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.

8 with each of the three groups having responsibility for profit and loss (i. Provide an outside perspective about resource balancing. February 9.11. In 2000. running the business) and R&D. unless otherwise indicated. . “Paul Otellini and others started to believe that the megahertz and gigahertz orientation (i. Paul Otellini. leaving Otellini as sole head of IAG. Otellini said. 2005. These studies confirmed that the market was more interested in what a product could do rather than discrete measures of performance. it’s not sufficient to drive the levels of growth and innovation that will allow our industry to prosper. 3. Help determine what the market really cared about. and Mobile Platforms Group (MPG).” Albert Yu and Paul Otellini co-managed IAG for several months before Yu transitioned to lead Intel’s optoelectronics group. 9 This paragraph is informed by David Perlmutter and Shlomo Maital.”11 7 Intel sometimes assigns two executives to co-lead a group.9 In an effort to aid the transition to this new structure. “Intel Inside Out: The Chip Industry Leader Adapts to Changing Consumer Demands. 4 Otellini.10 It turned out that the third task was to be particularly consequential. for example.. 2. Otellini alone led IAG. Chandrasekher described the purpose of the centralized planning and marketing group he was assigned to run: 1. 11 Chris Nuttall. we conducted a market segment analysis (MSA) to further explore the market needs.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.. Centralize marketing so general managers within IAG would not be distracted by marketing responsibilities such as product ramps. (After Yu transitioned to lead a different group. Otellini temporarily split out marketing and planning from IAG and centralized these functions under the direction of Intel executive Anand Chandrasekher. 2006. 8 Note: after the 2000 reorganization. p. gigahertz. Technology and Manufacturing (TMG) and Sales and Marketing (SMG) remained separate functional groups (Exhibit 1).) The combination that created IAG affected some 11. Chandrasekher explained. 10 All quotes from Anand Chandrasekher are from the authors’ interview on November 10. Speaking at the 2001 Intel Developer Forum.e. a relationship it calls “two-in-a-box. Desktop Platforms Group (DPG). As early as 2001.e. or the speed at which the processor executes instructions) was running out of steam.7 IAG would have both P&L responsibility and responsibility for R&D. then executive vice president of Intel Architecture Group. clock speed.” 2003.” 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.” Financial Times.000 Intel employees and brought together three separate business groups: Enterprise Platforms Group (EPG). “Israel Inside: The Evolution of Centrino/Banias/Mobility. “While this focus on raw processor speed [clock speed] is important. described the need for Intel to expand its focus beyond clock speed itself and take a more holistic approach. This led Paul to consider how we could best optimize our organization to take advantage of the market. MPG refers to Mobile Platforms Group and not Microprocessor Group.

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

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

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

Yet while Timna successfully met its primary goals. which meant reducing die size.000 at the time). it kicked off a critical discussion about alternative value propositions beyond 19 Email correspondence between Donald W. unless otherwise cited. This line of thinking was not only innovative for the successful design of the Timna chip. lower-power-consuming processor than the Pentium 4. yet it yielded several critical discoveries and drove important strategic discussions that would be crucial for Intel’s mobile strategy going forward. Timna relied on a memory technology called RDRAM from memory company Rambus. an unorthodox solution began to emerge.’”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. then vice president and general manager. Perlmutter recalled: “There was a group of people congregating near the bulletin board. ‘I have a solution for you for die reduction. related a corridor conversation from the fall of 1997 with an Israeli engineer. (Note: die size refers to the area used on a silicon wafer to make a chip. David (Dadi) Perlmutter. I can put everything you want [including the graphics and memory controller] into the Pentium III die size. more dies could be put on a wafer for roughly the same manufacturing cost per wafer. 8 was scheduled to hit the factory. who thought he could build a cheaper. it did not offer sufficient cost reductions compared with the Pentium III. IDC engineers had learned that by giving up a little bit of frequency.) Timna used the Pentium III core architecture while integrating other platform components such as graphics and memory controllers into the microprocessor. 20 All quotes from David (Dadi) Perlmutter are from the authors’ interview on February 9. Timna’s original goal had been to reduce cost. March 20. 2007. but you’ll have to pay a price for it. but you’ll have to let me lower the frequency by 10 percent. which Intel had been able to make more cheaply than it initially had planned. Subsequent quotes from this interview will not be cited. These setbacks left the IDC scrambling for its next project and the group feeling even more marginalized. Simcha Guchman. so ultimately it was not the best solution for the value PC market. While working along these lines. . Unknown to the IDC leadership at the time. but perhaps more importantly. Israel Development Center. which was pivotal not only for the Timna project. Nelson and authors. 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. 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. Economies were also realized by adding more components to a single chip.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. while maintaining the same die size. but ultimately had impact on Intel’s strategic direction for the mobile segment. 2006. they could free up a lot of die space (Exhibit 6). RDRAM cost more than twice as much as alternative technologies. Additionally. A die that is completed and cut from the wafer becomes a chip. One of the engineers said.19 Timna was never manufactured for sale. By decreasing the size of a die. Timna was targeted at the “Value PC” category (under $1.

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

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

allowed easy connectivity and did not require lots of infrastructure to deploy. and masks are created for the wafer fab) in mid-1999. But Otellini responded. Intel was clearly not a leader in Wi-Fi technology. The industry at that time was selling clunky. Banias products proved worthy. the Desktop Group. based on the decision to use an Intel Wi-Fi component. demonstrating strong performance and low power—a perfect mix for mobility. At tape-out (the final step in the design of a microprocessor where the product design is frozen in preparation for manufacturing. Intel’s then current desktop chip. First-time buyers were less savvy about what is valuable in a laptop. but we knew that people would eventually want standard features in thinner form factors. which required plenty of space. Some did not believe that Banias would deliver the performance the IDC promised. to supporting a dual mode (802. however. Using Wi-Fi was the basis for the four vectors of mobility discussion that Alex and Adi led earlier in 2001. We will use Intel wireless!” This entailed some risk because Intel’s competency in this technology was untested. believed that the frequency levels were too low to use with desktop computers. as the proposed alternative was to use Bluetooth technology to connect to cell phones.25 inches thick because they used standard desktop chips. “The Desktop Group said. Intel. “You don’t get it. most laptops in 2000 were over 1.” HEADQUARTERS TAKES A NEW LOOK AT MOBILE Conflict between Desktop and Mobile As Banias was emerging. however. The Banias marketing team also learned a valuable lesson by studying second. Adi Golbert recalled.11a and 802. used free spectrum.11b was more popular at the time and they were concerned that switching to a new protocol would weaken acceptance. as 802. In terms of form factor. MacDonald pointed out: “You couldn’t fit a Pentium 4 in what we now consider a mini-notebook. 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. ICG changed their Wi-Fi plans from supporting the 802. MPG learned that targeting experienced users for research was key to understanding the space.11b). had obtained Wi-Fi expertise in an earlier acquisition and had a group working with this technology in the Intel Communications Group (ICG). the Banias team engaged in a joint program with ICG. ‘with these frequency figures.” Wi-Fi was particularly attractive because it was an IEEE industry standard. That was after a big debate. we . After reviewing the Banias data. thinking Pentium 4 would fill the mobility gap as needed. it competed with the Pentium 4 development team for attention.11a protocol only. the Pentium 4 would not fit in thinner form factors. 11 by the fourth quarter of 2001: “We had analyzed multiple wireless technologies and closed on Wi-Fi in late 2001. third and fourth- time laptop buyers. Perlmutter said that.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. large notebooks. but experienced buyers better understood laptop feature set priorities. This is going to be Intel’s branded system. Initially the Banias team planned to use third-party elements for the wireless component.

Some Intel senior executives in the room wanted to move 175 Banias engineers to Desktop Group projects. After that reorganization. Now. that’s fine. Once approved by Otellini and CEO Craig Barrett. 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 In September 2001. 26 All quotes from Adi Golbert are from the authors’ interview on March 23. recalled: “At the time that Banias first appeared on the Intel roadmap for mobile products following the 2001 PLBP in September and October. Subsequent quotes from this interview will not be cited. protecting the team. however. MPG referred to Intel’s Microprocessor Group. In this meeting Otellini and others again stood strong for Banias. .” 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. Thus Banias was designed with the mobile PC in mind and was given funding and marketing support to move forward on its own. 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].28 The mobility team would no longer feel compelled to draft off the Desktop product roadmaps. Subsequent quotes from this interview will not be cited. Banias was fully funded and kept its headcount. 2006. ‘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.’”26 Banias became a target for criticism in many Intel forums early on. We grappled with questions like. its position on the Intel roadmap was seemingly secure and growing. Banias became the flagship product for the Mobile Processor Group (MPG). you cannot sell low megahertz products. 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. there still was enormous debate about whether Banias would succeed. Strategic Dissonance Revealed Anand Chandrasekher. 2006. Peleg outlined some of the early criticisms of Banias: “No one will want it. But we are selling gigahertz and you guys don’t have it. 12 can’t sell it. the Mobility team presented their PLBP to Intel’s top executives. 28 Until its reorganization in April 2000. we should kill Banias. PLBP meetings are critical at Intel because they determine funding and headcount resources for businesses. If you want to use it for mobile products. mobile platform and architecture would warrant dedicated resources. Intel’s Mobile Processor Group went by the MPG acronym. Otellini supported Banias and the project moved forward.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 27 All quotes from Alex Peleg are from the authors’ interview on February 22. would this hurt the company’s crown jewels?” Still.

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

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

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

We centered ourselves around ‘smaller’ and ‘thin and light. 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. Jim Johnson. but how could it be communicated? 32 All quotes from Jim Johnson are from the authors’ interview on February 22. reducing battery demands. megahertz. 16 Mobilizing the Communications Group In early 2002. 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. 33 Information provided by Philip Wennblom is from the authors’ interview held on Tuesday. 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. Centrino demanded that a new story be told.’ Centrino forced people to make important tradeoffs. 33 Much of the software available had not been written for low power. a new paradigm of importance had emerged. where in the past the group had only supported one.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. 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.” 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. Subsequent quotes will not be cited. 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. 2006. Mobile had something different. Subsequent information and quotes will not be cited. We finally stopped trying to justify ourselves. at the time vice president and general manager of the Wireless Networking Group within ICG. talked with the Centrino team to determine what was needed. August 22. 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. 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. MacDonald observed: “[New] branding was needed to help us differentiate ourselves both internally and externally.”32 Otellini also echoed this stance. and performance’ for so many years?” With Centrino. speed. communicating increasing microprocessor speed and capability as they became available with each new processor generation. Since 1996. there is no campaign. . including software. and without the brand. 2006. Intel used the “Intel Inside” brand as a way to directly influence the end user. 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. when you have been spouting ‘power.

The other part of the program involved articulating and forming the various deal structures. Funding was needed. Barrett and Otellini met with the Banias team and approved a $300 million budget to launch the Centrino brand. OEMs wanted flexibility to replace a piece of the Centrino technology with their own product (e. Hochberg worked with Mike Hoefflinger. but such computers could only be marketed as containing the Pentium M chip without the Centrino brand attached. Hochberg recalled: The development of the program concept was preceded by a significant market analysis done in MPG on the state of Wi-Fi. 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.. Yoav Hochberg.000 verified hot spots by the time of the Centrino launch in order to demonstrate momentum. but initial acceptance of the larger Centrino brand by OEMs was mixed. including TV ads. awareness) and proposed specific actions to address them. Internally. was in charge of the IEP and responsible for all MPG business deals with partners. the Wi-Fi service provider. but by doing so. signage. 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. Many OEMs were not excited about Intel’s introduction of a new brand.” Influencing the Industry In September of 2002. A Centrino sign communicated to the consumer: “Use your Centrino technology here. 17 As part of the branding strategy. These actions would usually involve three parties: the “Hot Spot” location owner. and Intel. which was being marketed and validated cross the world at thousands of hotspots to ensure connectivity. Ultimately consumer demand for Centrino resulted in significant support from all our OEMs. . followed by a more massive hot spot build-out worldwide.g. Some were early adopters of the Centrino brand. 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. MPG Director of Business Development at the time.34 The program goal was to reach 4. who was in charge of co-marketing the deliverables of the IEP program from the Sales and Marketing Group (SMG) point of view. The 34 All quotes from Yoav Hochberg are from the authors’ interviews on December 18 and 23. and training over 30 people in over 20 countries on the process. availability. access/authentication/accounting. security. a wireless card). 2006. defining and communicating the deal process. the Desktop Group wondered how the new Centrino brand would impact the Pentium brand and funding. “Just call it a Pentium. MacDonald recalled: “Acceptance of the Banias CPU [Pentium M] was pretty much ubiquitous.” But influencing the industry would not be cheap or easy. they would give up the right to use the Centrino brand.” some argued. etc.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. which included money for an Infrastructure Enabling Program (IEP). OEMs could use the microprocessor without the Centrino set. The idea was to look at the roadblocks for Wi-Fi adoption and try to remove them. We identified five key hurdles (ease of use. Some of the OEMs wanted to focus on taking markets from competitors rather than making markets with Centrino.

Intel announced it would devote $150 million of the $500 million Intel Communications fund specifically for investments in Wi-Fi companies worldwide. 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. Indeed. The industry seemed to be ready. bookstores. As an example. wary of delaying the product launch for want of an unproven communication feature. Intel focused on working closely with airports. This extended the mobility value proposition while allowing these partners to enjoy a new and lucrative revenue stream. invited their biggest customers to directly contact Otellini suggesting that Intel launch Centrino without the communication feature. it became clear that the Wi-Fi communications component would not be ready until March 2003 at the earliest.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Otellini received several e-mails from powerful customers suggesting that he launch as planned in January without the communications component. Intel focused on securing ease of use and seamless connection. . Many wanted to ship the technology without the wireless component. and others to build out Wi-Fi infrastructure and provide wireless services to consumers in various locations such as hotel rooms. As a result. Some Intel sales representatives. Building on the momentum. MANAGING THE PRESSURE TO LAUNCH By mid 2002. However. and other hotspots. by November 2002.000 hot spots verified for Banias. as component roadmaps demanded convergence to create the Centrino platform. Wayport. Marriott Hotels. Hochberg recalled that by March 2004 the Giants stadium in San Francisco had become the first sports arena to provide Wi-Fi access. 2003. 18 program initially focused on geographies with significant numbers of notebook users. enabling broadband in hotels. 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. resulting in slightly over 4. but the wireless component was not. The chipset was ready. the number increased to 30. 70 percent of notebook users were corporate employees. so the initial deployment targeted business travelers. hotel lobbies. cafes. hotel chains. Location partners or service providers paid for hardware. 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. for example.000 by the end of 2003.000 by the end of 2004. airports. and to over 100. Banias was ready. In October of 2002. At the time. Manufacturing capacity was committed for a launch in January 2003. 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. and national chains. 179 deals involving airports. quickly saw how connectivity would give them a competitive advantage in offering guests a new amenity: wireless connectivity. hotel chains. Pressure to launch Centrino mounted. and national chains were closed. it became clear that not all components were on track. which could have been provided by a third party source. Furthermore. Hochberg said that the program reached its stated objectives: between November 2002 and March 1. airports and other public or retail locations. and ready for the Centrino launch.

” which gave him the power to mobilize all the resources necessary. 2006. 19 Anand Chandrasekher recalled. had significant influence on the team and the opinion of many in the room.S.11a Wi-Fi features.) 35 All quotes from Sean Maloney are from the authors’ interviews on February 21. that strategic planning and marketing managers define the platform and the general manager of the business ratifies it. so we had to de- feature our product to 2.” The wireless team succeeded in delivering the more basic radio and the 802.’ Jeff’s leadership in that situation. who had hoped to leapfrog the competition. 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. Because of the almost black magic nature of RF.11a protocol was productized a year later. 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. Sean Maloney. executive vice president and GM of the Intel Communications Group (ICG) at the time.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. created a clash. but I think it is important to the brand that we wait until the Wi-Fi is ready. Whereas. 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. Eden’s responsibility was to coordinate across the Centrino platform to ensure that all the components would be delivered and the manufacturing ramp achieved. wherever they were located. Otellini decided to create the new position of “platform manager” and brought Mooly Eden from Israel to the U. making lots of tweaks and changes as they go.35 Others pointed out that Eden would go talk to the engineers down in the organization in order to get the “straight scoop.11b in order to keep up with our time-to-market priority. Ultimately. which cost him commission income.4 gigahertz and 802. amongst other things. (The 802. Eden pointed out that the platform manager job was execution only. recalled: “We quickly found out we weren’t ready yet on the 5-gigaherz radio and 802. to take on the job. Jim Johnson of ICG. Otellini made Eden responsible for delivering the first one million units of Centrino and gave him a “gold badge. Those two cultures. 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. the design methodology of microprocessors is much more monolithic: get it right and get it out of the door on a certain date.” and Eden himself pointed out that he had to use his gold badge only once. Hence. the radio guys tended to spin the design multiple times till it works. .11b Wi-Fi features for the target market launch date. ‘I know my customer wants us to launch without Wi-Fi.” 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. “There was enormous dissension in the organization when the decision was made to hold the microprocessor launch back to wait for the wireless component.” Chandrasekher remembered a particularly poignant meeting where Jeff Hoogenboom. the primary sales representative to one of Intel’s largest customers was asked for his opinion: “Jeff said.

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

Going forward. among others. In the next year. This was the topic of the 2003 SLRP meeting held in May of that year. This could have caused some anxiety among the Desktop Group that might have supported Tejas. 2006. but he was a big supporter of Merom. 38 Merom is the name of a lake in Israel and Tejas is the Spanish name for the state of Texas. According to one brand survey.37 Technical Success While Intel was dealing with capacity allocation questions.7%. Intel planned to continue expanding its platform strategy into. Shenoy came from Desktop and might have been expected to advocate for Tejas.38 Merom featured a converged core and followed the power-efficiency trajectory established by Banias. which would make scaling it to dual core and beyond relatively easy. Intel spent a little over $44 million. desktop PCs and servers. Top management wanted to make sure the key 37 Intel Centrino computer components marketing spend: 2005 ($44.2M). At the time of the 2003 SLRP. Otellini led this SLRP meeting. June 19. In particular. 39 The computational element of the Intel Core microarchitecture is described as a converged core. in 2004 Intel spent a little over $77 million marketing the Centrino brand. The relative merits of each were discussed at the 2003 SLRP meeting. which allows optimization of architectures and technologies adapted to usage demands. Intel had dropped Tejas in favor of Merom. 21 This success meant that Intel could reduce the amount of money it spent on marketing Centrino over time. the enterprise and digital home market segments. percent change: 42. “America’s Top 2000 Brands. Merom could be the basis for dual core microprocessors. delivering performance and energy efficiency. Tejas was developed with a focus on processing speed. it was also looking to the longer-term future. Anand Chandrasekher.178. Pentium 4 for desktop computers and Itanium for servers.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. recalled. Going forward. it was wrestling with finding the best avenues for growth.051. . He recalled that the dual core idea came about the moment the Banias team realized they had a small and power-efficient core. a reduction of nearly 43 percent. Within a year of this SLRP meeting. Perlmutter pointed out that being a much smaller and lower power design than Tejas. The launch of Centrino matched well with industry convergence toward broadband availability and mobility consumer demand.” CONCLUSION Intel Centrino technology was the company’s first attempt at developing a platform solution. 2004 ($77. “Two of the biggest advocates of Merom were (not surprisingly) Mooly Eden but also Sunil Shenoy. however. Intel produced three different microprocessor architectures: Centrino for mobile. Intel would align its architectures for mobile and desktop on Merom. 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). Interestingly.2M). This had been Intel’s traditional focus for microprocessors and the company was much further along in the development of Tejas than Merom.” Brandweek. Source. 39 It could be used in mobile PCs.

22 strategic. . 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.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. . 23 Exhibit 1 Intel Organizational Structure in 1998 and 2000 Intel Organizational Chart -.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.1998 Executive Office IABG SMG ICG MPG TMG (P&L) PMD BMD server desktop mobile Intel Organizational Chart -.

000 Sun’s Surface 1.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. . 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. given Intel’s current trajectory. 10.000 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.

291 Years Ended December 30.664 $ 7.141 11.382 2.946 2.463 13.145 4.826 $ 34.256 Losses on equity securities.049 19.612 Amortization and impairment of acquisition-related intangibles and costs 126 179 301 548 628 Purchased in-process research and development -.464 Impairment of goodwill -. 25 Exhibit 3 Intel Selected Financial Data 1998-2005 Years Ended December 31.487 Gross margin 23.379 Gains on investments.746 17.914 3.533 4.395 9.094 13.777 14.688 4.318 13.778 4. general and administrative 5.360 4.209 $ 30. (In millions--except per share amounts) 2000 1999 1998 Net revenues $ 33.068 Source: Company reports. . -.764 $ 26.616 9. net 565 289 192 194 393 Income before taxes 12.076 Amortization of goodwill and other acquisition-related intangibles and costs 1.622 17. general and administrative 5.535 $ 7.796 Operating income 12.389 $ 26.137 Provision for taxes 4.331 19. net -45 (2) (283) (372) (466) Interest and other.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.278 4.894 Operating income 10.273 Cost of sales 12.314 $ 6.228 9.726 $ 29.586 411 56 Purchased in-process research and development 109 392 165 Operating costs and expenses 23.509 Marketing.442 4.034 3.606 3.897 3.117 $ 1.801 1.959 9. net 987 578 573 Income before taxes 15.767 8. 2005 (In Millions--Except Per Share Amounts) 2005 2004 2003 2002 2001 Net revenue $ 38.836 12.089 3.516 $ 5.872 3.936 10.130 7.204 2.090 10. -.539 Cost of sales 15.334 4.069 Net income $ 10.183 Provision for taxes 3.052 Research and development 5.796 Marketing. net 3.650 11.141 $ 26.047 13.759 883 185 Interest and other. 98 Amortization of goodwill 1.111 2. 5 20 198 Operating expenses 10.417 7.659 4.901 1.641 $ 3.088 Research and development 3.087 892 Net income $ 8. 617 -.446 13.561 8.610 10.

991 Microprocessor revenue 5.017 Mobility Group 5.427 1.412 19.426 17. 26 Exhibit 3 (continued) Selected Financial Data by Business Group 2003-2005 (In Millions) 2005 2004 2003 Net Revenue Digital Enterprise Group 19. .141 Operating income (loss) Digital Enterprise Group 9.314 966 11.137 24.209 30.533 Source: Company reports.Intel Centrino in 2007: A New “Platform” Strategy for Growth p.075) Total operating income 12.826 34.059 Mobility Group Microprocessor revenue 8.092) (1.778 23. motherboard and other revenue 25.851 8.285 1.743 Flash Memory Group (154) (149) (152) All other (2.667 4.981 5.725 5.704 5.833 1.090 10.278 2.120 Chipset.006 8. motherboard and other revenue 2.131 6.130 7.068 Chipset.086 Flash Memory Group 2.405) (2.608 All other 280 165 388 Total net revenue 38.352 5.330 2.

Intel Centrino in 2007: A New “Platform” Strategy for Growth p. . 27 Exhibit 4 Global PC Volumes by Unit 1981-2009 (est.) 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.

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.) 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. . Impact on performance turned out to be minimal. trading speed for more room on the die to add new functionality. and so the trade-off was worth the risk. 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. Intel consciously made the decisions to give up frequency (and potentially CPU performance). Frequency Low Smaller Die Area Larger Source: David Perlmutter. Area High Freq.

. Value in the mobility space consisted in a platform that addressed four key areas important to mobility users. Adi Golbert.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. secure Longer battery life wireless connectivity Intel Secret Source: David Perlmutter. 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. Beth Eby. Alex Peleg. lighter notebooks (> 70% in MSA) • Wireless connectivity 11 Best 22 Thinner and Performance lighter In Form Factor 33 44 Simple. 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.

3 42. The PCI specifies a computer bus for attaching peripheral devices to a computer motherboard. 2003 .6 66.3 70. 31 Exhibit 8 Market Share and Total Revenue Data for Wi-Fi Sales Worldwide Wi-Fi Penetration in Laptop Computer. --- Others 19% 6% --. Wi-Fi Mini PCI (for laptop computers) IC Vendor Market Share. Source: ABI Research.9 61.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. Used with permission. --- Texas Instruments 11% --. . 2000 – 2006 2000 2001 2002 2003 2004 2005 2006 Laptop computers (millions) 26. --.7 37.9 Source: ABI Research.2 28. Wi-Fi Forecast Database: IC.4 Wi-Fi penetration rate 1% 4% 19% 46% 71% 89% 95% Laptops with Wi-Fi (millions) 0.2006 2003 2004 2005 2006 Atheros 18% 17% 11% 16% Broadcom 30% 30% 17% 20% Intel 11% 47% 72% 64% Philips 11% --.7 54.14 19. --. These devices can be an integrated circuit fitted onto the motherboard itself and/or an expansion card that fits in sockets.26 1.13 6. --- Total 100% 100% 100% 100% Note: PCI stands for Peripheral Component Interconnect. Used with permission.3 32.9 52. Table 3-27.

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. MSS stand for market segment share (or size). 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. SOW data from Gartner. Source: Intel internal reports.Intel Centrino in 2007: A New “Platform” Strategy for Growth p. SOW stands for share of wallet. .

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

Intel Centrino in 2007: A New “Platform” Strategy for Growth p.. Timna. Robert Burgelman.g. Philip Meza. Intel Confidential For use only by Dr. and appropriate Internal Intel Employees Source: Authors. . 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. Banias).