quotes from Paul S.

”25 In late July of 2001. it was not included in the original Banias plans and some engineering managers were set against making the weakest element of the nascent platform the center of the brand. Paul Otellini attended. Peleg and Golbert summarized MSA data and articulated a new value proposition for the mobility sector. IDC had been a part of the Microprocessor Group (the old MPG). After the reorganization. Otellini. they reconfirm and reinforce each other. a face-to-face meeting was held with the Mobile Platform Group’s staff. They presented these performance vectors to Otellini who agreed with the strategy and gave Banias initial headway as a project. wireless. Don MacDonald. The next day. two MPG strategists. Intel was concentrating on developing its Pentium 4 line. but the IDC team had learned through hard experience that Intel did not easily accept development paradigms that differed from its push for clock speed. Prior to its reorganization in 2000. later to be called Mobile Platform Group (the new MPG). however. It was clear to Perlmutter that Banias was of value to the mobile vector. 10 time. marketing manager of the Mobile Group at the time. and which Banias addressed well. After the Timna project was cancelled in September 2000. He went to see Frank Spindler. MSA data was gathered to determine what customers valued. Early positioning of Banias as a project to address the value segment of the desktop market was a safe play since that segment did not appeal to the desktop design centers of power. The team concluded that for future products. however. and found out about the importance of form factor and battery life. and form factor or size) that differentiated mobility products. that it was important for the IDC team to talk to the business side in the United States. Perlmutter. IDC was combined with MHPG (Mobile Handheld Product Group). the IDC group held a postmortem in order to draw lessons from the Timna experience. “If you can get people to care about some or all of these. and also work much closer with marketing in order to better define the elements that distinguished the mobile segment from its desktop cousins. MPG believed that it was the combination of four key factors (performance. allowing the IDC to continue its work with Banias along these four performance vectors (Exhibit 7). If we could get people to care about the combination of the four vectors. they would have to break out of “value” (low cost processor designs) and concentrate on fast- growing mobile products. With this change. Perlmutter recalled that 25 All quotes from Don McDonald are from the authors’ interview on March 22. MPG was empowered to drive and fund its own R&D processes and efforts focusing on the needs of the Mobility segment. worked into the night defining how best to articulate the Banias value proposition. By looking at this data MPG executives and engineers found key factors that were important to this segment. a group for whom lowering power consumption had not been an important concern at that time. then we’d have it. Alex Peleg and Adi Golbert. battery life.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 2006. Although wireless communication had been identified as one of the four vectors for mobility. had learned from the Chopaka and Timna experiences that potential technological advantages were not enough. decided to use wireless as the central element of the new platform brand. which included the “four vectors of mobility” that MPG had just identified. . In preparation for the discussion. director of marketing for MPG recalled. Subsequent quotes from this interview will not be cited.

however. First-time buyers were less savvy about what is valuable in a laptop. the Desktop Group. The Banias marketing team also learned a valuable lesson by studying second. used free spectrum. At tape-out (the final step in the design of a microprocessor where the product design is frozen in preparation for manufacturing. MacDonald pointed out: “You couldn’t fit a Pentium 4 in what we now consider a mini-notebook. Some did not believe that Banias would deliver the performance the IDC promised. thinking Pentium 4 would fill the mobility gap as needed. third and fourth- time laptop buyers. 11 by the fourth quarter of 2001: “We had analyzed multiple wireless technologies and closed on Wi-Fi in late 2001. demonstrating strong performance and low power—a perfect mix for mobility.25 inches thick because they used standard desktop chips. ICG changed their Wi-Fi plans from supporting the 802. We will use Intel wireless!” This entailed some risk because Intel’s competency in this technology was untested.11a protocol only.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Perlmutter said that. we . as 802. allowed easy connectivity and did not require lots of infrastructure to deploy. After reviewing the Banias data. That was after a big debate. to supporting a dual mode (802. The industry at that time was selling clunky. Banias products proved worthy. had obtained Wi-Fi expertise in an earlier acquisition and had a group working with this technology in the Intel Communications Group (ICG). Intel’s then current desktop chip.11b was more popular at the time and they were concerned that switching to a new protocol would weaken acceptance. the Banias team engaged in a joint program with ICG. but we knew that people would eventually want standard features in thinner form factors. the Pentium 4 would not fit in thinner form factors. but experienced buyers better understood laptop feature set priorities. Using Wi-Fi was the basis for the four vectors of mobility discussion that Alex and Adi led earlier in 2001.11b). MPG learned that targeting experienced users for research was key to understanding the space. Intel. But Otellini responded. Intel was clearly not a leader in Wi-Fi technology.” HEADQUARTERS TAKES A NEW LOOK AT MOBILE Conflict between Desktop and Mobile As Banias was emerging. which required plenty of space. based on the decision to use an Intel Wi-Fi component.11a and 802. as the proposed alternative was to use Bluetooth technology to connect to cell phones. Initially the Banias team planned to use third-party elements for the wireless component. believed that the frequency levels were too low to use with desktop computers. it competed with the Pentium 4 development team for attention.” Wi-Fi was particularly attractive because it was an IEEE industry standard. “The Desktop Group said. most laptops in 2000 were over 1. however. ‘with these frequency figures. In terms of form factor. Adi Golbert recalled. large notebooks. and masks are created for the wafer fab) in mid-1999. This is going to be Intel’s branded system. “You don’t get it. Public facilities such as coffee shops and other businesses could have Wi-Fi routers and thus provide wireless access as an amenity without investing more in infrastructure.

Thus Banias was designed with the mobile PC in mind and was given funding and marketing support to move forward on its own. In this meeting Otellini and others again stood strong for Banias. Intel’s Mobile Processor Group went by the MPG acronym. It was also in this PLBP meeting that the team discussed Banias as the microprocessor for a new mobility “platform”—linking Banias as the platform microprocessor to other platform components that would make up a full mobility solution that would later become Centrino. 27 All quotes from Alex Peleg are from the authors’ interview on February 22.” Peleg recalled a key meeting in preparation for the yearly business strategy review that Otellini attended in which Banias’ detractors “ganged-up” on the project: “Paul [Otellini] intervened stating flatly that Banias was not on the table for negotiation. Once approved by Otellini and CEO Craig Barrett. feeling that the low power and lower frequency direction Banias was taking would not work for Intel and that the resources could be better used by the core business. we should kill Banias. Otellini supported Banias and the project moved forward. After that reorganization. the Mobility team presented their PLBP to Intel’s top executives. Subsequent quotes from this interview will not be cited. protecting the team. that’s fine.’”26 Banias became a target for criticism in many Intel forums early on.”27 In September 2001. its position on the Intel roadmap was seemingly secure and growing. If you want to use it for mobile products.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. why do we need Banias when we have Tejas [an Intel microprocessor architecture project that targeted very high processor speeds but also consumed lots of power]. mobile platform and architecture would warrant dedicated resources. Peleg outlined some of the early criticisms of Banias: “No one will want it. would this hurt the company’s crown jewels?” Still. 2006. Strategic Dissonance Revealed Anand Chandrasekher. Subsequent quotes from this interview will not be cited. 12 can’t sell it. PLBP meetings are critical at Intel because they determine funding and headcount resources for businesses. . there still was enormous debate about whether Banias would succeed. however. Banias became the flagship product for the Mobile Processor Group (MPG). Now. Some Intel senior executives in the room wanted to move 175 Banias engineers to Desktop Group projects. 26 All quotes from Adi Golbert are from the authors’ interview on March 23. But we are selling gigahertz and you guys don’t have it.28 The mobility team would no longer feel compelled to draft off the Desktop product roadmaps. you cannot sell low megahertz products. MPG referred to Intel’s Microprocessor Group. ‘can you un-train customers to no longer focus on measures like megahertz and gigahertz after 20 years of marketing toward this feature?’ There was the concomitant concern that if you could un-train customers. We grappled with questions like. 2006. 28 Until its reorganization in April 2000. recalled: “At the time that Banias first appeared on the Intel roadmap for mobile products following the 2001 PLBP in September and October. Banias was fully funded and kept its headcount.

Chandrasekher explained.” But in 2001. the higher end mobile products would use Banias and desktop PC processors. “I visited our customers in Europe and found that a company that had the largest market share of notebook computers had been using our desktop microprocessors and putting them into larger. but our pricing and behavior did not support this bet. on the other hand. at this time. we thought some customers would be willing to pay more for notebook-specific microprocessors.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Chandrasekher recalled: “In 2001. and. Chandrasekher observed. Neither group really had visibility into the desktop processors that were effectively finding their way into notebook designs. Our 2001 plan was to completely replace these architectures between 2003 and early 2004. Intel had underestimated the total addressable market for mobile. overlooking the segment of users who wanted mobility but were very price conscious. and the lower end mobile products would use the Pentium III and Pentium 4 architectures. At that time. We made an intellectual bet on notebooks.” This segment of desktop chips used in laptop computers was a blind spot for Intel because it fell between two discrete product groups: Desktop and Mobile. the Mobile Group sold higher priced processors with lower voltage and power consumption levels. 13 One thing that might have swayed Intel’s decision to invest in the mobile segment was the perception. Rather. you can’t relegate your mobile architecture to the high-end niche. the mobile group defined the notebook market as only notebooks that used chips specifically designed for mobile products. cheaper notebooks.” .” Hence. Dadi Perlmutter. pointed out: Banias was never aimed at being a niche product in mobile. Chandrasekher explained: We thought Banias might serve a niche market of mobile customers who would be rather price insensitive.” In addition. according to Chandrasekher. Intel’s product roadmap did not call for all of Intel’s mobile products to converge on the Banias line. that “notebook PCs were more exciting than desktop PCs and had a faster replacement cycle. “The dissonance is that if you are betting on mobile. The debate was about how fast Banias will replace Pentium III and Pentium 4. This showed that the notebook market was in fact price elastic. This excluded a large segment of notebooks that used microprocessors that had been designed for desktops. it appeared that Intel had underestimated the addressable market for Banias. so we felt confident that the mobile segment would be very good for us. The overall notebook market at the time was around 20 million to 25 million units per year. it was initially not clear to Intel how the mobile market would develop. for a time. Chandrasekher said. Indeed. “The Desktop Group sold processors at a given voltage and power consumption level. We thought we would serve the high end with an expensive Banias chip and serve the remaining PC and low-end mobile markets with the Pentium III and later the Pentium 4 architectures. like we thought we were doing.

This plan reversed the order of product introductions that Intel had been accustomed to. In the past. By planning to ramp Banias from top to bottom. not a niche. In a metaphor often used at Intel. as well as the decision to drive the “attach rate” (selling the wireless component together with the Banias microprocessor) up from the then low single-digit level to a significantly greater than 50 percent level with the introduction of the platform brand.parks. and (2) ramp up Banias during 2003. For example Banias is the name of a spring and waterfall located at the Hermon National Park in Israel. and that consumers valued style over functionality. This resolved the strategic dissonance: Banias would indeed not be a niche product⎯just as the IDC team had planned. “We had four vectors of performance but did not articulate them coherently: We needed a brand. valleys or towns. Intel would have to dedicate marketing resources away from its core desktop microprocessor business during a downturn.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. but it would also “reverse waterfall” into the market: corporate customers. would be first to purchase Banias.” Intel now had its performance vectors defined. and then later by corporate customers. followed by the consumer market. Intel turned to a branding firm called Lexicon in Sausalito. it receives and is promoted under a commercial name. Banias would not only be a mainstream product. These names are often taken from geographical features such as rivers. Chandrasekher recalled. PCs were already in the doldrums and it risked further hurting PC demand.29 Once an Intel product is announced to the public. “It took courage to take marketing dollars and divert them to mobile. top to bottom. top management knew that the business market valued mobility and battery life. 14 Resolving Strategic Dissonance Intel’s strategy for mobile computing took further shape in the March 2002 Corporate Strategic Discussion (CSD) meeting. they are assigned code names.” Perlmutter concurred: “Doing the conversion back to mobility was considered a high risk.org. and eventually happened only in the second half of 2004 when Centrino successfully penetrated the consumer market segment (where desktop replacement was very popular). hungry for the latest and greatest technologies.” Determining the brand.il/ParksENG/company_card. Chandrasekher recalled. Lexicon came up with the brand name “Centrino” in late October 2002. that is to say. CEO Craig 29 http://www. From its MSA research. With this in mind. California. which risked leaving us with unsold PC chips. but something else was missing.php3?NewNameMade=0&from=116&CNumber=507388 . would be key outcomes of the March 2002 CSD meeting. when it most needed marketing resources. the executives at the 2002 CSD laid out the microprocessor roadmap the company would follow: (1) ramp up the Pentium 4 during 2002. To find a commercial name for the “platform” that Banias would become part of. Intel’s high-end desktop microprocessors had been purchased first by consumers. with their focus on mobility and battery life. CENTRINO® IS BORN While microprocessor designs are under development at Intel and not yet released to the public. top to bottom across all segments in the consumer space. high-end products would “waterfall” from consumer to corporate buyers.

Centrino was later shipped with a more robust chipset code-named Montera. 2006. in the face of initial opposition from the Folsom. Putting Centrino Together Intel® Centrino® mobile technology was essentially three components that. the narrower the circuitry. it would not be enough to just change the microprocessor. led the effort to develop a chipset. Ron Friedman. “We had the goal of getting 25 to 30 percent more performance from the same power envelope. 2003. the faster the processing and the more circuitry that can be crammed into the same amount of space. 15 Barrett approved this brand name in a special brand review meeting in November 2002.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Golbert recalled. We did this because we realized that. we had to optimize the CPU [microprocessor] and the chipset. which included a fully integrated graphics component built by a Folsom. the brand name was launched to the public. cache refers to memory that is available within the microprocessor.31 Mooly Eden. 31 All quotes from Ron Friedman are from the authors’ interview on March 23. . The components included: • Intel Pentium M Processor based on 90nm process technology featuring 2MB L2 cache (the microprocessor)30 • Intel 855 chipset family (the circuitry connecting the microprocessor and memory and communications elements) • Intel PRO/Wireless 2100 Network Connection (allowing mobile communications) The Centrino microprocessor and chipset design sought to optimize low power and performance to create a mobile-optimized product. we decided to produce in Israel a memory controller by revising an existing chipset made by an Intel group in the United States to reduce power consumption. In many cases up to a point. in order to extend the battery life of the device. but tweaked for mobility in order to meet the Centrino launch date demand. The ad hoc IDC chipset team was redeployed to work on Intel’s dual core microprocessor. 30 The 90-nanometer (90nm) process refers to manufacturing microprocessors with circuitry that is 90nm in width. Odem was essentially the desktop chipset. vice president and director of design recalled: As part of the Centrino effort. working together. On January 8. In this case. We dedicated around 25 engineers to the task of creating the chipset to reduce power consumption to increase overall battery life. California chipset group. Subsequent quotes will not be cited. California-based team. delivered platform capability to use computing power on the go. it became necessary to create an interim chipset that included graphics in order to meet Intel’s aggressive launch plans.” Developing an Interim Chipset Although early decisions to focus on low-power design allowed Intel to align resources. Engineering Manager in the IDC. code-named Odem.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. a new paradigm of importance had emerged. communicating increasing microprocessor speed and capability as they became available with each new processor generation.”32 Otellini also echoed this stance. He said that at first the wireless team was hesitant to abandon their branded channel strategy to solely focus on the Centrino Platform and OEMs. Centrino demanded that a new story be told. 33 Much of the software available had not been written for low power. reducing battery demands. and without the brand. 2006. Subsequent quotes will not be cited.” Branding for the Centrino technology revolved around a central question that would be asked by most customers. shipping wireless features with Centrino became a hot topic. Johnson credited Craig Barrett with providing much needed clarity to both his (Johnson’s) team and the Mobile team concerning the importance of wireless features to Centrino when he stated: “Without wireless there is no brand. Intel had been actively involved in power consumption reduction and had been working with the industry to reduce power consumption across all aspects of the PC. at the time vice president and general manager of the Wireless Networking Group within ICG. . Mobile had something different. MacDonald explained: “Who in their right mind would pay a premium for a product that was megahertz deficient? And how do you communicate this to the community—a completely new value. and performance’ for so many years?” With Centrino. Since 1996. speed. Intel asked key software vendors to continue their work to reduce power consumption and to make the mobility transition by developing applications that smartly managed power needs. Managing the Branding of Centrino® Intel’s branding strategy of the 1990s involved “Intel Inside” and the Pentium as flagship brands to establish trust and recognition with consumers and buyers. including software. 2006. Subsequent information and quotes will not be cited. there is no campaign. This realization put pressure on the wireless team to establish clear priorities to deliver on time for the platform launch and prepare to support approximately 20 OEMs simultaneously.’ Centrino forced people to make important tradeoffs. Mobilizing the Software Ecosystem One of the potential barriers to the successful launch of Centrino involved working with software vendors to create low-power-friendly software. We finally stopped trying to justify ourselves. Jim Johnson. but how could it be communicated? 32 All quotes from Jim Johnson are from the authors’ interview on February 22. Intel used the “Intel Inside” brand as a way to directly influence the end user. We centered ourselves around ‘smaller’ and ‘thin and light. 33 Information provided by Philip Wennblom is from the authors’ interview held on Tuesday. megahertz. MacDonald observed: “[New] branding was needed to help us differentiate ourselves both internally and externally. talked with the Centrino team to determine what was needed. when you have been spouting ‘power. 16 Mobilizing the Communications Group In early 2002. where in the past the group had only supported one. August 22.

We identified five key hurdles (ease of use.” But influencing the industry would not be cheap or easy. a wireless card). MPG Director of Business Development at the time. Ultimately consumer demand for Centrino resulted in significant support from all our OEMs.” some argued. awareness) and proposed specific actions to address them. followed by a more massive hot spot build-out worldwide.” Influencing the Industry In September of 2002. Many OEMs were not excited about Intel’s introduction of a new brand. Some of the OEMs wanted to focus on taking markets from competitors rather than making markets with Centrino. Internally. Barrett and Otellini met with the Banias team and approved a $300 million budget to launch the Centrino brand. access/authentication/accounting. signage. OEMs wanted flexibility to replace a piece of the Centrino technology with their own product (e.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. The idea was to look at the roadblocks for Wi-Fi adoption and try to remove them. Some were early adopters of the Centrino brand. Yoav Hochberg. “Just call it a Pentium.. 17 As part of the branding strategy. the Desktop Group wondered how the new Centrino brand would impact the Pentium brand and funding.g. the Wi-Fi service provider. identifying the people in Intel that we could utilize to negotiate the deals. others were more resistant because they had their own unique wireless solutions. These actions would usually involve three parties: the “Hot Spot” location owner. which included money for an Infrastructure Enabling Program (IEP). Intel envisioned utilizing signage similar to that used by Visa and Mastercard to announce to customers that their credit cards are accepted at certain establishments. MacDonald recalled: “Acceptance of the Banias CPU [Pentium M] was pretty much ubiquitous.000 verified hot spots by the time of the Centrino launch in order to demonstrate momentum. security. The key concept was that the location owner and service provider would build the hot spots and Intel would help in the verification and participate in the co-marketing. availability. including TV ads. which was being marketed and validated cross the world at thousands of hotspots to ensure connectivity. defining and communicating the deal process. and training over 30 people in over 20 countries on the process. 2006. etc. who was in charge of co-marketing the deliverables of the IEP program from the Sales and Marketing Group (SMG) point of view. they would give up the right to use the Centrino brand. but by doing so. The other part of the program involved articulating and forming the various deal structures. Hochberg recalled: The development of the program concept was preceded by a significant market analysis done in MPG on the state of Wi-Fi. but such computers could only be marketed as containing the Pentium M chip without the Centrino brand attached. The 34 All quotes from Yoav Hochberg are from the authors’ interviews on December 18 and 23.34 The program goal was to reach 4. A Centrino sign communicated to the consumer: “Use your Centrino technology here. but initial acceptance of the larger Centrino brand by OEMs was mixed. . OEMs could use the microprocessor without the Centrino set. and Intel. Hochberg worked with Mike Hoefflinger. was in charge of the IEP and responsible for all MPG business deals with partners. Funding was needed.

Hochberg said that the program reached its stated objectives: between November 2002 and March 1. Indeed. for example. MANAGING THE PRESSURE TO LAUNCH By mid 2002. the number increased to 30. 70 percent of notebook users were corporate employees. airports and other public or retail locations. which could have been provided by a third party source. wary of delaying the product launch for want of an unproven communication feature.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Wayport. Marriott Hotels. but the wireless component was not. As a result. and also worked with the world’s largest software security vendors such as Verisign and Checkpoint to ensure that hot spot users had the best protection available as they utilized the Centrino hot spots across the world. The industry seemed to be ready. and national chains. and to over 100. As an example. and other hotspots. Intel focused on securing ease of use and seamless connection. the $300 million Centrino brand campaign funds were also entirely separate from the Intel Inside rebates given to OEMs for their Banias (Pentium-M) processor purchases. as component roadmaps demanded convergence to create the Centrino platform. Location partners or service providers paid for hardware. cafes. 179 deals involving airports. it became clear that not all components were on track. so the initial deployment targeted business travelers. enabling broadband in hotels. Some Intel sales representatives. Hochberg recalled that by March 2004 the Giants stadium in San Francisco had become the first sports arena to provide Wi-Fi access. Intel focused on working closely with airports. 2003. Intel worked closely with these service providers and their customers to establish standards so Centrino users would experience smooth connectivity at all Centrino-branded hot spots. Banias was ready.000 by the end of 2003. Manufacturing capacity was committed for a launch in January 2003. Building on the momentum. Furthermore. and national chains were closed. bookstores. quickly saw how connectivity would give them a competitive advantage in offering guests a new amenity: wireless connectivity. Otellini received several e-mails from powerful customers suggesting that he launch as planned in January without the communications component. This extended the mobility value proposition while allowing these partners to enjoy a new and lucrative revenue stream. hotel lobbies. The chipset was ready. .000 by the end of 2004. 18 program initially focused on geographies with significant numbers of notebook users.000 hot spots verified for Banias. hotel chains. and ready for the Centrino launch. and others to build out Wi-Fi infrastructure and provide wireless services to consumers in various locations such as hotel rooms. Pressure to launch Centrino mounted. invited their biggest customers to directly contact Otellini suggesting that Intel launch Centrino without the communication feature. Many wanted to ship the technology without the wireless component. resulting in slightly over 4. In October of 2002. However. hotel chains. Beyond the $300 million budget (this was a budget size typical for brand launches at Intel) the Centrino war chest included additional funds invested by Intel Capital in companies such as STSN. airports. Intel announced it would devote $150 million of the $500 million Intel Communications fund specifically for investments in Wi-Fi companies worldwide. by November 2002. At the time. it became clear that the Wi-Fi communications component would not be ready until March 2003 at the earliest.

. This had been a painful decision to make because we had been successful in the 1990s in the Ethernet market by supporting both 10 and 100 Mbps standards.11a Wi-Fi features.11b Wi-Fi features for the target market launch date. making lots of tweaks and changes as they go.” Chandrasekher remembered a particularly poignant meeting where Jeff Hoogenboom. Eden’s responsibility was to coordinate across the Centrino platform to ensure that all the components would be delivered and the manufacturing ramp achieved.” The wireless team succeeded in delivering the more basic radio and the 802. Because of the almost black magic nature of RF. amongst other things. Hence. 2006. Otellini made Eden responsible for delivering the first one million units of Centrino and gave him a “gold badge. Whereas. the primary sales representative to one of Intel’s largest customers was asked for his opinion: “Jeff said. Those two cultures. Sean Maloney. Ultimately. wherever they were located.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.” which gave him the power to mobilize all the resources necessary. recalled: “We quickly found out we weren’t ready yet on the 5-gigaherz radio and 802.S. that strategic planning and marketing managers define the platform and the general manager of the business ratifies it. recalled that Eden was very good at the platform manager job because he was very experienced in engineering management and could confront people without offending them. Maloney also pointed out that Intel’s wireless team was new to the wireless business: There was considerable difference between the way that you execute a project around a microprocessor and the way you do it around radio subsystems. Otellini had to make the decision and he held his ground: “You offer all of it or you do none of it!” In August 2002. Eden pointed out that the platform manager job was execution only. “There was enormous dissension in the organization when the decision was made to hold the microprocessor launch back to wait for the wireless component. the radio guys tended to spin the design multiple times till it works.11a protocol was productized a year later.” Adding to the pressure on Intel was the fact that the decision to delay the launch by a quarter would shorten the life of the product line and would likely cost the company upwards of hundreds of millions of dollars. 19 Anand Chandrasekher recalled. Jim Johnson of ICG. so we had to de- feature our product to 2. had significant influence on the team and the opinion of many in the room. who had hoped to leapfrog the competition.4 gigahertz and 802. (The 802.35 Others pointed out that Eden would go talk to the engineers down in the organization in order to get the “straight scoop. to take on the job.’ Jeff’s leadership in that situation.) 35 All quotes from Sean Maloney are from the authors’ interviews on February 21.11b in order to keep up with our time-to-market priority. Otellini decided to create the new position of “platform manager” and brought Mooly Eden from Israel to the U. which cost him commission income. ‘I know my customer wants us to launch without Wi-Fi.” and Eden himself pointed out that he had to use his gold badge only once. but I think it is important to the brand that we wait until the Wi-Fi is ready. the design methodology of microprocessors is much more monolithic: get it right and get it out of the door on a certain date. executive vice president and GM of the Intel Communications Group (ICG) at the time. created a clash.

Intel could not accurately predict demand for Centrino. Jim Johnson recalled: “First. it required growing new competencies. Intel believed that Centrino was largely responsible for more than doubling the compounded annual growth rate (CAGR) of notebook sales after 2002. MPG faced the new challenge of working with OEMs in a way that greatly differed from Intel’s experience working with traditional Intel channel partners. the phrase “wafer starts” is synonymous with product mix because wafers contain only one type of microprocessor. chipset and wireless controllers) together to speed their time to market. OEMs are much more educated on features of Intel versus the competition. . we had to change the team’s focus from branded channel partners to OEMs. Because the product was new. provides nevertheless some insight into the rapid growth of the success of the “platform” approach to mobile computing. In order to make sure Intel had sufficient capacity. on the first generation. By 2006. Ultimately.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. one measure of the market success of Centrino (set in the March 2002 CSD meeting) was the attach rate: the percentage of laptops sold worldwide that contained the entire platform (mobile microprocessor plus the Wi-Fi capability). Intel thought that the premium pricing it was able to achieve with Centrino. Channel players are less so. whereas Channel players don’t require this. we had to change some of our design managers from personnel with channel experience to people with OEM experience who could bring the skill set needed to work in the OEM environment. on March 13. increasing from a CAGR of around 10 percent between 2001 and 2002 to 22 percent in 2003 and beyond. 2003. OEMs expect more rigorous validation for all three components (CPU. These requirements pushed back our launch. OEMs want to see a functional sample. our wireless components needed to meet the same expected validated quality as our CPU and chipsets. Centrino managers wanted to keep enough supply on hand to act as a buffer in case demand exceeded expectations. Intel would have to reduce the number of wafer starts36 dedicated to Pentium 4 microprocessors. Questions of manufacturing capacity and product mix struck at the heart of Intel’s strategy and hence were made by top management. Paul Otellini supported the Centrino team’s request for a wafer buffer and gave Centrino its additional wafer starts.” Finally. 36 Microprocessors are made in batches on large wafers of silicon (which are cut into individual microprocessors toward the end of the manufacturing process). Centrino quickly repaid the confidence its supporters placed in the platform. Basically. Not surprisingly. where Intel’s microprocessor businesses had established strong relationships and process methods. and only one type of wafer can be run through a manufacturing plant at a time. added $2 billion in revenue in the two years following Centrino’s introduction (Exhibit 9). In fact. Exhibit 8. when average selling prices for other chips made by Intel and its competitor were declining. while not Centrino-specific. Centrino was launched. Further. Hence. 20 In addition. To create this buffer. THE SUCCESS OF CENTRINO® Market Success A real test of the confidence that Intel had in Centrino came in late 2003.

which would make scaling it to dual core and beyond relatively easy. The relative merits of each were discussed at the 2003 SLRP meeting. “Two of the biggest advocates of Merom were (not surprisingly) Mooly Eden but also Sunil Shenoy. Shenoy came from Desktop and might have been expected to advocate for Tejas. Intel had dropped Tejas in favor of Merom. Top management wanted to make sure the key 37 Intel Centrino computer components marketing spend: 2005 ($44. Perlmutter pointed out that being a much smaller and lower power design than Tejas. 39 It could be used in mobile PCs.051. recalled. Intel spent a little over $44 million. but he was a big supporter of Merom.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Merom could be the basis for dual core microprocessors. desktop PCs and servers. however. Anand Chandrasekher. in 2004 Intel spent a little over $77 million marketing the Centrino brand.2M). Intel would align its architectures for mobile and desktop on Merom. In particular. 38 Merom is the name of a lake in Israel and Tejas is the Spanish name for the state of Texas.7%.2M). The launch of Centrino matched well with industry convergence toward broadband availability and mobility consumer demand.178. Interestingly. Pentium 4 for desktop computers and Itanium for servers. 2004 ($77. percent change: 42.37 Technical Success While Intel was dealing with capacity allocation questions. Within a year of this SLRP meeting. He recalled that the dual core idea came about the moment the Banias team realized they had a small and power-efficient core. “America’s Top 2000 Brands. . a reduction of nearly 43 percent. delivering performance and energy efficiency. At the time of the 2003 SLRP. Otellini led this SLRP meeting.38 Merom featured a converged core and followed the power-efficiency trajectory established by Banias. 2006. This could have caused some anxiety among the Desktop Group that might have supported Tejas. Intel produced three different microprocessor architectures: Centrino for mobile. which allows optimization of architectures and technologies adapted to usage demands. Tejas was developed with a focus on processing speed. This had been Intel’s traditional focus for microprocessors and the company was much further along in the development of Tejas than Merom. 21 This success meant that Intel could reduce the amount of money it spent on marketing Centrino over time. it was wrestling with finding the best avenues for growth. The company was developing two new architectures that promised to provide good future growth opportunities: Merom (from the IDC) and Tejas (developed in the United States).” Brandweek. Going forward. This was the topic of the 2003 SLRP meeting held in May of that year. Intel planned to continue expanding its platform strategy into. According to one brand survey. Source. 39 The computational element of the Intel Core microarchitecture is described as a converged core. the enterprise and digital home market segments. June 19.” CONCLUSION Intel Centrino technology was the company’s first attempt at developing a platform solution. it was also looking to the longer-term future. Going forward. among others. In the next year.

organizational and managerial competencies garnered from its Centrino platform experience would be used in future platform developments.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 22 strategic. .

2000 Executive Office ICG EPG DPG MPG SMG TMG EPG Enterprise Platforms Group DPG Desktop Platforms Group IABC Intel Architecture Business Group ICG Intel Communication Group MPG Mobile Platform Group P&L Profit and Loss PMD Performance Microprocessor Division SMG Sales and Marketing Group TMG Technology Manufacturing Group Source: Authors.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. .1998 Executive Office IABG SMG ICG MPG TMG (P&L) PMD BMD server desktop mobile Intel Organizational Chart -. 23 Exhibit 1 Intel Organizational Structure in 1998 and 2000 Intel Organizational Chart -.

000 Rocket Power Density Nuclear 100 (W/cm2) 8086 10 4004 Hot Pentium® 8008 8085 386 processors 8080 286 486 1 ’70 ’80 ’90 ’00 ’10 Year Source: Pat Gelsinger. 10.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. given Intel’s current trajectory. .000 Sun’s Surface 1. 24 Exhibit 2 Hotter Than the Sun This graphic (used by Pat Gelsinger in 2001) theoretically illustrates future chip temperatures of microprocessor heat dissipation over time.

net 987 578 573 Income before taxes 15.331 19.561 8. general and administrative 5.089 3.539 Cost of sales 15.650 11.586 411 56 Purchased in-process research and development 109 392 165 Operating costs and expenses 23.897 3.946 2. net 565 289 192 194 393 Income before taxes 12.314 $ 6. (In millions--except per share amounts) 2000 1999 1998 Net revenues $ 33.360 4.068 Source: Company reports.256 Losses on equity securities.464 Impairment of goodwill -.533 4.204 2.796 Operating income 12.069 Net income $ 10.379 Gains on investments.726 $ 29.959 9.836 12.088 Research and development 3.047 13.094 13.936 10.487 Gross margin 23.278 4.767 8.764 $ 26.273 Cost of sales 12.141 11.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.894 Operating income 10.130 7.034 3.291 Years Ended December 30.087 892 Net income $ 8.777 14.052 Research and development 5.141 $ 26.183 Provision for taxes 3.901 1. net 3.606 3.442 4.509 Marketing. -.801 1.759 883 185 Interest and other.872 3.446 13.417 7.664 $ 7.334 4.117 $ 1.209 $ 30. 2005 (In Millions--Except Per Share Amounts) 2005 2004 2003 2002 2001 Net revenue $ 38.914 3.516 $ 5.145 4.641 $ 3.535 $ 7.463 13.688 4. 5 20 198 Operating expenses 10.610 10. 98 Amortization of goodwill 1.612 Amortization and impairment of acquisition-related intangibles and costs 126 179 301 548 628 Purchased in-process research and development -.049 19.395 9.622 17. .228 9. general and administrative 5. -.382 2. 617 -.659 4.746 17.090 10.616 9.111 2.389 $ 26. 25 Exhibit 3 Intel Selected Financial Data 1998-2005 Years Ended December 31.778 4.796 Marketing.076 Amortization of goodwill and other acquisition-related intangibles and costs 1.826 $ 34.137 Provision for taxes 4.318 13. net -45 (2) (283) (372) (466) Interest and other.

017 Mobility Group 5. motherboard and other revenue 2.130 7.137 24.141 Operating income (loss) Digital Enterprise Group 9.833 1.533 Source: Company reports.068 Chipset.778 23.405) (2.667 4.352 5.086 Flash Memory Group 2.006 8.330 2.090 10.826 34.725 5.981 5.704 5.427 1.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.743 Flash Memory Group (154) (149) (152) All other (2.412 19.120 Chipset.092) (1. .991 Microprocessor revenue 5.075) Total operating income 12.426 17.314 966 11. 26 Exhibit 3 (continued) Selected Financial Data by Business Group 2003-2005 (In Millions) 2005 2004 2003 Net Revenue Digital Enterprise Group 19.851 8.209 30.285 1.059 Mobility Group Microprocessor revenue 8.131 6. motherboard and other revenue 25.278 2.608 All other 280 165 388 Total net revenue 38.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. .) Worldwide PC Shipments 300 300 (M) 250 208 200 150 121 100 50 0 1988 1989 1991 1993 1994 1996 1997 1999 2001 2003 2004 2006 2007 2008 1986 1987 1990 1992 1995 1998 2000 2002 2005 2009 1985 1981 1982 1983 1984 Forecast Source: Gartner. 27 Exhibit 4 Global PC Volumes by Unit 1981-2009 (est.

) 70% United States 60% Western Europe 50% Japan 40% 30% Asia/Pacific 20% 10% 0% 2002 2004 2006 2008 2010 Source: Gartner.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 28 Exhibit 5 Notebook PCs as a Percentage of PC Market by Unit by Region 2002-2010 (est. .

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Frequency Low Smaller Die Area Larger Source: David Perlmutter. Intel consciously made the decisions to give up frequency (and potentially CPU performance). . Area High Freq. 29 Exhibit 6 Relationship of Frequency to Die Size The graphic below diagrams − in concept − how small reductions in frequency significantly reduce needed die area in chip design. trading speed for more room on the die to add new functionality. and so the trade-off was worth the risk. Impact on performance turned out to be minimal.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Value in the mobility space consisted in a platform that addressed four key areas important to mobility users. secure Longer battery life wireless connectivity Intel Secret Source: David Perlmutter. Beth Eby. 30 Exhibit 7 Four Vectors of Mobility This slide was used early on by the Mobility team to articulate how “value” to the consumer meant more than just speed. lighter notebooks (> 70% in MSA) • Wireless connectivity 11 Best 22 Thinner and Performance lighter In Form Factor 33 44 Simple. . Adi Golbert. Alex Peleg. 8 Notebook Market Trends (driven by Mobile MSA) 4 Vectors of Mobility • Performance remains top purchasing criterion • Battery life is #1 improvement requested for mobile PCs • Most users prefer thinner.

Used with permission.7 54. These devices can be an integrated circuit fitted onto the motherboard itself and/or an expansion card that fits in sockets.26 1.3 42.7 37. 2003 .9 52. 2000 – 2006 2000 2001 2002 2003 2004 2005 2006 Laptop computers (millions) 26. Used with permission.3 70.6 66. --- Others 19% 6% --.9 61.9 Source: ABI Research.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Table 3-27. --. The PCI specifies a computer bus for attaching peripheral devices to a computer motherboard. . --- Texas Instruments 11% --. Wi-Fi Forecast Database: IC.2 28. Source: ABI Research.3 32. 31 Exhibit 8 Market Share and Total Revenue Data for Wi-Fi Sales Worldwide Wi-Fi Penetration in Laptop Computer. Wi-Fi Mini PCI (for laptop computers) IC Vendor Market Share.14 19.13 6.2006 2003 2004 2005 2006 Atheros 18% 17% 11% 16% Broadcom 30% 30% 17% 20% Intel 11% 47% 72% 64% Philips 11% --.4 Wi-Fi penetration rate 1% 4% 19% 46% 71% 89% 95% Laptops with Wi-Fi (millions) 0. --- Total 100% 100% 100% 100% Note: PCI stands for Peripheral Component Interconnect. --.

MSS stand for market segment share (or size).Intel Centrino in 2007: A New “Platform” Strategy for Growth p. SOW stands for share of wallet. ASP stands for average selling price.9% March 2003 Launch March 2003 2002 2003 2004 2005 2001 2002 2003 2004 FORECAST Note: TAM stands for total addressable market. Source: Intel internal reports. . SOW data from Gartner. 32 Exhibit 9 Success of Centrino Platform = (K TAM + K MSS + K ASP + K SOW) Success Mobile TAM Mobile MSS Growth Rate = 22% +6% GROWTH Growth RateRATE = 10% 10% ™ Centrino Centrino Launch TM Launch March 2003 March 2003 2001 2002 2003 2004 2005 FORECAST FORECAST Q1’03 Q1’05 Mobile ASP Mobile SOW* Intel Mobile Revenue as % of Notebook System Revenue Launch +4.

33 Appendix Key Events Fall 1997 David (Dadi) Perlmutter. Brand name made public on January 8. October 2002 Centrino brand name created in late October and ratified by CEO in a brand review on November 21. March 2002 Corporate Strategic Decision (CSD) confirms the targeted four performance vectors.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Banias is put on the Intel roadmap. May 2003 SLRP Meeting led by Paul Otellini. Chandrasekher and Perlmutter conclude that Centrino is not ready to go to market as defined. Otellini delays the launch of Centrino until the Intel-produced wireless component of the platform is ready. Israel Development Center (IDC). Sets target attach rate of greater than 50 percent. Source: Authors. Paul Otellini protects Banias from efforts by other senior executives to redeploy the team. The “Jazz the PC” SLRP identifies notebook computers as a growth opportunity for Intel. November 2002 In their November Operations Review meeting. May 2001 Intel Strategic Long Range Planning Process (SLRP) meeting discusses two topics: “Jazz the PC” and “Power Wall”. Banias-based Merom is chosen over Tejas as Intel’s microprocessor architecture going forward. Sept-Oct 2001 Product Line Business Plan (PLBP) meeting. Banias in corporate space top-to-bottom and (reverse) waterfall it to consumers in the following year. irrevocable combination of CPU + chipset + Wi-Fi. 2003. March 2003 Centrino is launched with all three components. During an urgent follow-up strategy meeting with top management. April 2000 Intel reorganization combines the Microprocessor Group (MPG) and Intel Architecture Business Group (IABG) into the Intel Architecture Group (IAG). 2001 PC market declines in absolute units for the first time ever. learns from an Intel IDC engineer about a potentially valuable tradeoff: for a 10 percent reduction in microprocessor frequency. then vice president and general manager. more components can be placed within the Pentium III die size. .

34 Appendix (continued) Centrino Key Decision Points This illustration shows the key decision points leading to successful launch of Centrino and Intel’s transition to a single core strategy using architectures founded in early designs (e. Intel Confidential For use only by Dr. Banias)..Intel Centrino in 2007: A New “Platform” Strategy for Growth p.g. . and appropriate Internal Intel Employees Source: Authors. Robert Burgelman. Timna. Philip Meza.