Intel Corporation: The Evolution of an Adaptive Organization

"Intel is the most important company in the history of the microcosm."

-- George Gilder, 1989

"A common thread in all of Intel's success has been technology. Technology has tied it all together across the epochs."

-- Gordon Moore, Chairman Emeritus, 1999

By almost any measure, Intel Corporation has been among the most successful companies in history. According to Time Magazine, "the microchip has become -- like the steam engine, electricity, and the assembly line -- an advance that propels a new economy... Intel is the essential firm of the digital age."

Since its founding in 1968, Intel's corporate strategy has driven, or in other cases adapted to, a rapidly changing technology and industry context. Intel's history can be divided into three prominent epochs: "Intel the Memory Company", which lasted from 1968 until about 1985; "Intel the Microprocessor Company", which lasted from about 1985 until today; and "Intel the Internet Building Block Company", which is now beginning to unfold. The three epochs correspond, to a large extent, to the tenure of the company's top leaders. Robert Noyce and Gordon Moore, Intel's founders, directed the company during Epoch I. Gordon Moore and Andy Grove ran the company during Epoch II. And Andy Grove and Craig Barrett were leading the transformation to Epoch III. Exhibit 1 summarizes some of the key events in Intel's history.

Throughout its history, Intel's strategy has been centered on technological innovation and leadership. In Epoch I, Intel was a broad-based semiconductor supplier that primarily focused on developing and selling memory products. In Epoch II, Intel became a focused microprocessor company at the center of the PC revolution that eventually came to drive the PC industry. The transformation between Epoch I and II clearly changed Intel's substantive corporate strategy (from memory to microprocessor company), however the company's generic corporate strategy remained intact (differentiation through product leadership). The characteristics that make Intel's organization highly adaptive enabled it to successfully navigate from Epoch I to Epoch II.

As Intel enters Epoch III, it faces two major cross-currents: the ongoing pressures from the rapid technology and marketing changes in the PC marketplace; and the emergence of the Internet as a powerful new driving force in high technology. Intel is once again challenged to adapt its organization in the face of major change. Additional new competencies will become important, while some existing ones will become less so. Intel must not only develop new competencies, but also succeed at integrating these capabilities with the existing organization.

Across the three epochs, Intel's challenges can be translated into three strategic management issues: (i) exploiting opportunities in and addressing challenges to its core business, (ii) pursuing new opportunities outside the core business, and (iii) balancing the relative focus on (i) and (ii) over time. A key issue Intel management has continually faced is identifying when a new opportunity or challenge is truly of epoch proportions.

 

Epoch I: Intel the Memory Company, 1969-1985

"We are really the revolutionaries in the world today -- not the kids with the long hair and beards who were wrecking the schools a few years ago."

-- Gordon Moore, Intel founder, 1973

In its formative years, Intel was a wellspring of new technical opportunities. Intel became the first company to specialize on producing large scale integrated circuits, and Intel engineers invented most of the important building blocks of the digital age. Intel was the first company to develop the SRAM (static random access memory), the DRAM (dynamic random access memory), the EPROM (electrically programmable read only memory), the EEPROM (electrically erasable programmable read only memory), and the microprocessor.

In the 1970s and early 80s, Intel had many different products and relatively autonomous business lines. Intel's technical sales force played a major role in applying and selling products, and they provided input about the market and customer needs to the product planning groups. These dynamics allowed Intel to spontaneously take advantage of many new opportunities.

Organization

Intel was founded in 1968 by Robert Noyce and Gordon Moore, two legends in the emerging field of solid state electronics. Nearly a decade earlier, Noyce had shared credit for inventing the integrated circuit, while Moore was regarded as one of the great chemists of the century. Together, Noyce and Moore had helped found Fairchild Semiconductor. Robert Noyce was Intel's first CEO. Gordon Moore became Intel's second CEO in 1974. From the beginning, Intel was positioned to attract the most talented research and development engineers in the field.

Noyce and Moore were soon joined by their Director of Operations, Andy Grove, who had been Moore's Assistant Director of Research at Fairchild. Grove took responsibility for building the organization, and shaped the culture of the company with his focus and aggressiveness. Recalled Moore, "Andy always made it hard on me. I would be all excited that we were under budget or ahead of schedule on a product Ė and heíd ask why we couldnít do it faster and cheaper. He got very interested in the art of management, and that served us very well." Grove became Intelís president in 1979.

Gordon Moore described how he, Noyce, and Grove complemented one another:

"We generally complemented one another pretty well. With Bob Noyce and me, Noyce was more outside, while I was more inside. For example, he organized the second and third financing rounds completely by himself without any real input from me. I focused on the technology and the products, and I had a sound feel for the products and the markets. The thing about Bob was that everyone liked him. He had a tremendous personality, he was very bright, and he was full of ideas. The tough thing was that these ideas were usually too good to ignore... Noyce was what I consider to be a real leader. He would suggest the right things to people and then go away. Andy is a true manager. He's very detail oriented. He has strong follow-up -- he never trusted that anyone would do what they were asked unless there was follow up -- and he is strongly data driven. I was somewhere in the middle."

Andy Grove added,

"At no point in Intel's history has it been a solo show. It's never been only one person leading the organization. Our tradition is somewhat of a shared power structure."

During the 1970s and early 80s, Intel's business lines were relatively autonomous business groups that were run by general managers. Intel had a strong research and development group that generated many new ideas and product innovations. These products were brought to market by a talented technical sales force. According to Dennis Carter,

"The sales force was in direct contact with customers, and served as a lightning rod for marketplace information. They provided input about the market and customer needs, and gave feedback to the product planning groups. You could not make a product successful without the support of the sales force. They represented the perspective of the customer, and they were a unifying force across the various divisions... They became informally part of the development process and provided input to the design teams."

Internal Selection Environment

Culture. Intel's culture was established in its formative years and has remained a powerful force throughout the company's history. Intel's core values were shaped by the behavior of the members of the founding team, who had a deep appreciation of the importance of technical excellence and depth. They insisted on discipline in thinking and action, they focused on results and output, and they wanted to create an egalitarian meritocracy in which knowledge power was not subjugated by position power.

Within Intel, they created a culture of openness that encouraged intense intellectual debate. This key aspect of Intel's culture came to be known as "constructive confrontation." This meant that people should deal with issues openly and directly. During a period of debate, disagreements would be forcefully aired, but once a decision was made, those who disagreed were still expected to commit to the decision. The saying, "I disagree, but I commit", manifests constructive confrontation. Andy Grove explained in 1989,

"Intel is a very open system. No one is ever told to shut up, but you are asked to come up with better arguments. People are allowed to be persistent."

Intel paid great attention to hiring and promoting people that would be compatible with this culture. The importance of recruiting to Intel was illustrated by the fact that Andy Grove personally interviewed all exempt hires until the early 1970s. Speaking of how Intel's culture was established, Andy Grove said,

"There are two key things. Who you hire and who you promote. Who you promote is probably even more important, because it's the most visible statement that you can make about the selection environment... Intel has a strong personality... For many years I was part of the hiring process for every exempt hire. That's 25 years gone, but that influenced generations of hiring managers. We still select people pretty damn well. People may look very differently, but they all have some affinity... As managers, we tend to replicate ourselves... through the people that we hire and the people that we promote... It's like a form of cultural reproduction."

Asked about the selection criteria used during the recruiting and promotion processes, Grove continued,

"I could say all of the usual things. Results orientation. Openness. Dealing with issues aggressively and openly. It's all true, but it somehow falls flat. People each bring their own personalities. The characteristics look different in each person... It's quite instinctive."

A high level of rigor also went into measuring and rewarding performance. Intel used a form of "management by objectives" whereby an employee identified approximately 10 objectives with quantifiable measures. Performance against those objectives became the basis for measuring performance. Intel used a "ranking and rating" process that was applied by each manager to each of his/her direct reports. The process started early in the calendar year and concluded sometime in April, with each employee receiving a rating of "outstanding", "successful", or "improvement required". Dennis Carter, vice president of marketing, explained:

"It is taken bloody serious. If you are unprepared in a ranking and rating meeting to talk factually about the employees you are reviewing, you are bludgeoned. I donít think anything we do is taken more seriously."

Intel's compensation system, which included salary, bonus, and stock options, was tightly linked to the ranking and rating process.

Finance & Resource Allocation. Throughout its history, Intel has always had a strong profit and shareholder-value orientation and the finance group has held a prominent position within the company. For example, during the 1981-82 recession, Intel had instituted the "125% solution", which required all salaried employees to work an additional 10 hours per week without additional compensation. When the recession persisted, Intel instituted a 10% pay cut in addition to the 125% solution. This allowed Intel to minimize lay-offs during the period.

In the early days, when Intel was still more of a start-up, the founders and key team members could rely on their own expertise and intuition to make decisions. The link between the concept of corporate strategy and strategic actions was immediate and direct. Les Vadasz recalled some of the key factors for Intel's success in the early days:

"Intel was a successful start-up in the late 1960s... The reasons for success were embedded in the combined talents of the group that was in charge. We had a "sense" about the technology and the business, which led to a series of correct decisions."

As the company grew larger and more complex, this link became less direct, and Intel senior management developed a series of management tools and mechanisms to implement corporate strategy.

The rules concerning manufacturing resource allocation were a potent part of Intel's control structure. Intel's product divisions shared fabs, and when manufacturing was constrained by high sales, Intel would allocate capacity in such a way as to maximize profitability. The maximize margin-per-activity (wafer-start in this case) became the key allocation rule, which was implemented by mid-level product division managers.

In contrast, the resource allocation rule for R&D was controlled by top management. In the late 70s and early 80s, Intel's three major development areas were DRAM, EPROM, and microprocessors. In 1985, each of these three development groups had approximately equal numbers of personnel and budget allocations. Gordon Moore commented that resource allocation did not necessarily reflect the market fortunes of the product families:

"Allocation of resources to the different technology development groups is centralized by Andy and me."

Product & Technology Development

There were three major types of technological competencies involved in semiconductor products: (1) circuit design (can we design it?), (2) process technology development (can we make it?), and (3) manufacturing engineering (can we manufacture it in large volumes with high yields?).

Throughout its history, Intelís strategy has focused on technology leadership and first mover advantage. Intel engineers invented a wide range of semiconductor devices, including the SRAM (1969), the DRAM (1970), the EPROM (1971), and the microprocessor (1971). Intel established itself as a leader in semiconductor memories by pioneering a new process technology called metal-oxide-semiconductor (MOS).

Process Technology

Intel became the first successful mover in DRAMs because its new MOS process technology pushed manufacturing yields above the threshold for viability. Other companies had designed working DRAMs, but had been unable to develop a process technology to manufacture them successfully. According to Gordon Moore,

"We made just the right technology choice in the early days -- silicon gate MOS. This was a new version of MOS. To make complex circuits required several steps, and it took real work to get by. We had seven years using this technology without competition, and this really helped us financially. We did a variety of different memories, and we were first to market with each major memory technology."

From the beginning, Intel forged tight links between development and manufacturing, and Gordon Moore insisted that Intel perform all process research directly on the production line. The ability to tweak the manufacturing process on the production line had been consistent with Intel's strategy of being first to market with premium priced products. In addition, it had been a source of pride to manufacturing people that Intel could run manufacturing processes that were not inherently stable. As more fabs were added, however, it became increasingly difficult to transfer technology from development to manufacturing, or among fabs. Each fab performed its own process experimentation on the line, and developed its own tuning processes. This ultimately degraded consistency and manufacturing yields.

The Development of DRAM

In 1970, Intel introduced the world's first DRAM, which came to replace magnetic cores as standard technology for computer memories (which store computer programs and data). According to Gordon Moore,

"We started out wanting to find a hole in the product line of established companies, especially Texas Instruments and Fairchild. We wanted to do something different, and not compete directly. We saw the opportunity to do something in semiconductor memory to make a complex chip. This was very difficult to do economically. TI and Fairchild both were making simple chips. We saw memory as a function used in all digital systems. We thought that semiconductors would be very competitive for small memories where magnetic core memories didn't compete well, so we started out making memory circuits."

By 1972, the 1103, Intel's original DRAM, was the world's largest selling semiconductor and accounted for over 90% of Intel's $23.4 million in revenue. DRAMs remained the core product line for Intel throughout the 1970's. DRAMs were initially a very attractive business. Demand grew rapidly as innovations in production processes and memory design led to steadily increasing product performance and decreasing cost. Intel's leadership in design and fabrication yielded strong profits and healthy market segment share throughout the 1970's.

The Development of EPROM

The EPROM was invented by Dov Frohman, who had joined Intel in 1969 from Fairchild. Frohman was credited with inventing, designing, and fabricating the first EPROM, and he used a crude but compelling demo to convince Gordon Moore of its usefulness. Recalled Frohman,

"We put together a 16-bit array with primitive transistor packages sticking out of the 16 sockets, an oscilloscope and pulse generator, and we carted all this into Gordon's office. There were red bulbs to indicate the bits. This was all new to us, and we were thrashing around. We showed Gordon that by pushing the button you could program the device, and we demonstrated that it would hold a charge."

The EPROM market segment was greatly stimulated by the growth of the microprocessor market segment. Unlike ROM (read only memory) chips, which had to be programmed at the factory, EPROM could be easily reprogrammed by customers. Thus, designers could develop and revise microprocessor programs quickly and easily.

Intel had a significant competitive advantage in the EPROM manufacturing process, and retained a majority market segment share through the late 1970s. In 1980, an industry downturn, combined with the entry of several competitors, including Japanese companies, caused average selling prices for the 16K EPROM (introduced in 1977) to drop by 75%. Intel regained leadership with the 64K EPROM (1981). According to Gordon Moore,

"We mis-guessed the role that EPROM would play. EPROMs were more expensive than masked ROMs, and we thought they would only be used for prototyping. We expected all the volume would still be in masked ROMs, so we kept the prices high for EPROMs. It turned out that EPROMs were the engineers' security blanket. The engineers would always want to keep tweaking their code, so people didn't design them out! So the quantities got to be pretty high, but not as high as DRAMs. We ended up hiding this as well as we could. EPROMs were our most profitable product line through 1985."

The Development of the Microprocessor

The microprocessor was invented by Ted Hoff. Intel had been approached by the Japanese firm Busicom to design and build a set of chips for a number of different calculators. Busicom had envisioned a set of around 15 chips designed to perform advanced calculator functions. Hoff suggested building a simpler set of just a few general purpose chips which could be programmed to carry out each of the calculator functions. He was the architect of the chip set which Federico Faggin and a team of designers implemented. The set included four chips: a 4-bit central processing unit (CPU) called the 4004, a read only memory (ROM) chip with custom instructions for calculator operations, a random access memory (RAM) chip, and a shift register for input/output buffering. It took nearly a year to convince Busicom that the novel approach would work, but by early 1970 Intel signed a $60,000 contract which gave Busicom proprietary rights to the design. The CPU chip was eventually called the microprocessor.

While Intel produced chips for Busicom which were successfully made into 100,000 calculators, a debate within Intel developed about whether the company should try to renegotiate the rights to the chip design. Hoff believed that Intel could use the devices as a general purpose solution in many applications ranging from cash registers to street lights, and he heavily lobbied within the company. Eventually, Intel decided to offer reduced prices to Busicom in exchange for non-calculator rights to the design. Top management, however, did not yet understand the full potential of the microprocessor. Ed Gelbach, Senior Vice President of Sales at the time, remembered the decision:

"Originally, I think we saw it as a way to sell more memories and we were willing to make the investment on that basis."

For the next several years, Intel's microprocessor development steadily progressed. The company introduced the first 8-bit processors, the 8008 (1972) and 8080 (1974), and then the first 16-bit processor, the 8086 (1978), followed by the 8088 (1979). At the time, Intel's primary competitors in this market segment were Motorola and Zilog. Motorola's 68000, also a 16-bit processor, had been designed-in at Apple computer and seemed to be gaining momentum. Recognizing a potential threat, Intel, in a project called CRUSH, mobilized SWAT teams of engineering, applications, and marketing people to travel anywhere in the world to get design wins. The group set-out to produce 2000 design wins in 1980, and ended up producing 2,500, the most notable of which was IBM's decision to use the 8088 in its first personal computer. In 1981, IBM accounted for 13% of Intel's sales (including memory and systems sales). Gordon Moore recalled the IBM PC design win,

"Any design win at IBM was a big deal, but I certainly didn't recognize that this was more important than the others. And I don't think anyone else did either."

The 80186 and the 80286 (1982) microprocessors were developed next. The 286 product features were developed through extensive field interviews, which generated over 50 potential applications ranging from business systems to industrial automation. Ironically, the applications list did not include the personal computer, which became the largest application by far. Recalled Gordon Moore,

"In the mid-1970s, someone came to me with an idea for what was basically the PC. The idea was that we could outfit an 8080 processor with a keyboard and a monitor and sell it in the home market. I asked, 'What's it good for?' And the only answer was that a housewife could keep her recipes on it. I personally didn't see anything useful in it, so we never gave it another thought."

The 286 received a cool response from the market after its introduction, and Intel launched Project CHECKMATE, modeled after the CRUSH concept, to produce design wins for the 286. Intel was able to stress specific features that had been included as a result of its market research. Dennis Carter recalled:

"The design wins completely turned around. When we went into CHECKMATE, some market segments were three or four to one in favor of Motorola. By the time we finished, it had turned around the other way."

IBMís continued use of the Intel architecture for its PC line was once again the most important design win. Several factors were important to IBM in making this selection. These included the strengths of Intel's microprocessor technology, development tools, and marketing support, as well as cross-licensing agreements that allowed multiple manufacturers to source these microprocessors. According to Gordon Moore,

"The 80286 was introduced in 1982, and we were given huge forecasts. This was a new generation of products, and we had IBM using several other things too. It looked like we would need more than we would be able to make ourselves. So we ended up sourcing the 286 technology very widely. We had AMD in the United States, Fujitsu in Japan, and Siemens in Europe. We did lots of sourcing. The actual demand was about 1/3 of what we had been led to believe, so this turned out to be a very competitive business. We essentially gave away the profits on several generations of product."

By 1984, Intel believed it had the best chip design capability in the world. The 32-bit 386 processor, the successor to the 286, was introduced in 1985.

Strategic Inflection Point in the Core Memory Business

Soon after Intel's early success in DRAMs, competitors entered the market and began to erode Intel's position. Every 3 years or so, a new generation was introduced with four times as much capacity as its predecessor. By the mid 1970s, Intel was one of several companies vying to be first at introducing the next generation of DRAM memories.

Product Market Strategy. Intel was first to market with the 1K and 4K DRAM. However, with the 4K generation, start-up MOSTEK introduced design innovations that made DRAMs much more user-friendly. Intel redesigned its 4K DRAM to include these same features, and captured a significant share of the 4K DRAM market segment.

With the 16K generation, Intel was late to market and MOSTEK introduced the first volume quantities. In the 16K generation, Intel used process technology innovations to remain a player. Intel developed a new process, called HMOS (high performance MOS), that enabled it to introduce the single power supply 16K DRAM in 1979 (other DRAMs required three separate power supply voltages, which made them considerably more difficult to apply). Intel was able to capture a price premium of about 2 times relative to average prices for the conventional three power supply DRAMs. While Intel captured significant market segment share in the single power supply 16K DRAMs, even by 1981 the single power DRAMs constituted less than 3% of the overall 16K DRAM market.

Intel concentrated on the single power supply 16K DRAM at the expense of development on the 64K generation. Intel management believed that the single power supply would become critical in the memory industry, and that technical challenges would slow the introduction of 64K DRAMs. However, in late 1979, Fujitsu began volume shipments of the first 64K DRAM.

For the 64K generation, Intel tried to compensate for poor yields in manufacturing by building-in redundancy at the chip level. When a fatal defect in the redundancy design showed up late in development, Intel's entry was delayed further. Meanwhile, Japanese competitors were quickly adding capacity. As a result, Intel was not a factor in the 64K generation.

With the 256K generation, Intel was again late to market and again used process technology to compete. Intel developed CMOS (Complementary MOS) process technology that greatly reduced power consumption in working chips. This made CMOS chips well suited for certain applications, and Intel introduced 64K and 256K DRAMs using this technology. Intel management pursued a niche strategy and attempted to differentiate its products from other offerings and sell them on features. The CMOS DRAMs were introduced at prices approximately 1.5 to 2 times higher than conventional DRAMs. Unfortunately, the DRAM market softened in 1984, prices plummeted, and Intel's differentiation message got lost. By late 1984, Intel's market share in 256K DRAM was down to 4%, and it had lost its position entirely in 64K DRAMs.

Environmental Forces. By 1984, DRAMs had taken on the characteristics of a mature product. Product features were highly standardized, and performance progressed along a steady, predictable path, which favored the manufacturing oriented players. Equipment suppliers had become more important in process innovation, which led to a leveling of process capabilities. As a result, both design and fabrication technologies were well understood by all major players in the industry. Moreover, the primary buyer of DRAM changed from engineers to procurement officers, who were more concerned with cost, quality, and reliability than product features.

The competitive nature of this market made low manufacturing costs a key competitive advantage. Manufacturing costs were driven largely by processing yields, which improved as a result of the experience curve. The experience curve caused production costs to drop by about 30% for every doubling in cumulative volume. This was an area where Intel competitors, especially the Japanese, had the advantage. Japanese firms under the direction of MITI had targeted semiconductors as a strategic industry, and this led to aggressive capacity expansion and price competition. In 1984, Hitachi and NEC had combined DRAM market share that was 8 times greater than that of Intel.

Production Allocation. Intel allocated production capacity based on the maximize-wafer-start allocation rule. For Intel, there was as much as an order of magnitude difference between the margin per activity for DRAMs and that for its high margin products. As a result, an increasing percentage of Intel's capacity was assigned to microprocessors and EPROMs. This reflected Intel's competitive advantage in both microprocessors and EPROMs. Bob Reed, Intel's first VP of Finance, insisted at one point that the DRAM manager sign a symbolic check equal to the margin foregone whenever high margin products were bumped by DRAM.

Changes in the DRAM industry had shifted the basis of competition from process technology, where Intel had the advantage, to large-scale precision manufacturing, where Intel competitors, primarily in Japan, had the lead. In terms of production allocation, Intel's exit from the DRAM business was already well under way and DRAMs accounted for less than 5% of Intel's revenue in 1984 (see Exhibit 2). Significantly, Intel's internal selection environment caused this shift before any change was made in Intel's official strategy.

Position in 1984. Intel hoped to leapfrog the competition by being first to market with the 1M DRAM, and Intel believed it was 1.5-2 years ahead of the competition. While Intel believed it had the lead in developing the 1M device, there were questions within Intel as to whether they could compete in this commodity market over the long term. Moreover, production of the 1M DRAM would require at least one $150 million manufacturing facility.

The strategic context for DRAMs at Intel had always been clear. The memory business had "made Intel", as one senior manager put it, and DRAMs were considered to be the technology driver for the company. Intel's sales force thought it was essential for Intel to offer a full product line (though this was in some ways belied by the continued success of Intel's microprocessor businesses), and memories were still the largest market segment in integrated circuits. Moreover, the DRAM development group was considered Intel's best corporate resource for process development. The DRAM development group was working on the 1micron process while the microprocessor group worked on the 1.5 micron process. The R&D investments in DRAM helped keep alive Intel's perception of itself as a memory company. Ed Gelbach, Intel's VP of sales at the time, explained:

"I was in favor of keeping DRAMs from a marketing strategy standpoint. A full line supplier has a basic advantage in any sales situation... Since customers often pay particular attention to their highest dollar volume vendor, it also pays to offer the commodity product since it is generally purchased in high volume. A more subtle reason had to do with reputation. Intel had been known to drop unprofitable products, sometimes leaving customers high and dry. In board meetings, the question of DRAMs would often come up. I would support them from a market perspective, and Gordon [Moore] would support them because they were our technology driver. Andy [Grove] kept quiet on the subject. Even though it wasn't profitable, the board agreed to stay in on the face of our arguments."

In 1984, Intel considered several options: dropping out of DRAM altogether; licensing the technology to another company; staying in as a niche player; or investing in DRAM capacity at the 1M generation and committing to a low margin business. Gordon Moore explained in 1999,

"In mid 1984 the semiconductor business collapsed. The Japanese had discovered EPROMs... DRAM volumes were way under capacity, so we were all losing our shirts. And with the Japanese discovery of EPROMs, our golden goose was plucked rather badly. We had gotten caught back up in DRAM with a new design and a new process. To build this business back to where we would be a significant player we had to build 2 new fabs. But everyone was losing their shirts. So we said, this is ridiculous!"

DRAM Business Exit

The decision to exit DRAMs was a difficult one. Andy Grove recalled a pivotal discussion with Gordon Moore,

"I recall going to see Gordon and asking him what a new management would do if we were replaced. The answer was clear: Get out of DRAMs. So, I suggested to Gordon that we go through the revolving door, come back in, and just do it ourselves."

In November 1984, Intel top management decided not to invest in the 1M DRAM manufacturing facility. In the ensuing months, it became clear that the company's future lay in the microcomputer business. Grove took personal charge of the implementation of the DRAM exit. He made several organizational and personnel changes, and sought to retain key capabilities from the DRAM group and transfer them to the microprocessor group. Grove visited the DRAM groups affected by the change, and greeted them with a presentation entitled: "Welcome to the Mainstream Intel" -- that is, Intel the microprocessor company.

 

Epoch II: Intel the Microprocessor Company,
1985-1997

"Put all of your eggs in one basket, and then watch that basket!"

--Andy Grove, quoting Mark Twain

Some of the greatest challenges facing Gordon Moore and Andy Grove in 1985 were exiting the DRAM business and transforming Intel from a memory company to a microprocessor company. From a manufacturing and production perspective, the transformation was already well under way. Intel's powerful internal selection environment, as manifested in the maximize-wafer-start rule, had already allocated most manufacturing capacity toward the more profitable microprocessor products, and away from commodity memory products.

However, Intel still defined itself as a memory company, and a significant percentage of its research, development, and engineering resources were devoted to memory products. Intel's management team had risen and developed in a broad-based semiconductor company that focused primarily on memory products. In a discussion of Intel's executive staff at the time, Gordon Moore said,

"You know, if we're really serious about this [becoming a microcomputer company], half of our executive staff had better become software types in five years' time."

Corporate Transformation

Competing in the microprocessor business required different competencies from the memory business. For example, DRAMs required a tight relationship between circuit design and process technology development, whereas the key value add in microprocessors was in logic design and product design (see Exhibit 3). Andy Grove explained that by the mid-80s,

"Intel had moved from a silicon-based distinctive competence in memory products to a distinctive competence in implementing design architectures in logic products."

Intel was able to exit effectively from DRAMs and preserve and redeploy important corporate competencies without having to rely on extraordinary foresight on the part of top management. Intel's internal selection environment had caused managers to allocate scarce manufacturing capacity toward high margin products, reflecting the competitive reality of these businesses. This put the burden of proof on managers who wanted to forego profits for "strategic" reasons. Top management's strategic recognition capacity led them to draw decisive conclusions about the company they wanted Intel to become. Gordon Moore said,

"With the decision to drop out of DRAMs, the technical teams were increasingly diverging... We had three generations of products in three product categories (DRAM, EPROM, and microprocessors). In DRAMs, we were out. In EPROMs we were pushed out. So we had 3 groups to focus on the microprocessor business. The timing of that focusing was very fortunate, given the big growth in the microprocessor business."

The transformation of Intel from a memory company to a microprocessor company required significant strategic leadership. According to Andy Grove,

"The Grove leadership approach consisted of trying to persuade and sell the new strategic approach to the management team... After some period of time, the new strategy had traction with some managers and it did not have traction with some others. The people who did not get traction -- they may have provided lip service to the new strategy, but their actions were not so supportive -- the approach was to remove these people from positions where they could choke progress. We moved them around to other positions where they couldn't impede progress. This worked for a period of time. But when it became obvious that they were in a position that was not so important or influential, several of them left. We didn't actually have to fire anyone, nor were we happy that they left. But they were not happy being in a non-core activity."

Grove continued,

"It's not always clear why you do certain things. You do a lot of things instinctively, without knowing why you're doing it. I knew we had to get out of DRAMs and put all of our best and brightest on microprocessors. It was easier for people to accept that we had to get out of memories, than it was to face what that meant. It was easier to convince people that we needed to focus on the microprocessor business, than it was to get them to face what that meant. We eventually put people in places where they would have a good chance of achieving the strategic objectives. I screwed around for about a year trying to make the existing incumbents in the organization work."

Grove believed that the greatest challenge of his tenure was shifting the focus and direction of the company. According to Grove,

"The most significant thing was the transformation of the company from a broadly positioned, across the board semiconductor supplier that did OK, to a highly focused, highly tuned producer of microprocessors which did better than OK."

Organization

In 1987, Andy Grove became CEO, while Gordon Moore remained chairman. During the second epoch, Grove and Moore worked closely with each other in running Intel. Moore explained how they complemented each other:

"Andy tends to see things in black and white, whereas I tend to see things in delicate shades of gray. In running the company, he was a very good amplifier for me, and we had a good relationship. He ran everything internally, and I worked through him and others... When he became CEO, he really jumped on the opportunity to organize the industry. I wasn't so inclined to do this. He likes public exposure more than I did, and he has a stronger feeling about where it fits in... Andy has had a tremendous impact on what's going on outside. He needed someone who was focused internally, and that's where Craig [Barrett] came in."

Craig Barrett was named executive vice president and chief operating officer in 1990.

Strategic Long Range Planning (SLRP). One key element of Intel's corporate structure was the Strategic Long Range Planning (SLRP) process where Intel's senior executives met to discuss important issues related to corporate strategy. Commenting on this process in 1989, Les Vadasz said,

"As the company grew, we tried to replicate the environment that had led to 'correct' decisions by forming relatively small business units and creating a bottoms-up strategic planning system. However, that became unwieldy. The notion of pushing decisions down may have been a good one, but the task relevant maturity was not great enough. Managers started gaming with the system."

A key element of strategic leadership and corporate transformation is communicating the new strategy throughout the organization. As CEO, Andy Grove made significant changes to the SLRP process. Grove explained,

"In 1987, we blew up the SLRP process. Formerly it had been a very bottoms-up process, but there was no strategic framework. Each of the different groups was supposed to come up with the strategy for their group, and then we would try to piece them together like a jigsaw puzzle. By '87, I was so frustrated with the whole thing that I started the process of turning the SLRP process on its head. I said, 'I'm going to tell you what the strategy is.' I started with a detailed discussion of the environmental issues, which led to a series of strategic mandates. I did not consult the organization. I did this myself, along with the help of my TA at the time, Dennis Carter... I became very directive in prescribing the strategic direction from the top down. This defined the strategy for all of the groups, and it provided a strategic framework for different groups at different levels of management. It's very hard to reach through several layers of management to communicate the strategy and the vision. SLRP became a tool for doing that."

Organization Structure. Under Grove's leadership, Intel became an extremely focused organization. Craig Barrett explained,

"We've had two hugely different environmental periods. In the first case, we had IC expertise and we were looking for a place to apply it. We were looking for problems to which we had the solution... In the second case, this evolved into a focus on the microprocessor business. Quickly, the environment changed to where we knew as much about the system and platform as the customers. We became the industry driving force rather than just facilitating the industry with our technology... With this, we became much more verticalized behind IA and related businesses."

Strengthening the position of the Intel microprocessor in the computing industry became the company's top priority, and Andy Grove made sure that everyone within Intel considered this "Job 1".

Intel was organized in a matrix structure consisting of both product and functional groups. The product groups included Desktop Products Group (DPG), Mobile & Handheld Products Group (MHPG), Enterprise Products Group (EPG), and Internet & Communications Group (ICG). The functional groups included Sales & Marketing Group (SMG), Corporate Business Development (CBD), Technology & Manufacturing Group (TMG), Content Group (CG), and Microprocessor Products Group (MPG).

Core Business Strategy, 1984-1997

The mid 1980's to mid 1990's were very good years for Intel. As sales of IBM and IBM-compatible personal computers skyrocketed throughout the 80's and 90's, Intel prospered. Industry analysts estimated that by 1993, Intel's 486 microprocessor accounted for 75% of the company's revenues, and 85% of its earnings. Personal computers based on the Intel Architecture represented 75-85% of the entire PC market (see Exhibit 4).

During the 10-year period between 1987 and 1997, Intel achieved exceptional financial performance (see Exhibits 5-8). Sales of Pentium® microprocessors and related board-level products comprised about 80% of the Company's revenues and a substantial majority of its gross margin in 1997. Revenue increased from $1.9 billion to $25.1 billion, an increase of 29.4% per annum. Net income grew from $248 million to $6.9 billion, an increase of 39.5% per annum. $100 invested in Intel stock in 1987 would have been worth $2078 in 1997. Intel operating activities generated $8.7 billion in cash in 1996, up from $4.0 billion and $2.9 billion in 1995 and 1994, respectively. In 1997, Intel was the sixth most profitable company in the world, and its net profit margin of 28% was 2.5 times higher than any other company in the top 10. Exhibit 9 shows Intel's market capitalization over time.

Relationship with Microsoft

IBM brought Intel and Microsoft together in 1980 to provide the key components in the original IBM PC. Intel and Microsoft's rise in the computer industry had been highly symbiotic since then, and the two had become the clear leaders in the worldwide PC industry. Microsoft's MS DOS, Windows 3.1, and Windows 95 operating systems ran on over 85% of the world's PCs. These operating systems had a tremendous installed base of available applications that ran on them. And to Intel's benefit, these operating systems were written with the largest market in mind and therefore ran only on Intel Architecture microprocessors.

The success of Windows 3.1 and Windows 95 were key factors in Intel's growth and profitability. These operating systems transformed the "IBM PC" from a character based interface to a graphical environment, which required considerably greater processing power and stimulated the overall market for PCs and microprocessors.

Microsoft introduced Windows NT in 1993 and targeted the new operating system at high-end desktop PCs, workstations, and servers. Although Windows NT was designed to be easily ported to multiple microprocessor architectures and had been ported to several, the vast majority of systems that utilized NT did so on Intel microprocessors. From 1993 to 1999, Windows NT, combined with high-performance Intel processors, made steady and significant advances in the market segments for workstations and servers, increasingly displacing RISC/UNIX systems.

Compatibility & Applications

To drive demand toward the newer products, Intel cultivated new businesses and worked with the software industry to develop compelling applications that would take advantage of this higher performance. The availability of new applications, combined with the compatibility of new Intel processors with previous ones (and therefore the ability to run the existing base of applications), were key elements of Intel's strategy. According to one senior technical manager:

"Our biggest advantage is compatibility. As long as we can offer world-class performance, our customers will continue to choose our architecture to maintain compatibility with the huge software base written to the Intel architecture."

In many respects, compatibility was more important than performance or cost. The technical manager continued:

"Compatibility costs us 10% to 30% in terms of performance because we have to have a more complex design. Complexity costs in two ways - 1) more transistors lead to a larger chip which is more expensive to make; 2) it hurts on performance because a more complex design is often slower or harder to optimize. However, the difference in size isn't as great as the competition is advertising, we are currently compacting the Pentium processor to decrease the number of transistors and move it too a smaller (line width) process. This will reduce the size of the chip and the amount of heat it generates. When we finish this, about the time our competitors will have their chips available, we'll be on par. But sure, because they are starting from scratch, they don't have this added burden of complexity."

He added,

"In the early Pentium processor development we did extensive investigations and asked our customers, 'What would it take to get you to switch?' They said it would take more than a 2X difference in performance."

The importance of compatibility was not originally obvious to Intel. According to Paul Otellini, who managed Intel's relationship with IBM in the early 1980s,

"We didn't really understand the importance of compatibility between different generations of processors. I remember a meeting in 1982 with [the person who ran the IBM PC division]. We had just finished the 286 and were shipping it to IBM. IBM called me and Gordon [Moore] up to San Francisco and lectured us on the importance of compatibility -- the 286 was not perfectly compatible with its predecessor. As a result of this experience, we finally got it. Perfect backwards compatibility is really critical."

Complementors

In order for Intel's technology leadership to be relevant, it was essential that customers demand ever increasing processing power. Thus, Intel focused on stimulating the supply of and demand for high-performance content and applications. Intel had pursued strategic alliances with developers of multimedia and communications applications, as well as computer based video games. Compatibility became a key link in the virtuous circle that emerged between the installed base of Intel Architecture-based systems, and the availability of applications that ran on these systems.

Sole-source Strategy

During the early 1980's, IBM and IBM-compatible PCs became the leading PC platform, and the Intel Architecture became the highest volume microprocessor architecture by a wide margin. Since the 1970s, Intel had licensed its processor designs to other chip manufacturers to enable them to second-source Intel processors. This included the 8080, 8085, 8086 and 286 designs. As a result, in 1984 the Intel Architecture accounted for 59% of all 16-bit microprocessors, but Intel's share of the market segment was only 14.5%.

In 1985, beginning with the 386 generation, Intel changed its licensing methods. Intel decided that prospective licensees would have to pay to help defray the costs of technology development. However, none chose to do so. As a result, Intel decided to go it alone. The company no longer actively licensed the intellectual property necessary for others to produce the 386 or future generation processors. Competitors could still develop Intel-compatible microprocessors, but they would have to do so without the benefit of Intel intellectual property. Competitors would have to reverse engineer these products, a time consuming and difficult process. As a result, Intel served as the sole source for the 386 for four years, and after introducing the 486 in 1989, served as sole source until mid-1993.

Technology Leadership & First Mover Advantage

Intel's strategy was to continually introduce higher performance microprocessors, and the company did so at a relentless pace. Between 1989 and 1997, Intel launched members of the 486 family, the Pentium processor family, the Pentium processor with MMXÔ technology and the Pentium II processor. (Exhibit 10 shows the evolution of Intelís microprocessors over time.)

Albert Yu provided some background on microprocessor design:

"Versions of each generation of the microprocessor architecture follow a road map which usually contains three dimensions: Performance, Packaging, and Features. Performance and Packaging are rather obvious. For example we offer a range of clock frequencies (i486 25 MHz, 33 MHz, 50 MHz). The faster the clock, with all else equal, the faster the processor. There is also a well-known set of packaging options that customers demand. Features are much less obvious. These refer to specific functions which may be included on the microprocessor chip which aren't necessarily part of the core processor. Our products serve three markets, Servers, Desktop PCs, and Mobile Computers. These markets look similar, with many of the same players (OEMs). However, the details (feature set required) are very different."

Manufacturing Capacity and Capabilities

To implement its strategy, Intel aggressively invested in both R&D and production capacity for its high-performance chips (see Exhibit 7-8). Production investment decisions were usually made long before demand for these chips could be ascertained. Intelís capital investments in property, plant and equipment had grown from $302 million in 1987 to $4.5 billion in 1997, three times that of any other chipmaker in the world. According to Intel president Craig Barrett,

"It's a risk to go out and spend billions of dollars on these manufacturing plants. But if we didn't, we couldn't possibly reap the benefits. We're going down the road at 150 miles per hour, and we know there's a brick wall someplace, but the worst thing we can do is stop too soon and let somebody else pass us."

Intel's Technology and Manufacturing Group (TMG), led by Gerry Parker, was a functional organization that managed Intel's manufacturing sites, or fabs. TMG engineers were closely matrixed with product groups because of the close link between product and process design.

Economies of Scale

The cost and complexity of microprocessor development increased dramatically with each new generation of product. This placed a premium on volume and scale. According to Albert Yu, head of MPG, speaking in the early 90s:

"Volume is key to everything. A leading edge design will take 50 to 100 top engineers two or three years to develop. Total development costs will probably range from $50 to $100 Million. In addition to that, the processor must make use of the latest manufacturing technology to be cost effective. A leading edge fab can require $700 to $800 Million in capital investment. You have to sell a lot of processors to recoup those costs."

Albert Yu continued,

"Another trend we're facing at MPG is the increasing number of engineers required for each new generation of the architecture. There was as much innovation in the administration of the Pentium architecture development project as there was in the technology. The 386 team consisted of about 30 engineers, the 486 was roughly 80 engineers, and the Pentium processor design team was over 100 engineers. Right now we have even more engineers working on the P6. That's about as large as we can go. We have to find a way to handle the increased complexity of future generations through automating the design process rather than by simply adding more people."

Design Tools

Intel bought off-the-shelf tools for designing standard elements of the microprocessor, but for the most complex and innovative elements, Intel developed its own design automation tools. Intel made large investments to develop these tools, and this gave Intel an important advantage. According to Albert Yu,

"We have made a big investment in proprietary design tools. For the last decade we have led the industry in the complexity of logic design. When we were putting 1 million transistors on a logic chip, the industry was at 500k. We have no choice but to do our own tools. This is also an advantage over our competition - they have to either build their own or work with less advanced, generally available tools."

Another senior technical manager elaborated,

"We are constantly co-developing new design tools along with our leading-edge process technology. One of our key tools helps us move a design from the drawing board to our most advanced process technology faster than anyone else in the industry. However, the best tool is still by hand. We still lay out the critical elements of the processor by hand (less than 10% of the total transistors). No tool can absolutely optimize as well as the right person."

Corporate Marketing

The group that became Corporate Marketing Group (CMG) began as the End-User Components Division within MPG in 1989. By 1992 it had undergone significant growth and change and became a functional group of its own known as CMG. Intel launched its first end-user advertising program in the fall of 1989 to encourage end users to adopt the latest Intel technology. This included programs to help speed the transition from the 286 to 386 (which was known as "Red X" because it featured print ads with a large red X crossing out a 286 logo), and from the 386 to 486. Intel would later pursue two much more prominent marketing programs: the Intel Inside program and the microprocessor branding strategy. From 1990 to 1993, CMG invested over $500 million in programs designed to increase Intel's brand equity with end users.

Intel Inside. Intel Inside was the company's first comprehensive effort geared toward turning the Intel name into a consumer brand. Launched in April 1991, Intel Inside was designed as an ingredient brand program intended to work with the OEMs while communicating directly to end users. The program sought to influence PC buyers to specify Intel microprocessors when they purchased a PC. This effort marked the transformation of Intel from a microcomputer component supplier targeting OEMs to a branded products company that also marketed directly to a non-technical mass market. Intel Inside was composed of two broad thrusts. The first involved advertising and merchandising activities with OEMs, which included cooperative advertising that displayed the Intel Inside logo and promotions where the OEMs' point of sale displays and packaging materials would include the Intel Inside logo. The second thrust was a series of print and TV ad campaigns focused on developing the Intel brand directly with end customers. The key brand attributes emphasized by Intel included high performance, advanced technology, and software compatibility. According to Intel's market research, the Intel Inside campaign significantly increased end user awareness of the Intel brand name and enhanced the perception of Intel as a technology leader. By the end of 1991, over 300 OEMs participated in the program. However, not every OEM was enthusiastic about Intel Inside, and major OEMs such as Compaq and IBM went through periods of participating and periods of not participating in the campaign. These OEMs felt that Intel Inside decreased their ability to differentiate products from their competitors.

Microprocessor Branding -- the Pentium Processor. Intel's microprocessor branding strategy furthered the company's transition toward a branded products company. Through the 70s and 80s, Intel's microprocessors had names such as 4004, 8008, 8086, 286, and 386. These numeric names were not properly trademarked, therefore when competitors introduced Intel-compatible processors, they followed the same naming convention. After an unsuccessful attempt to prevent other suppliers from using the 386 name, Intel decided to develop a new naming convention that could be registered and protected as a trademark. This led to the selection of the Pentium name for Intel's fifth generation processor.

Relationship with OEM Manufacturers

As a result of Intel's successful marketing programs, the company became far more central in the PC industry. Many industry analysts believed that this, combined with the limited R&D investments made by OEMs, reduced the OEM's ability to differentiate their products. Fierce competition among PC makers rapidly lowered the cost of systems based on Intel's processors, and Intel and Microsoft captured a significant percentage of the industry's overall profits.

Legal Actions

Intel significantly increased its legal staff during the 1980s in order to better protect its intellectual property. According to Bob Reed,

"If our primary value added is in our design capability, we've got to protect that with vigilance. We have a strict policy of pursuing anyone or any company that appropriates our intellectual property -- design or process."

Key Challenges in the Core Microprocessor Business: 1984-1997

Intel faced many challenges in the second epoch, the most important of which follow:

Competitive Rivalries

During this period, Intel faced considerable competition in its core microprocessor business. The first and most direct competition came from other manufacturers who produced Intel compatible processors. The second came from vendors who were promoting a new microprocessor design methodology known as Reduced Instruction Set Computing (RISC).

Intel Architecture Competition. The most consistent competitors that Intel faced who manufactured Intel Architecture-compatible microprocessors were AMD and Cyrix. AMD was founded in 1969 and had become one of the largest semiconductor companies in the US. The company claimed to be #1 or #2 in every market that it served. Since the early 1970s, AMD had manufactured Intel-compatible microprocessors based on licenses from Intel. Beginning with the 386 generation, this was no longer the case. AMD introduced a 386 compatible processor, which it reverse engineered, in 1990, and by 1992 AMD was the 386 leader with over 50% market segment share. AMD introduced its 486 compatible processor in 1993, and its Pentium compatible processor, which it dubbed K5, in 1994. In 1996, AMD acquired NextGen, Inc., whose president, Vinod Dham, had been a senior technical manager at Intel. The combined company collaborated on the K6 processor, with which AMD had some success, gaining design wins with Digital Equipment Corporation and IBM, among others. Aggressive pricing was a key element of AMD's strategy, and AMD's stated goal was to achieve 30% market segment share in Intel-compatible processors by 2000, up from its 1997 share of approximately 15%.

Cyrix was founded in 1987 by a group of managers from Texas Instruments. Cyrix's first product was a math coprocessor that worked with Intel microprocessors. Cyrix introduced a 486 compatible processor in 1991 that fit into a 386 socket. This chip was attractive to OEM manufacturers who had excess inventory of 386 motherboards in the transition from the 386 to the 486 generation. Cyrix also introduced a 486 processor and a 586 (Pentium compatible) processor. In February 1997, Cyrix introduced the Cyrix MediaGX Processor, which reduced system costs by incorporating multimedia and systems functions on the CPU -- a first for the industry. This innovation enabled the introduction of a complete multimedia system with a 180 MHz processor, monitor included, for $999. Cyrix outsourced the production of its processors to IBM Microelectronics and SGS-Thompson, providing Cyrix with access to leading-edge process technology and volume capability.

RISC Competition. Reduced Instruction Set Computing was a term used to describe a microprocessor design philosophy that emerged in the early1980's. Microprocessor instructions are the lowest level commands processed by a microprocessor, and include functions such as "retrieve from memory" or "compare two numbers". RISC processors support fewer instructions (70-80) than CISC (100-150). As a result, RISC chips offered better performance over a narrow range, and could be optimized for a specific purpose. Also, RISC processors were much simpler to design and much cheaper to manufacture. The primary RISC vendors were Sun (SPARC), HP (PA-RISC), IBM and Motorola (PowerPC), Digital (Alpha), and SGI (MIPS).

Intel countered claims of superior RISC performance in general use by pointing out that the 486 and the Pentium processor incorporated many RISC features and offered competitive performance in important benchmarks. According to Albert Yu,

"There is a lot of marketing hype from our competitors about RISC processors being superior to our architecture. RISC vs. CISC is not a technical discussion. RISC means simplifying as much as possible. This is good in any setting. RISC is a set of techniques that can be applied to microprocessor design to increase overall throughput. We continue to apply these techniques to the Intel Architecture. The 486 and Pentium architectures incorporate many of these design concepts."

Apple Computer was the primary competitor to Intel-based systems in the PC industry. Apple used Motorola's PowerPC, one of the leading RISC processors, in its personal computers. According to Dennis Carter,

"In 1994, Motorola and Apple ran ads about the PowerPC that showed RISC growing in performance over time, and CISC and Intel running into a brick wall in a few years. They essentially said that Intel has a clumsy architecture, and no future. Every press interview I did, the press said, 'you guys are dead'... From a marketing perspective, it was a disaster for Apple. Apple had always previously focused their positioning on the ease of use of the whole system. Now suddenly they were playing our game and calling out the microprocessor as the most important part of the PC. And on top of that, they were way over promising which cost them their credibility."

In the late 1980's, RISC architectures dominated the market for technical workstations. The leading RISC vendors were Sun Microsystems, IBM, HP, Digital, and Silicon Graphics, and these companies each developed their own proprietary architectures which included both a RISC microprocessor and a proprietary version of the Unix operating system. These systems were not compatible with all the applications written for Microsoft's DOS and Windows operating systems. Hence, workstations with RISC processors did not gain a significant presence in the much larger market for PCs. Conversely, much of the software written for RISC/UNIX workstations would not run on the PC and the PC was not perceived as being powerful enough to move into the workstation market segment. According to Dennis Carter,

"In the early 90's in the very high end processor segment we were perceived as technology laggards versus the RISC vendors, and this hurt our growth into the workstation market segment."

The Pentium Processor Flaw

After introducing the Pentium processor in March 1993, Intel discovered that the chip had an obscure flaw that caused extremely rare computation errors when performing certain types of mathematical calculations. Intel had evaluated the problem and determined that the average user would encounter the problem only once every 27,000 years or so. As a result, it was extremely unlikely that most users would ever encounter the error and this so-called floating point flaw was not considered particularly significant or urgent.

After all, microprocessors were extremely complex chips with millions of transistors and they commonly had bugs, especially when they were initially released. Intel had been far more careful and concerned about product quality with the Pentium processor than it had with any previous processor. A senior technical manager explained, before the Pentium processor flaw surfaced:

"When the 486 came out, there were many bugs -- bad bugs -- that we didn't catch. One bug that was found by a customer surfaced after we had begun production -- 50,000 486s became key chains instantly. That's a lot of very expensive key chains! We were at our customers' mercy -- they were finding the errors in our products... For the Pentium architecture we assigned a team to be responsible for interfaces only (where most of the 486 bugs had occurred). As the project was nearing completion, after each of my teams assured me that there were no bugs, I offered to pay anyone $100 on the spot if they found a bug... This was a lot cheaper than finding the bug later or having a customer find the bug for us."

Relatively speaking, the Pentium processor flaw looked quite innocuous. This bug could be corrected in later versions of the chip as part of the normal succession of changes that were made to reduce the chip size or cut manufacturing costs.

However, in October 1994 a mathematics professor discovered the bug, and a week later hundreds of users had posted messages to Internet news groups discussing the problem. For a heavy-duty user that was likely to be affected by the flaw, Intel agreed to replace the chip with a corrected one. For the typical user, Intel explained how unlikely it was for the error to surface, and offered to provide a detailed white paper that explained its analysis. Initially, Intel's approach seemed to be working.

Then on November 7, 1994, the Electrical Engineering Times ran a cover story on the Pentium processor flaw. On November 22nd, CNN ran a feature story that discussed the issue in depth. Then on December 12th, IBM announced that it had stopped shipments of all Pentium processor based computers, claiming incorrectly that errors due to the flaw were more common than Intel had reported. According to one analyst,

"IBM's ban on Pentium sales has little technical merit... The ban appears to be a marketing ploy to make IBM look good and kick Intel when it's down."

Nonetheless, the ensuing publicity caused call volumes to Intel's customer support lines to skyrocket. The problem would not go away. On December 19th, Intel's board met to discuss the issue, and decided to reverse the old policy and replace any Pentium processor affected by the flaw regardless of the user. Intel took a $475 million charge against earnings for the quarter to cover the expense.

The success of Intel's marketing campaigns -- Intel Inside and the Pentium processor branding efforts -- had helped make Intel into a consumer products company rather than just a semiconductor component supplier. The Pentium processor flaw crisis made this transition painfully manifest.

Changes in the PC Market Segment

Several significant changes occurred in the PC market segment that influenced Intel's strategy:

The Network Computer (NC). In 1997, the network computer (NC) emerged as a potential threat to Intel. Companies like Sun Microsystems, Oracle, Netscape, and IBM were evangelizing network computing, a new computing paradigm based on so-called "thin-clients" and "fat servers", that they claimed would dramatically reduce the total cost of ownership (TCO) of desktop computers Ė especially maintenance and management costs. The NC was expected to include low-end processors running Sunís Java software and an Internet browser Ė they would not include Intelís microprocessors or Microsoftís Windows operating systems. According to one analyst,

"In the NC worldview, a desktop computer will consist of an inexpensive, relatively low-powered processor, a minimum amount of memory, and a browser, connected to a variety of servers that store applications and data."

A key issue driving the viability of the NC concept was the maintenance expense of the PC. The Gartner Group, in a highly publicized report, estimated that the cost of maintaining a PC over a 3 year period was approximately $25,000. The NC was targeted primarily at corporate customers, though it was also positioned as a low-cost alternative for consumers that could not afford a conventional PC.

Response to the Network Computer: the NetPC. Intelís response to the perceived threat of the network computer (NC) was focused and swift. A division within Intel's Desktop Products Group (DPG) was given the charter to design a low-end PC, called the NetPC, which would include a Pentium processor, PC management software from Intel, and no floppy disk drive. DPG quickly ramped up efforts for the NetPC, growing staff to approximately 70 engineers. As a result, the NetPC competed for some of the same resources as other Intel ventures, including Hood River and the networking business.

In addition to the NetPC project, Intel made significant efforts, primarily in the area of PC management software, to reduce the total cost of ownership of Intel PCs. These efforts by Intel contributed to the demise of the NC concept, which by 1999 was no longer a threat. One Intel engineer noted, "We respond extremely well to threats, but perhaps not so well to opportunities."

Emergence of the sub-$1000 PC. In early 1997, PC makers began introducing stripped-down, low-cost PCs, priced below $1000. The implications of this development were not completely clear at the time, but there was the potential that this could disrupt Intel's business model. Moreover, there was concern that low-end pricing could affect pricing at the high-end as well.

New Business Ventures, 1984-1997

During Epoch II, a number of new business ventures were initiated at Intel, some of which became successful and some of which did not. The most significant of these efforts are discussed below:

Response to RISC: the i860 Processor

In the late 1980s, the Intel Architecture was the focus of the company's microprocessor franchise. However, the Intel Architecture had been challenged internally by a RISC design developed by a small team within Intel's own microprocessor design organization. Led by Les Kohn, the project had been demonstrated as a coprocessor for the 486, rather than as a standalone processor. Les Kohn explained,

"I realized that RISC architecture had some definite technical advantages... Between 1982 and 1986, I made several proposals for RISC projects through the Intel product-planning system, but I wasn't successful... In 1986, I saw that our next generation processors would have 1 million transistor chips, and I started working on the idea of a RISC-based processor that would take full advantage of that technology... Several people... got interested in the idea... We positioned it as a coprocessor to the 486 and made sure it could be justified on that basis. We designed it as a standalone processor, but made it very useful as an accessory to the 486... The real fortuitous part came when presentations to several large customers generated a lot of positive feedback to senior management. Feedback helps because at a technical level, senior managers are not experts."

The i860 was an advanced processor that had a 64-bit architecture and enhanced graphics capabilities. According to Kohn, the chip utilized design concepts found in supercomputers. Kohn positioned the i860 as the ideal processor for Intel to target advanced workstations. In February 1989, Intel announced the i860 as a stand alone RISC processor. According to Craig Barrett at the time,

"There has been competition between the RISC and CISC architectures for some time... Intel's bread and butter clearly is still the x86 family... But if there are enough people out there who wan to buy YARPs (yet another RISC processor), then... the i860... is the highest performance RISC processor on the market."

The i860 created marketing challenges for Intel. According to Dennis Carter,

"From a marketing perspective, this became a major issue for me. There were some people that advocated introducing the RISC processor as the 486R, while the CISC version would be known as the 486C. Despite the similarities in the proposed names, the two chips were not at all compatible! This would destroy everything that we had created. I became pretty hysterical about it. In the end, the i860 was launched at the same time as the 486, but nothing about 86 was associated with it. The key issue was 'don't confuse the brand'!"

According to Claude Leglise, who ran the engineering and marketing groups for the i860,

"In the days of RISC versus CISC, we were hedging our bets. Everyone told us that the 386 was a technical aberration and a dead duck. Some people inside the company were convinced of this as well. So we hedged our bets. In turns out that there was a lot of hype, and we got caught up in it. The reality is that there was not enough business to sustain the investment necessary to be competitive... We won a number of deals, but all of them were low volume. We won Stratus Computer, which was a pretty big deal. We won every high end 3D graphics accelerator deals. But at the end of the day, there was no volume."

Dennis Carter added,

"As a company, we waffled on the CISC versus RISC issue. RISC ended up being sort of a red herring, but we were affected by the hype. As a technology leader, this made the debate even more wrenching, and it tore the company into two camps... In the end, Andy resolved the debate. He essentially did a compromise that favored CISC. But the RISC efforts siphoned off hundreds of people just on the marketing side. A lot of people worked with partners such as software vendors, which was necessary to launch a new architecture. RISC was a hedge bet by Intel. The i860 ended up petering out. There were not enough customers."

After a protracted internal debate, the Intel Architecture won, and by 1993, Kohn and most of his team had left Intel. However, Intel succeeded in retaining most of the people associated with the program. The ability to retain competencies by redeploying human resources is an important ability at Intel. Claude Leglise explained,

"After the i860, I shut down the operation and found jobs for everyone else. Andy [Grove] says that most CEOs move money around, but not people. We've institutionalized the redeployment of people. We move people around. I went from 225 employees to 3 in about 90 days. This was a very interesting experience, and it's something that Intel does quite well. This leads to less fear in the organization of doing something risky."

Systems and Motherboards

Intel's Systems Business got its start in the mid-1980s, under the direction of Les Vadasz. The Systems Business provided technology that enabled the growth of Intel's microprocessor business. Vadasz explained,

"We were providing our customers multiple choices at different levels of integration. If they wanted microprocessors or board level products, we could provide either."

With its systems business, Intel enabled many OEMs, who otherwise would have been unable to compete, to introduce their own PCs with leading-edge systems sourced from Intel.

At times, the systems business had great strategic value to Intel. A key element of Intel's strategy was introducing and accelerating the adoption of new processor generations. This strategy was not always well-aligned with OEM manufacturers, who were often content to extend the life of a processor generation, especially if they had invested significant R&D in the development of their own systems for that generation. Andy Bryant, who was the controller for the systems group in the early 90s, provided a poignant illustration:

"One major customer told us that we were moving too fast from the 486 product line to the Pentium product line. Grove then told our motherboard division to stop building boards for the older technology 486 processors and to exclusively produce boards for the Pentium generation, and that they were to aggressively sell the newer technology products. He reiterated that their task was to move advanced technology products into the market place."

Dennis Carter added,

"With the introduction of the Pentium processor, the motherboards became a big deal. Both Compaq and IBM decided not to design Pentium processor based systems for the consumer segment. We had a motherboard which enabled smaller companies to introduce Pentium processor based systems, who otherwise wouldn't have been able to. Despite IBM and Compaq, we sold lots of them in 1994, the year of the Pentium processor ramp."

At other times, the strategic value of the systems business was less significant, and Intel managed the group differently in that circumstance. Andy Bryant explained,

"At the point you are no longer getting a long term strategic benefit from an activity, you have to make sure you are getting good short term returnsÖ We manage businesses differently at different times. Systems and boards were a vehicle for bringing advanced technology to the market quickly. If the technology is already moving into the market fast enough, then the products that serve this purpose have less long-term value. When this happens, current profitability is more important, and we manage with a more short-term return focus."

Bryant continued,

"We do our best to quantify the return of the strategic value. We say, let's define the long term benefit, and get a sense for what the cost is to accomplish this. Then we ask, does this make any sense?"

Bryant provided another example of how optimizing the strategy for the core microprocessor business could negatively impact the financial performance of a non-core business:

"At a time when motherboard pricing was extremely competitive, the motherboard division decided not to use Intelís chipsets because they were more costly than third party alternativesóeven though they provided superior performance. Again, Grove said that the key mission of the group was to move advanced technology into the market place, and that saving $10 a board to ship inferior technology was unacceptable (it should be noted that this was a substantial portion of the margin on certain boards). If Intelís motherboards, which were targeted for high end PCs, had the superior performance, the rest of the market would need to respond with similar high technology products. Finance made a presentation that supported the use of the cheaper chipset. Short-term profitability would be enhanced by cost cutting. Grove ruled that the long term interests of the company remained moving advance technology into the market place, and that we would forgo short term returns for the long term benefits."

Chipsets

Chipsets are an important part of the motherboard. Chipsets are used in conjunction with the microprocessor and implement the PC bus, which provides the interface between the microprocessor and the rest of the system. During the 1990 Strategic Long Range Planning (SLRP) meetings, Intel managers discussed the potential problem associated with the system architectures that Intel's customers were using. While the performance of Intel processors had advanced at a steady and rapid pace, the PC bus had not changed in more than a decade. One manager explained,

"It was like we were putting a Ferrari engine in a Volkswagen. We realized that we had to take more of a leadership position in the industry. Originally, IBM and other OEMs had set system designs standards. With OEMs under extreme pressure to cut costs, most aren't thinking about this -- they are focused on executing today to survive. Only a handful of OEMs have the ability to do research on next generation system architectures."

Intel Architecture Labs (IAL) introduced the PCI (Peripheral Components Interconnect) bus in the early 1990s along with the Pentium processor. The PCI bus greatly improved bus speeds and overall system performance. One technical manager explained,

"We developed PCI technology, and are giving it to the industry. While PCI as an architecture does not directly generate significant revenue for Intel, it is an important enabler for Intel's next generation high-performance CPUs."

Networking

Intel's networking business got a jump in the early 1990s when a new manager wrote a business plan for selling network adapter cards. 3Com was selling adapter cards for about $400 at the time, and this manager believed that Intel could build them faster and cheaper. Intel had success with its early network adapter products, and quickly captured about 8% market segment share. 3Com responded aggressively and held Intel at 8% share.

Intel's networking group took advantage of a market transition and developed a strategy that took advantage of its strengths in silicon technology and OEM channels. 3Com sold most of its network adapters on fiberglass cards through reseller channels. Intel worked to integrate networking functionality into the PC platform, and then drove design wins and sales through OEM channels. Mark Christensen, the head of Intel's networking business, explained,

"In 1993 we were not winning the market segment share battle with 3Com, and we saw that we needed to change the ballgame to make significant progress. The best time to do it was during a market transition. The Fast Ethernet project started around this time and allowed dual 10 MB/100 MB performance. We were first to market in 1993, and were consistently 6-9 months ahead of the competition. Our product went from 15 chips to 3 chips to 2 chips to one chip, and in the process costs dropped rapidly. The pricing went from $300 in September 1993 to $179 a year later, then to $149 and then to $89. This rocked 3Com."

In spite of its successes, Intel's networking business was not considered strategic during this period and the business was forced to make trade-offs that favored short term profitability over long term strategic considerations in order to justify its continued survival. According to Mark Christensen,

"For the first 6 years, from 1991 to 1997, it was basically pay your own way for growth. If you didnít grow, you had the threat of getting downsized. Much of the funding was being funneled into programs that would help microprocessor growth (Job1)"

In addition, the group was prevented from making acquisitions in large part because acquisitions would dilute Intel's price/earnings ratio. According to Mark Christensen,

"We could have bought [a number of companies] in the early 1990s and turned them into multi-billion dollar networking businesses had we had a bigger vision earlier. This shows that we were not thinking like a networking company at that time."

Coprocessors, Upgrade Processors, & Boxed Processors

In the late 1980s, Intel built a significant business selling math coprocessors, which provided floating point math capabilities for the 286 and 386 processors. In 1991, this product was part of the End-User Component Division, whose manager saw that the functionality of the math coprocessor would be integrated into the core processor beginning with the 486 generation. The division therefore adapted its product strategy, and it proposed the creation of a business around the sale of microprocessor upgrades. The proposal went through Intel's traditional POR process successfully, and the business came to be known as OverDrive. Explained Tom Lacey, later a senior manager in the division and still later a vice president in Intel's sales and marketing group:

"The funding argument began with the growth of the PC's installed base. A certain percentage of the installed base would upgrade, and by making different assumptions about the percent that upgrade, we generated ROI calculations. There's also about a four to five year depreciation cycle for PCs in businesses, and yet we rev the microprocessor every 18 months or so. Thus, the idea with the OverDrive® processor is that for a few hundred dollars, you can buy the latest technology and put it in your couple year old computer. It was a compelling business model -- it amounted to many hundreds of millions of dollars. We also asked the question, what if new PC sales growth slows? When that happens, we'll have an installed base of upgradeable PCs. Wouldn't that be a great business. It was a nice insurance policy against dips in our mainstream business."

The OverDrive processor business became quite successful, but not just because of upgrade sales. Jim Yasso, who became general manager of the OverDrive division, explained,

"We saw that there were lots of "upgrade" sales in Asia. But the business was too large to be just upgrading. We found out that many small integrators were buying OverDrive processors and using them as CPUs. An OverDrive processor offered several key features for integrators. It was offered as a single unit, it came in a retail box from Intel, it had a heat sink attached, and it came with a good warranty. Thus, it lended itself well to this channel, and was perceived by end user customers as being 'genuine Intel'."

The OverDrive processor business eventually became impractical and the group transformed itself once again. Tom Lacey explained,

"The idea behind the OverDrive processor was that a consumer could plug a new engine into his or her computer and get at least 50% more performance as a result. As our microprocessors became more and more powerful, it became increasingly more difficult to achieve this 50% performance boost, and the investments needed to realize this performance gain also increased. So the OverDrive business was transformed into the retail microprocessor business that Intel now runs. Rather than being an upgrade microprocessor, we now provide a boxed microprocessor."

The original proposal for the boxed processor business was made in mid-1994 and approved later that year. This group came to be known as the Reseller Product Division (RPD). Jim Yasso, vice president and general manager of RPD, explained,

"Our mission was, as the OverDrive processor business declined and the [boxed processor] business grew, to change our focus from end-users to resellers and grow the box business, the brand, and the channel... We had an ongoing business with OverDrive processors that was profitable, which allowed us to shift resources from one place to another... When we got started, my boss was a strong supporter, and we also had support from Andy [Grove] and Craig [Barrett]. The support from the top was important to our success... The critical things to make a new business successful [at Intel] are having top down support, and having adequate funding."

RPD sold boxed microprocessors through distributors and resellers to small integrators. By 1999, RPD worked with over 40,000 dealers around the world, and made up a significant percentage of Intel's total microprocessor business in terms of unit volume, particularly for desktop PCs.

ProShare

In the early 1990s, Intel viewed videoconferencing on the PC as a major business opportunity, as well as an important horizontal application that would consume considerable microprocessor cycles. Andy Grove was a major proponent of Intel's efforts in this area. In January 1994, Intel introduced ProShare® software, a desktop video-conferencing system that was designed to work with ISDN networks. ProShare utilized many of Intel's own technologies, including the video compression chip. Rather than participating in the market just as a component supplier, Intel would compete as a systems player. Andy Grove explained,

"Twice before (with the PC and network cards) we sat by and played the components supplier as a system business grew haphazardly. These things aren't an Erector set that jumps together on its own. Somebody has to package our technologies and sell them. This time we're going to go out and help create the market, build the system as well as the components, and compete in the market place."

Asked whether market research had convinced Intel that there was a huge potential market for desktop videoconferencing systems, Andy Grove said,

"Who the hell are you going to ask? This is a brand new market. There's just something about live video on a computer screen. I can't walk away from it."

ProShare became one of the largest commitments ever made by Intel in a non-microprocessor area and the effort eventually included 700 technical people. Intel invested considerable marketing resources behind ProShare technology, and in 1994 the marketing budget for ProShare was second only to that of the Intel Inside® program. Pat Gelsinger, who had been in charge of Intel's P6 processor development after leading 486 development, was put in charge of ProShare. Grove noted,

"We can't expect to succeed if we aren't willing to put our best people on the project."

Intel priced the ProShare systems aggressively relative to competitive products, but the market for the product did not materialize as expected. After 5 years and about $750 million in investments, the effort was scaled back dramatically. Discussing the learnings from the experience with his class at Stanford Business School, Andy Grove said,

"We assumed that just because it could be done technically there would be high demand. I was an enthusiastic user and supporter, but I've stopped using it. The novelty wore off. It was difficult to set-up and so forth, so the benefits were outweighed by the costs. Videoconferencing doesn't want to break out of its vertical orientation... If the application had resonated in the market, we would have been heroes. We did all the things required for strategic leadership, it's just that we were wrong. If we were to do it over again, our approach would be not so much like the Normandy invasion, but more of a vertical market focus... We brought a style and conceptual approach to an area where it did not work."

Hood River

Intel's Hood River project was a seed program that had been initiated in early 1996 with the objective of penetrating the family room with the personal computer. The project had been driven in part by the Intel motto, "The PC is IT". Hood River was positioned at the high end of the family room PC market, selling for $2,500 or higher, without a monitor. The Hood River design called for the use of Intelís 233 MHz Pentium II processor, their highest performance CPU at the time, and integrated a wide spectrum of applications, including communications, consumer electronics, and entertainment applications. The Hood River team at Intel had engaged a number of potential OEM customers, including Sony, Matsushita, and Philips. However, in early 1997, none of these companies had committed to making volume shipments based on the Hood River design. Compaq and Gateway 2000 were also developing PC entertainment systems for the family room, and had collaborated with Intel in this area (though neither used Hood River systems).

By early 1997, the Hood River team believed that they had made significant progress on the design. There were, however, several weaknesses in the first generation product. Hood River had been dependent on Microsoft for several important features, including "Instant On", which were delayed along with Memphis, Microsoft's successor to Windows 95. The system also had shortcomings in terms of its user interface, which was much more like a PC than a TV in terms of its simplicity.

Hood River reported into the Desktop Products Group (DPG), which was led by Mike Aymar in 1997, who said,

"Originally we expected the [Hood River] project to go smoothly and to generate demand for another 1 million PCs per year. But market projections were for various vendors worldwide to ship only in the tens of thousands of units in '97 and '98. Suppliers could be counted on a single hand. Several OEMs seemed interested, but more in terms of keeping engaged with the experimentation. This was insufficient. So, thus far we had failed the market test."

As a result, Aymar decided to cut the funding for Hood River.

Despite its cancellation, Hood River provided Intel with insight and information about the converging computer, communications, and consumer electronics industries. According to Ron Whittier, senior vice president and general manager of Intel's Content Group,

"With Hood River, we explicitly articulated for the first time that content would move into a different room. With this awareness, new issues came up and usability became very salient... We became more cognizant of the issues that might arise in different areas. This is consistent with the mass segmentation of the PC into every niche in the market. Intel is now looking at computers in cars, in each room in the house, and other places. Hood River was the leading edge effort in this direction. From a monolithic market, with 100 million units of the same thing, we move to a segmented market. The content also segments, so now we need to make sure content plays in each segment and environment."

In the aftermath of Hood River, Intel continued to pursue the transition of the PC into new market segments and to refine its strategy for the family room.

Looking Ahead In 1997

During its second epoch, Intel became one of the most successful and profitable companies in the world. Moreover, the strategic actions of the company during this period reinforced the strength of the company's position in the existing industry structure. This success, however, had not come without costs. In 1989, Gordon Moore, then Intel Chairman, said,

"We can do variations on present businesses very well, but doing something new is more difficult. Today, the likelihood of someone killing an effort like the one of Dov Frohman (inventor of the EPROM) is very high, because you need a well-defined application to a market from the outset. This is especially so because we are not looking for additional opportunities. There is still a lot of evolution left in the current technology. If you consider the possibilities with reducing line-width, you can see another twelve years of evolution along the same curve."

By 1997, Intel was dominated by the microprocessor business, and the company's focus on this business was singular. Despite the emergence of many new opportunities across the computing industry, and the emergence of many new ideas within the company, Intel had had difficulty turning these opportunities into successful stand-alone businesses. According to Dennis Carter,

"It was difficult for new ideas to evolve because of the internal infrastructure, not because of a lack of new opportunities. The business climate actually created many opportunities, but the question was how to pursue and exploit those opportunities. Andy tried to get Intel into other spaces, such as videoconferencing. But none ever quite jelled. None of the options that looked good 4 or 5 years ago have come to pass... Les Vadasz through his CBD group knows every trend that's happening in the industry, but there is no mechanism for turning this knowledge into products. There is no serendipity."

Dennis Carter discussed another issue:

"In the 1990s I remember being tired constantly. There was no respite. After every success, Andy would say 'that's nice, but what are we going to do next.' For example, when the Intel stock reached $100 for the first time, we were in a meeting, and upon hearing the news, people were excited and started to cheer. But Andy looked very glum. He pounded the table with his fist and said, 'the shareholders that bought our stock at $100 expect us to bring it to $200.' This created a sense of urgency, and I went from elation to almost a sense of panic. We have been successful largely due to efforts -- extraordinary efforts. And we've been driven. But this is very tiring, even exhausting. So it is not sustainable."

 

Epoch III: Intel the Internet Building Block Company, 1998 and Beyond

"The industry that we're in has changed. We made a transition from the 70s to the 80s from being an integrated circuit company to being a microprocessor company. Now we're transitioning to an Internet building block company. If you look at our mission statements, you'll see that they reflect this transition from being a microprocessor company to being an Internet building block company. Every third word makes a reference to the Internet."

--Craig Barrett, CEO, April 1999

"Intel's goal is to be the leading supplier of building blocks to the connected computing industry worldwide."

--Intel 10-Q Report, May 1999

In 1999, Intel faced two major cross-currents: the company's core business was becoming more segmented and its growth appeared to be slowing; and the Internet, rather than the PC, had become the driving force in high technology. Intel was still the leading player in the microprocessor industry, and one of the most influential companies in the computer industry. However, there were worrisome signs even in the core business. Some analysts projected that growth in the PC industry was slowing. Additionally, a new class of low-cost PCs with less powerful processors was growing in popularity. Furthermore, there was a widely held belief frequently reported by the press that processor power was not important for Internet browsing. The combination of these events could add up to a significant shift in the consumption of high-end processors. Intel top management questioned whether the company could sustain its record of outstanding revenue and profit growth without diversifying and developing new businesses.

Intel continued to aggressively pursue opportunities in the microprocessor business. However, Craig Barrett had strongly urged Intel's senior managers to think of their businesses in the context of the emerging Internet economy rather than the PC industry. At a meeting of Intel's top management, Barrett described Intel's strategy to become the building block supplier to the Internet economy by focusing on four areas: client platforms, networking infrastructure, server platforms, and solutions and services. Barrett said,

"The world changes and the center of gravity shifts. We need to shift with it. We want to be the center of gravity in each of these four areas."

Of these four areas, Intel's core business already addressed client and server platforms. Additionally, by 1999, Intel's networking business had become much more prominent both from a financial and strategic perspective. And to address Internet solutions and services, Barrett had organized a new business group that would focus entirely on developing new businesses. According to Andy Grove,

"Intel is undergoing an adaptation today that is a different version of the adaptation that occurred in the mid-80s. And it's almost as dramatic. The adaptation today is to the connected computer universe... It remains to be seen how well our senior management is tuned to that. I have some concerns. We have a number of people that were very competent in the 2nd epoch of Intel. Right now the actions are not convincingly in tune with the new epoch."

Organization

In early 1998, Craig Barrett was named CEO of Intel and Andy Grove became Chairman. In 1999, Intel was organized in a matrix structure that consisted of both product and functional groups. The product groups included: the Intel Architecture Business Group, which had responsibility for Intel's industry leading microprocessor franchise; the Computing Enhancement Group, which produced chipsets, StrongARM processors, and flash memory products; the Network Communication Group, which developed products that provided network and Internet connectivity; and the New Business Group, which focused on growing opportunities in new and emerging market segments. The functional groups included Sales & Marketing Group (SMG), Corporate Business Development (CBD), Technology & Manufacturing Group (TMG), Microprocessor Products Group (MPG), the finance group, and the legal group. Exhibit 11 shows Intel's organization chart and Exhibit 12 list Intel's officers. Intel's management ranks were deep, and 29 of the top 30 officers had been with the company for over 15 years.

In the third epoch, Intel's finance group continued to be a very strong part of the company's organization. The finance group was organized to mirror the operating groups, with most of the general managers or division heads having a dedicated controller as a counterpart. All finance personnel have a solid reporting line to Andy Bryant, CFO, and a dotted line to their respective operating groups. The finance group balanced the interests of shareholders, finance, and the operating groups. According to Andy Bryant,

"I tell the finance people that their first job is to look after shareholders, the second job is to look after finance, and their third job is to look after the operations. Their proximity to operations means that 80% of the influence comes from the operations. I worry about the proximity overweighing their perspective... They should complement the skills of the GM, and... play a broker role. They have to be willing to tell me that I'm wrong, or that the GM is wrong. Always, they must look at what creates the best cash return to the company."

Key Challenges in the Core Business, 1998 & Beyond

In 1998, Intel's revenue grew 5% to $26.3 billion (after 20% growth in 1997), while net income fell 13% to $6.1 billion (see Exhibits 5-6). Intel's gross margin percentage fell in '98 to 54%, from 60% in '97 and 56% in '96. Intel operations generated $9.2 billion in cash during 1998. In late 1998, Intel's market capitalization exceeded $200 billion for the first time (see Exhibit 9). According to Forbes, 80% of Intel's revenue, and all of its profits, come from microprocessors (see Exhibit 13). In spite of its extremely strong financial condition, Intel, as in earlier epochs, faced significant challenges.

The PC Market Segment

By 1998, low cost PCs had captured a significant percentage of consumer sales within the highly visible retail channel (see Exhibit 14). Many of these machines initially used low-cost microprocessors from AMD and Cyrix. As a result, AMD's market segment share of Intel-based microprocessors grew from 6.7% to 16.1% between the fourth quarters of 1997 and 1998, while Cyrix's grew from 3.5% to 5.8% over the same period. Intel's overall market segment share dropped from 86.2% to 76.1% during this period. On the other hand, PC manufacturers such as Dell and Gateway who sold direct via the phone or the Internet continued to focus on high-end PC sales and were gaining market segment share away from the retail channel. Nonetheless, financial analysts reported that the growth of the low-end market was putting tremendous downward pressure on the average price of desktop microprocessors. Speaking to his class at Stanford Business School, Andy Grove said,

"The 2nd law of thermodynamics, applied to the computer industry, is that everything gets commoditized eventually. Grove's Law is that the last one to get commoditized wins."

Many analysts were predicting that the PC was entering its twilight years. According to Paul Horn, senior VP of IBM Research,

"After more than 15 years as the center of the computing universe, the PC is about to give way to a new breed of... [specialized] devices that will dramatically change the way people communicate and share information."

Other industry analysts expected annual PC sales to peak at around 100 million units, only slightly higher than 1999 levels. Although it was a widely held belief that appliances would eventually replace PCs in at least some applications, there was not yet evidence of this happening.

The Internet

The Internet was causing revolutionary changes in the computing industry, and these changes presented both threats and opportunities to Intel. In 1999, the primary driver of PC sales was the demand for Internet connectivity. It was widely held that processor power was not important to a user's Internet experience, and this trend threatened to devalue desktop processing power. But there was a potential opportunity for Intel as well. The Internet caused more processing power to be needed in high-end servers that distributed content and processed online transactions, and this was a potential growth market for high-end Intel processors. According to Renee James, former technical assistant to Andy Grove,

"We have struggled for the last 2 years with how to make an impact on the Internet... This is one of Intel's biggest long-term strategic challenges. There has been a view that the Internet is happening without us."

Strategy & Action in the Core Business, 1998 & Beyond

Intel's core business consisted of both client platforms and server platforms. Intel's long-term strategy was to introduce ever higher performance microprocessors tailored for the different segments of the computing market. Moore's Law predicted that processor power would increase geometrically over time. This process had in the past resulted in tremendous improvements in the value delivered to PC users. For example, a Pentium® III processor based system selling for about $1500 in early 1999 had roughly 10 times the performance of a comparably priced Pentium processor based PC circa 1994. That rapid increase in value and performance had made it possible to offer tailored processors for different segments of the computing market.

There were many key elements to Intel's strategy in its core business:

Market Segmentation Strategy

In response to the changes in the computing industry, Intel reorganized its microprocessor line with new products and brands in each of the major market segments. The Intel® Celeron™ processor was targeted at the low-cost, entry-level part of the market, the Pentium II and Pentium III processors were targeted at performance desktops and entry-level servers and workstations, and the Pentium II Xeon™ and Pentium III Xeon processors were targeted at mid-range and high-end servers and workstations. All of these processors were derivatives of Intel's P6 microarchitecture. This segmentation allowed Intel to maintain a huge pricing range, with Celeron processors selling for about $63 each and high-end Pentium Xeon processors selling for about $3700 each. This strategy had helped sustain Intel's historical average sales price for microprocessors (see Exhibit 15). Intel's product breadth made it more difficult for competitors to pursue niche strategies by exploiting holes in Intel's product line. Exhibit 16 shows projections for IA microprocessor unit growth and market segment share for desk-bound PCs, mobile PCs, servers and workstations, and upgrade processors.

Client Platform Products. The Celeron microprocessor, optimized for the low end of the market, was introduced in April 1998. The Celeron processor design was optimized to reduce manufacturing costs, whereas the Pentium processors were optimized for performance. The Celeron chip was the result of Andy Grove's recognition, alarmed by the idea of "disruptive technologies," that the low-end market had become a major potential threat to Intel's strategic position. Intel reassigned a number of engineers in 1997 for a crash effort to accelerate Celeron's development. Celeron was priced far more aggressively than the Pentium processors, and it had slowed Intel's market segment share losses on the low end.

Intel's Pentium II and Pentium III microprocessors remained the core of Intel's business. The Pentium II was Intel's largest volume microprocessor in 1998. In February 1999, Intel launched the Pentium III microprocessor, which the company positioned as being ideal for Internet users. As part of the launch, Intel offered owners of Pentium III processor based systems a free online service, called Intel® WebOutfitter(sm) Service, that offered plug-ins that were optimized to provide improved Internet performance.

Intel developed specialized versions of its microprocessors that were targeted at the mobile computing market segment. These processors were modified versions of the Celeron, Pentium II, and Pentium III processors, but had lower power consumption and size requirements compared to the standard processors.

The market for digital appliances, such as cell phones, hand held computers, cable set top boxes, and automobile PCs was rapidly taking shape in early 1999. These products were relatively inexpensive, and generally required processors that cost around $20 or $30. Intel competed in these markets with both Pentium and StrongArm processors. The Pentium processors targeted at these segments were generally several generations old. StrongArm, acquired as part of a legal settlement with Digital Equipment, was noted for its high performance, its low power consumption, and -- with a price between $21 and $33 in volume -- its low price. These characteristics made StrongArm well suited for many applications within this market segment.

Some analysts were skeptical about Intel's resolve to compete in these markets. Craig Barrett admitted,

"Over the next three to five years our main focus will be on traditional computing devices, including more powerful workstations and servers. There may be some overlap, but we still think the main processor market will be for a computer that has the ability to do a wide variety of tasks."

Speaking of the appliance market, Les Vadasz said,

"We'll compete, but this is not a major growth area for Intel. There's not that much silicon value and no segment will match the PC in volume. The appliances market will consist of many smaller categories rather than singular winners."

Server Platform Products. On the high end, Intel was increasingly targeting the computers that ran corporate data centers and the World Wide Web. The company believed it could establish a standard processor design in this segment, much like it had years earlier in the market for PCs, and more recently in the market for workstations and entry-level servers. The Pentium II Xeon microprocessors targeted at this market segment offered improvements in performance, reliability, and security, and were designed for use in multi-processor systems. In 1999, 50% of Intel's development dollars were being spent on servers and workstations, despite the revenues in these segments being much less than 50%.

Product & Technology Development

Technology leadership had been central to Intel's strategy throughout its history. According to Andy Grove, "Intel has pursued a technology driven strategy, which we've executed relentlessly." In 1998, Intel spent $2.7 billion for R&D, up from $2.3 billion in '97 and $1.8 billion in '96 (see Exhibit 7). Intel's R&D spending allowed it to design multiple generations of new microprocessors concurrently. In 1998 and 1999, Intel continued its tremendous pace of product development (see Exhibit 10). In 1999, Gordon Moore spoke again about the future applicability of Moore's Law and the industry's ability to continue making technology advancements and line-width reductions:

"We can still see 12 years down the road. I can generally see 3 to 4 generations ahead. Beyond that, it always looks like we will hit a wall, but we've been fortunate so far. As we move along, because of the physics of semiconductors, we will eventually hit one. Matter is made of atoms, and eventually we'll reach the limit of what the technology can do. It's amazing that the technology has been able to go this far... I think we'll move on more or less the same curve until about 2010 or 2020. Then we'll be approaching the limits of the atomic nature of matter. It will take a tremendous investment in technology to continue the advance."

Intel had also strengthened the quality of its product designs as a result of the highly publicized Pentium processor flaw issue of 1994.

Intel increasingly incorporated greater amounts of functionality into its microprocessors, allowing it to add more value and capture a higher percentage of industry rents. Gordon Moore addressed this issue in general:

"One of the really amazing things about this industry is that we assimilate the value add of our customers and give it back to them for free. The ones that recognize this force do well. The ones that don't, and try to protect their little niche, get steamrolled. The technology is phenomenal in terms of the economics. It commoditizes everything."

In February 1999, AMD introduced its K6-III processor, a rival to the Pentium III processor. AMD had dramatically reduced its time to market disadvantage -- in earlier years it had taken AMD as long as 4 years to match Intel's new product releases. Even so, AMD had often come up with microprocessor designs that theoretically matched Intel's performance, only to fall short in the transition from the lab to the factory floor.

Intel had partnered with HP to develop the technology for 64-bit microprocessors. The first of the IA-64 product family, code-named Merced, was expected to be available in sample volumes in late 1999 and initial production volumes in mid-2000. In addition to running Microsoft's Windows NT, the Merced processor would also run most of the major versions of Unix, including HP-UX, Solaris, AIX, and Linux.

Intel faced challenges in penetrating the high end market segment, and Merced had been delayed several times. According to Paul Otellini,

"Overreach is our biggest challenge. We were working on the IA-64, the Xeon processor, and 8-way systems, but were running behind schedule. We're moving this forward, but we've encountered technical challenges. This is truly rocket science, and there are very few companies who do this well. No one has successfully commoditized the high end."

Microsoft had similar ambitions to Intel with its Windows 2000 operating system, but it too had been delayed by more than a year. As a result of the delays with Merced, HP and SGI had both decided to extend the lives of their own chip architectures.

Manufacturing & Production

Intel was renowned for its ability to manufacture its chip designs, and the company had developed the practice of optimizing the manufacturing process for a given chip, and then rolling out that process to Intel's other fabs in a process called "Copy Exact." In addition, Intel had closely integrated product design with process development. According to Albert Yu, senior vice president of Microprocessor Products Group,

"Having in-house manufacturing is very important. Design engineers in MPG work very closely with TMG (Technology and Manufacturing Group) to incorporate the design and manufacturing process... This is a key element to success - the heart of the business. This is important to our ability to get the most into the smallest chip. If you look at our competition, the ones with their own manufacturing have chips that are compact and well designed. Competitors who don't (have their own manufacturing) have bigger chips."

Gordon Moore spoke of the importance of manufacturing and process technologies:

"One of the amazing things about our industry is that the next generation of technology is always much more cost effective than previous generations. Making things smaller makes everything else better simultaneously. That's a pretty unique characteristic with important implications. If you stop spending during a slow down, it's a big drawback if you get behind. We have a saying: "you never get well on your old products." We've been able to continue the march forward. Other companies have been in the same vein of technology, but nowhere near as successful. Over the years, we have made more profits than the whole industry combined."

Intel competitor AMD had traditionally had great difficulty in transferring its processor designs from the lab to the factory floor. On the other hand, IBM's manufacturing capabilities were generally considered world-class and IBM had recently announced several developments that put it at the forefront of manufacturing technology.

Intel continued to invest heavily in state-of-the-art manufacturing facilities, spending $4 billion on capital additions in 1998, compared to $4.5 billion in '97 and $3.0 billion in '96 (see Exhibit 8). Most chip companies shopped prototype designs to potential customers before committing to build, and many others outsourced manufacturing rather than build their own fabs. However, according to one analyst,

"Only Intel builds [manufacturing] capacity before it creates demand."

Complementors

Though the so-called "Wintel" alliance remained strong, Intel's relationship with Microsoft had changed significantly, particularly on the low-end and the high-end. On the low-end, Microsoft had released Windows CE which targeted handheld PCs and other information appliances. Unlike other versions of Windows which were closely integrated with Intel processors, Microsoft initially ported Windows CE to several non-Intel processors. On the high end, Intel's Merced and future IA-64 processors were designed to work not only with Microsoft Windows NT, but also with each of the major versions of Unix.

Intel continued to work actively with development partners, providing them early access to new technologies so that Intel's new products would have strong applications that supported them at launch.

Corporate Business Development (CBD) performed an important function within Intel by doing acquisitions or making equity investments in strategic companies. Since 1996, Intel had invested about $2.5 billion into 200 companies, making it one of the largest venture firms in the US. Some notable success stories included Inktomi, Broadcom, eToys, and Broadcast.com. By 1999, CBD was making significant investments in Asia, Europe, and Latin America to support Intel's international growth objectives. In May 1999, Intel established a $250 million venture fund targeted at hardware and software companies that supported Merced. Speaking of CBD's role, Les Vadasz, senior vice president and director of CBD commented,

"CBD seeks to accelerate the creation of new market ecosystems, in part by using our financial resources... [For example], we've made a number of investments in hardware and software companies to accelerate their plans to adapt their products to the new Intel platforms."

Relationship with OEM Manufacturers

Through its brand strength, its influence over design standards, and its vertical integration into chipsets and motherboards, Intel became far more central to the PC industry. Some analysts likened the OEMs more to distributors than technology companies. According to Tom Yuen, cofounder of PC maker AST Research,

"You no longer buy a Compaq computer, you buy an Intel computer from Compaq."

According to Paul Otellini,

"In a sense, PCs are now commodities, because most of the intellectual property is now tied up in the microprocessor."

Since the early 1990s, Intel had been a major player in the market for chipsets and motherboards. This allowed Intel to support OEM manufacturers who were less competitive in these areas, improving their time to market, reducing their R&D requirements, and allowing them to focus resources in other areas. This leveled the playing field among OEM manufacturers, it reduced the value-add provided by OEMs, and it offset some of the power and competitive advantage of the major OEM manufacturers.

Intel's ability to influence design standards impacted the balance of power in the industry. Case in point: Intel's entry into the server market segment. In 1996, Compaq earned 27% of its sales -- and all of its profits -- from server sales, based on the success of their unique system designs. These systems cost between $10,000 and $20,000, and add-on processor cards cost about $16,000 more. When Intel entered the server market segment with proposals for comprehensive system design standards, the playing field quickly leveled. By 1999, dozens of OEMs using Intel technology offered multi-processor servers between $5000 and $10,000, with add-on processor cards at about $2500.

Sales & Marketing

International markets had relatively low PC penetration and were viewed by Intel as an area of major potential growth. In 1998, more than 50% of Intel sales were outside the US. Craig Barrett had focused significant effort on growing international sales, opening new sales offices in a number of countries.

Intel's technology leadership and time-to-market advantage allowed the company to recoup its R&D investments. Intel's practice was to aggressively reduce prices over time, especially as competitive products were introduced, which allowed the company to maintain market segment share and stimulate market growth. This pricing strategy also put enormous pressure on Intel competitors. Exhibit 17 shows Intel's microprocessor pricing for several products over time.

Intel's outbound marketing efforts had become a key driver of its microprocessor business. Since launching the Intel Inside campaign in 1991, Intel had invested heavily in marketing, advertising, and promotions. For the Pentium III processor launch in early 1999, Intel spent about $300 million in advertising, more than double its investment on any previous chip launch.

Competition

In early 1999, Intel's primary competitors in the desktop microprocessor market segment were AMD and Cyrix. Other competitors included Sun (Sparc), Compaq (Alpha), and Motorola (PowerPC).

AMD. Between 1997 and early 1999, AMD had design wins with HP, IBM, Compaq, Gateway, Sony, Toshiba, and others. In January 1999, for the first time ever, AMD chips outsold Intel chips in the US retail desktop PC market segment, with a 43.9% unit market segment share, based on strength in the sub-$1000 market segment. However, AMD suffered significant operating losses in 1996, 1997, and 1998 ($253 million, $91 million, and 164 million, respectively), as well as in the first quarter of 1999 ($128 million). In June 1999, AMD announced second quarter operating losses of $173 million.

Cyrix. Cyrix helped launch the sub-$1000 PC market segment in early 1997, and in July 1997, National Semiconductor acquired Cyrix. In May 1999, after making significant market share gains in 1998 and early 1999 but also suffering steep operating losses, National announced that it would exit the PC chip business. In June 1999, National announced that Cyrix would be acquired by Via Technologies, a Taiwanese developer of chipsets that already had a clone of Intel's Celeron processor in development.

RISC Competition. In the high end of the microprocessor market segment, Intel faced stiff competition from the established RISC/UNIX systems from Sun, IBM, HP, SGI, and others. Despite gains by Windows NT and Intel systems in the low end of the workstation and server markets, these RISC/UNIX systems continued to hold market share on the high end. In particular, Sun Microsystems was perceived by many as the leader in the Internet server segment. However, according to Tom Lacey,

"Sun has done a great job of marketing and hyping their association with the Internet. McNealy says that "we're the dot in dot.com". But there's a big difference between perception and reality."

Strategy & Action Outside the Core Business, 1998 & Beyond

The Creosote Bush

Outside the core business, Intel focused on developing business around both networking products and Internet services and solutions. Intel's singular focus on the microprocessor had been a critical factor in the company' success during the second epoch. However, Craig Barrett did not believe that the company's historical growth rates and profitability could be sustained solely with microprocessors. According to Barrett,

"Microprocessors by themselves will not be the growth engine that they've been in the past."

During the second epoch, Intel's business focus on microprocessors had made it difficult for new ventures to thrive inside Intel. Craig Barrett likened the company's microprocessor business to the creosote bush, a desert plant that poisons the ground around it, preventing other plants from growing nearby. Although much of Intel's R&D investments went into technologies that complemented the processor and thereby offered opportunities to launch other new business units, the company rarely attempted to do so. Even when it did, the creosote effect made it difficult for the new business to thrive. The reason for this was simple. Any technology advance that enriched the PC environment was likely to create more demand for microprocessors. Thus, it was generally more valuable for Intel to give away technology and quickly disseminate it in the market, rather than try to build a business around it.

Faced with the changes being caused by the Internet and what appeared to be a new epoch, Barrett began actively pushing Intel to diversify into new businesses. In 1999, the most prominent manifestations of this diversification strategy were the Network Computing Group (NCG) and the New Business Group (NBG).

The Network Computing Group (NCG)

During Epoch II, Intel's networking group struggled to get resources, was not considered strategic by senior management, and was prevented from making acquisitions. As a result, the networking business had not grown as rapidly as it might have. Despite its slow start, by 1999, NCG had become a sizable and strategic business for Intel, with sales of approximately $1.2 billion. Since 1991, Intel's networking business had grown 58% per year and was expected to grow another 75% in 1999. NCG consisted of 4 distinct businesses: home networking, network adapters, small to medium sized systems for business, and communications silicon. According to Mark Christensen, vice president and general manager of NCG,

"Looking forward, every segment that we play in is different. We have different competitors, different channels, different customers, etc. We're taking a targeted, segment by segment approach, and we've developed a unique strategy for each of these segments."

Christensen continued:

"Craig came up with the creosote bush image to illustrate to the company the challenge of trying to grow new businesses in one of the most highly successful companies of all time. He encouraged people to develop new businesses rather than just supporting "Job 1". The creosote bush was deep here..., [but] we've gotten lots of support from Craig and have now made significant progress in growing several new areas."

Craig Barrett discussed the charter of NCG:

"The charter of Christensen is to be the silicon building block supplier, not just to the LAN, but also to the WAN, and to also be a network building block supplier to the home and small business. This ties into the vision of a billion connected computers. We want to have the hearts and brains of the clients, servers, and some of the networking components. We're not into competing with the Ciscos and Nortels... We want to supply the Nortels, Ciscos, Lucents, etc. with building blocks."

Home Networking. In 1999, the home networking market segment was still developing, but Intel expected it to become a $1 billion market within a few years. Intel had recently introduced a product called the Any Point™ Home Network that allowed consumers to use the phone line in their home to connect computers and printers together, and to provide Internet access to any PC in the house through a single ISP account. The Any Point product combined technology developed in the Intel Architecture Lab (IAL) with technology obtained in the acquisition of Digital Semiconductor. 3Com was Intel's primary competitor in this segment.

Network Adapters. Intel had become a major player in the market segment for network adapters, which had previously been dominated by 3Com. Intel continued its strategy in this segment of leveraging its strengths in silicon and OEM channels, which the company believed positioned it favorably vis-a-vis 3Com. In 1999, Intel expected to ship 25 million units and earn about $500 million in revenue. Intel's market segment share had grown from 8% in 1993 to 40% in 1999. 3Com had about a 40% share as well.

Network Systems for Small and Medium Sized Businesses. The network systems business consisted of hubs, switches, and routers targeted primarily at small businesses. Intel did not consider this segment to be their strongest area, but the business had grown from 0 to $300 million. Intel had made several acquisitions in this area. Intel competed primarily with 3Com and Cisco in this segment.

Communications Silicon. Intel viewed communications silicon as a major strategic growth area. The market for communication chips was growing about twice as fast as the microprocessor market segment. Intel believed that the communications and networking industries would undergo a similar transformation to that undergone previously by the PC industry in transforming from a vertical to a horizontal industry structure. Intel hoped that its chips would become successful in this industry just like they had with PCs. This would not be an easy task. Networking companies understood well the history of the PC industry. In addition, the Intel brand was associated with the computer products, not networking products per se.

Networking chips came from dozens of players, including Lucent Technology, IBM and Broadcom. In March 1999, Intel acquired Level One Technologies, the premier supplier of network interface chips, in a transaction valued at $2.2 billion. Level One's products were used by network equipment makers such as Cisco, Lucent, 3Com, and Hewlett Packard. Broadcom, a Level One rival, said it expected to gain market share as a result of the Intel acquisition. Broadcom believed that network equipment companies, such as Bay Networks, 3Com, and Cisco, would prefer to buy chips for their hubs, switches, and routers from a non-competitor.

Relationship with Cisco. Intel's advances in the networking market had in some ways complicated the company's relationship with Cisco. Intel's approach was to leverage its own strengths in silicon, OEM channels, and consumer brand awareness, and to avoid competing directly in the enterprise market, where Cisco's strengths in direct sales to large accounts and servicing of end users were advantages.

Acquisitions. Acquisitions had become a central aspect of Intel's strategy in the networking business. This was a departure for Intel. According to Mark Christensen,

"Around 1996, we said that we needed to do acquisitions, but this was very painful since the internal view was that any acquisition would fail due to the strong culture of Intel. But, we convinced the board of directors, Andy, and Craig that we could do them successfully. Since then, we have continued to get a lot of support from Craig and Andy and we have now done 7 or 8 acquisitions with very good results thus far. It remains a critical element of our growth strategy."

After the Level One acquisition, Craig Barrett said,

"We're continuing to look for other acquisition candidates. We are deadly serious about our networking communications group."

The New Business Group

Under Craig Barrett's leadership, Intel had significantly expanded its emphasis on building new businesses inside the company. In 1997, Barrett had commissioned a course on "Growing the Business". Each of the top 100 managers in the company attended a week of classes that focused on new business models and developing a common vocabulary to discuss them. Intel's executive staff also was exposed to a condensed version of ideas on how to manage new ventures, as well as how to capitalize on disruptive technologies. As a result of this course, Intel managers began to refer to its core microprocessor business as the "blue" business, and new business ventures as "green" businesses. According to Barrett,

"If Intel wants to continue to occupy a central position [in high tech], it's just not enough to build the hearts and brains of computers."

Barrett organized many of Intel's "green" businesses into a single organization, called the New Business Group (NBG), and he asked Gerry Parker, one of two executive vice presidents at Intel, to lead the group. According to one of Parker's direct reports,

"Gerry is very process oriented, heís very senior, and heís one of the few people that can effectively manage Craig and Andy. Heís working on a process to develop a portfolio of new businesses. Heís combining a formula of getting great people, training them, practicing, etc., and he ends up being a lot like the coach on the sidelines."

Parker recruited senior Intel people from throughout the organization. Enthusiasm to join the new effort was very high, especially from managers who had completed the "Growing the Business" training. Said Parker,

"Intel is moving on Internet time. And we don't think it is much different from chip time."

Intel Data Service (IDS). In early 1999, Intel announced plans to develop a major Internet hosting service. As part of the plan, Intel would host Web applications and data in huge "bit factories" around the world with thousands of servers. Mike Aymar, a 25 year Intel veteran, was chosen to lead IDS, reporting to Gerry Parker. According to Parker,

"Our ideal dream is to host a significant percentage of the world's digital content on our servers, and to be number one or number two in this market."

Renee James, the groups marketing manager, had been instrumental in developing the unit's business plan. As the former technical assistant to Andy Grove, she had been asked to research new business opportunities around the Internet, and to present her findings to the executive team at SLRP. James discussed three different opportunities. Internet hosting was one of them, and it got the go-ahead. Explained James,

"We knew we were never going to be a portal, a content company, or a phone company. But we wanted [the opportunity] to be big enough to capture interest. Some people in IT [at Intel] had similar ideas, and we've retained some of what they had been thinking. They were looking at outsourced IT, but were not necessarily Internet focused."

According to The Wall Street Journal, Intel planned to invest more than $1 billion in IDS, and although Intel senior management was enthusiastic in its support, Intel's control structure created challenges for IDS. For example, while building the first data center, which would cost approximately $250 million, the finance group stopped the purchase orders for the equipment. According to Executive VP Gerry Parker,

"Three levels down finance stopped the PO. They said 'you don't have Andy [Grove]'s approval'. I said, 'who gives a [damn]'. We need to be able to balance moving fast with the discipline of the company. The real issue in an internal bureaucracy is that if it thinks upper management doesn't like it, they will stop it at every level just to ask more questions and look at more alternatives, whereas if Grove says "do it", no one in the bureaucracy dares slow it down. In my case Craig acted hesitant at the wrong moment and it was a license for every level of finance to slow it down for more justification. We are working our way through this but I'm constantly amazed at how difficult it is to do something different in an area where none of us have a good experiential base or good intuition."

Parker continued,

"The finance guys are the hardest, because they think it's their money. I've been given a fair amount of rope on this thing... I'm trying to be fiscally responsible. But with the data center for example, I try to avoid pitching a real detailed plan that locks you in. [I say], 'don't get too enamored with any part of the plan.' "

Intel believed that its competencies in streamlining complex processes, running tight operations, and managing large investments would give it an advantage in this business. Intel planned to refine one data center, and then replicate the data center following Intel's "Copy Exact" process for building chip factories. According to Ellen Hancock, CEO of potential rival Exodus Communications,

"I think it's a stretch for them to say they have some expertise here. We've taken years to set up our operations. I'm befuddled that they think this is like building chip factories."

Digital Imaging. The Digital Imaging division, led by Don Whiteside, combined two previously separate organizations. The first group, based in Oregon, was more consumer marketing and software oriented, whereas the second group, based in Phoenix, was more engineering and hardware oriented. Don Whiteside discussed some of the challenges he had encountered since taking over the group:

"We have a tendency to take a technology driven view rather than a market driven perspective to things... The basic thinking is that we're a technology company, this is technology, so we should be doing it. What we should be asking is what value does Intel provide to the market opportunity."

Much of the funding for the Digital Imaging group came out of discretionary funding, which had a less rigorous allocation process than the formal POR process. Nonetheless, the group had encountered some challenges with funding. According to Lorie Wigle, marketing manager of the group,

"New things kept getting added to Digital Imaging... but we never got more resources, which meant that we needed to cut existing projects... We were one of the first green businesses and there was a lack of clear understanding: do we get new incremental resources or do we have to carve it out of existing budget."

Intel Play. Intel® Play™ toys were the result of a collaborative partnership with Mattel to develop and produce "smart toys" that could be used with the PC. The toys would be manufactured, distributed, and marketed by Mattel under the "Intel Play" brand, and developed in a joint lab in Oregon staffed by both Intel and Mattel engineers and product designers. The underlying technology for the products came from the Intel Architecture Lab. According to Michael Bruck, who co-managed the partnership for Intel,

"The idea behind the project is to take advantage of Intel's strengths in technology and Mattel's strengths in toy design."

The effort came about as a result of joint work being done between Intel's Content Group and Mattel on CD-ROMs. Michael Bruck explained,

"[Mattel] approached us with the idea to apply the Intel brand to electronic learning toys... Jill Barad, Mattel's CEO met with Andy Grove and pitched Mattel and their capabilities... Intel did a lot of research on the opportunity. We did market research to identify how the Intel brand was perceived by parents and their kids... We found that parents recognized the Intel brand, but that there needed to be a connection to the PC to support the brand extension."

Business Communications Product Division (BCPD). BCPD's charter was to design and build PC-based videoconferencing systems for desktops and conference rooms. BCPD was the successor to ProShare, and by early 1999, the group had become profitable, but revenues remained relatively low. The group was leveraging PC technology to increase the accessibility and decrease the costs of video conferencing.

New Business Investments (NBI). NBI was established in late 1998 with the charter of creating new businesses within Intel. NBI played a role similar to that of a venture capitalist, and focused on capitalizing on ideas that were emerging within Intel. NBI worked closely with CBD, but had different goals and objectives. While CBD emphasized external investments and acquisitions, NBI was more internally focused. NBI invested relatively small amounts of money in a venture's early stages of development (as in a seed or 1st round), whereas CBD invested relatively larger amounts in later stages of development. This also implied that NBI would take a relatively active management and advisory role with its investments compared to CBD.

Intel Service Operation (ISO). Intel Service Operation was run by Jim Johnson, an executive with a long history of entrepreneurial initiatives at Intel. ISO provided PC help services to small businesses and consumers. Callers were serviced by live technicians who offered technical support for both hardware and software, including all of the most popular desktop applications. ISO reported a 96% customer satisfaction rating. However, very few of the customers who had subscribed to the service actually used it. The value of the service seemed to be viewed by the consumer as more of an insurance policy. Johnson believed that the service would not be successful without significant investments to promote the service and increase customer awareness.

Intel Architecture Lab (IAL). For most of the 1990s, IAL' s primary role was to support Intel's microprocessor business, and many important developments came out of the lab, including chipsets, motherboards, software, and many of the company's networking products. In 1998, the charter of the group expanded to include the goal of generating new businesses for Intel. For example, IAL engineers contributed to the development of the electronic toys made in partnership with Mattel.

Pandesic. Established in August 1997, Pandesic was a 50/50 joint venture between Intel and SAP, based in Sunnyvale, California. Pandesic had developed an electronic commerce software and service package that allowed companies to outsource the development and hosting of their e-commerce systems. The Pandesic product was based on SAP's enterprise resource planning software, and ran on Intel systems. Pandesic was led by Harold Hughes, former CFO of Intel, and many of Pandesic's senior executives were former Intel employees. Pandesic software was being used by a number of Internet start-ups, including DVDExpress.com and ALLHerb.com, as the backbone of their e-commerce infrastructures.

Challenges in New Businesses

In discussions with Intel managers across the company who were involved with new ventures, several themes came through. There were certainly many advantages to these ventures of being associated with Intel. However, there were also many challenges and disadvantages.

The Creosote Bush - The Gravity of the Core Business. The enormous success of Intel's core business created challenges for the new ventures. Being inside a $26 billion company that made over $6 billion in profits caused even the most promising new ventures to struggle to become relevant. Reflecting the view of many Intel managers, Jim Johnson said,

"Building new businesses within Intel can be very hard. The microprocessor business is so large that it wipes out anything that gets close to it."

Claude Leglise continued,

"Everyone's thinking PC around here. The review systems and management processes, while they're the right process for the existing business, they're a pain in the butt for what I'm doing... The gravity of the organization is strong."

According to Patty Murray, Intel's head of human resources,

"Intel is the world's largest single cell organism."

Craig Barrett and the Executive Staff, however, had made changes to make the environment more supportive of the new businesses. According to Claude Leglise,

"The CEO has said 'do this'. Everyone else, leave them alone. It's become socially acceptable to do what I'm doing -- not PCs."

Don Whiteside described how these changes had influenced the Digital Imaging group:

"We entered both of these initiatives as Job 1 businesses. They both supported our desire to sell more and faster PCs... Now we're looking at both of these businesses in the following way: if they happen to support Job 1, that's nice, but that's not one iota of justification for doing these businesses. We're looking at them as standalone, money-making businesses."

Claude Leglise added,

"The core business is 95% of what Intel's doing, and it's not broken, so let's not fix it. The chaos that I'm fostering (in my new business focus) would be disastrous in the PC business. We need to adapt the management approach to the demands on the organization."

Marketing & Brand Strategy. The consensus view was that the Intel brand created many advantages for new ventures, but that there were significant challenges to using it effectively. The Intel name inspired trust and confidence, and helped open doors with potential customers, partners, and suppliers, giving Intel new ventures a significant advantage relative to typical start-ups. Jim Johnson explained:

"The Intel brand was very useful. When we launched our service, the corporate marketing group did a phenomenal job. We received almost as much activity from our launch as a microprocessor launch typically receives. They put together press kits, radio spots, and TV spots, and we ended up getting phenomenal press... Also, the Intel web site put a pointer on our home page, and this generated 10,000 hits per day of visitors coming to our section of the web site. From a marketing point of view, we did a wonderful job of using the Intel name."

There were also major challenges related to marketing and branding. According to Jim Johnson (ISO),

"Although the brand name is a huge asset, it's hard to use the asset unless you're a part of the microprocessor business."

For example, Mark Christensen (NCG) noted,

"Intel's brand stands for a lot of things... [but] we're probably not known as a networking company. It's probably easier for us to extend the brand to 'the Internet building block company' than it is to extend it to 'the networking company'."

Dennis Carter, VP Marketing, discussed how the role of marketing had evolved across the three epochs:

"During the first epoch, marketing allowed us to differentiate our products from others. We made a number of innovations in sales and marketing techniques that allowed us to sell high value semiconductors. We were virtually unknown outside of the engineering community, but were innovators in business to business marketing. During the second epoch, we were again able to differentiate our product from a marketing perspective, but this time via Intel Inside®. We successfully took our brand to the end-user. Marketing was a key element in the success of our microprocessor business. During the third epoch marketing must again play a key role, we will need to move our brand to include a broader set of products, and we need to make the Intel brand relevant in the Internet space."

Sales & Marketing Channels. For the most part, the new ventures relied on Intel's existing sales forces to sell and market their products in the field. This created challenges in terms of mind share, and conflicts in funding occasionally arose. According to Jim Yasso, a veteran of Intel's sales force before his tenure as GM of RPD,

"The sales force is part of the blue process. For anything that's green, it's very hard to get resources, unless they are explicitly included as part of the original plan and budget allocations. You also need to have marketing people in place in the geographies... We have quarterly POR processes where we get funding approved or adjusted. Within a division or geography, we can be subject to headcount or spending cuts, and in some cases these budget decisions can conflict [with funding for the new ventures]. This can be a real killer for green businesses... With NBG, we've kept the [green] businesses separate and shielded them from the ongoing [blue] business."

Lorie Wigle mentioned another issue:

"Getting mind share of the sales force is difficult."

The new emphasis at Intel on non-core businesses created challenges for the sales force. According to Tom Lacey, a vice president in Intel's sales and marketing group,

"A lot of the sales reps at Intel have been here for a long time. Most of them grew up on the IA bandwagon. We're now trying to sell more Intel products through the same channels. This is requiring people to learn more about some of our other products... This has been a bit of a transformation for the sales force... We're in the process right now of reorganizing the sales force. We're creating a separate sales force that is focused on non-CPU businesses."

Finance & Resource Allocation. Intel's financial resources created a tremendous advantage for the company. At the same time, Intel faced challenges in terms of allocating resources to new businesses. Speaking of the funding challenge that new ventures faced at Intel, Tom Lacey explained:

"We apply the POR process to establish budgets for all of our projects... Virtually every single quarter, the requests outweigh the willingness to spend. We would end up 'ZBB-ing' the lower ROI projects. The larger ROI projects were almost always related to the mainstream CPU business. Therefore, if you were not part of the mainstream business, you needed to be very spirited and very perseverant to drive your projects through that POR process every quarter. In many cases they were great businesses by any other metric, just not compared to the microprocessor business. Three or four years ago I asked Craig [Barrett], 'shouldn't we be diversifying more?' His answer was 'Absolutely not. It takes every bit of our energy to execute on the microprocessor business.' If you were in a non-CPU business, it was tough."

Les Vadasz summarized part of the challenge that Intel faced in this area:

"One of the biggest dangers for big companies is not being able to manage small amounts of money. For some top managers, these new ventures are trivial stuff. But these ventures are necessary to stay on the bleeding edge. On the other hand, you can easily overspend too early. I have done this various times in my life!"

A pronounced tension existed between Intel's strong control structure and the patience that many Intel managers believed was necessary for new ventures to thrive. Subjecting new ventures to the POR process created difficulties. During Epoch II, the highest ROI projects were typically connected to the microprocessor business. Hence, few opportunities outside the core business survived the funding process. In Epoch III, being subject to the quarterly POR process still created challenges. According to Jim Johnson,

"When ISO was established, I tried to get a fixed amount upfront. Then if we run out of money, it's our problem. It turned out that we were funded out of the quarterly POR, which means that our budget is reviewed every quarter. This creates major challenges. I had taken over 1 year to negotiate an advertising arrangement with AOL that included key pages on AOL sponsoring our service. [The deal] required us to make a $2 million commitment over a 12 month period. At the end of the negotiation, [my boss at the time] said he wasn't going to sign it... The $2 million investment was key. If we had been given our funding upfront, rather than quarterly, we could have saved money in other areas and found enough to do this deal."

Though most managers gave their finance counterpart high ratings, one senior executive said,

"My controller runs my business. He spends very little time on it, and he doesn't understand my strategy, my customers, etc. Yet, he controls my purse strings, and gives me very little flexibility. With the POR process, you go in with your request by department. He says which department gets funded and which one does not. This happens every quarter. We spend the last 30 days of the quarter preparing for the budget review, then the first 30 days of the next quarter appealing the budget cuts. So we end up spending 60 out of 90 days negotiating our budget."

Intel's hard-nosed culture and financial discipline served it well in the core business, but could be a liability to the new businesses. For example, there seemed to be a common perception that Intel lacked patience with new ventures. According to Scott Darling,

"We tend to want to go from zero to a multi-billion dollar business very rapidly... We've been too mechanistic. We are an engineering company, and we've approached new businesses like you'd solve an engineering problem. But there is serendipity in new ventures. We're often wrong about new markets... And we have tended to micromanage, although Gerry [Parker] has avoided this."

Renee James discussed the objectives of Intel Data Services,

"The edict from Craig [Barrett] is to make this a $1B business in 3 years".

Dennis Carter added,

"A key challenge that we face in new businesses is patience -- the popular conception is that we don't have any. Craig says that these green businesses will be successful within 2 years. People respond, 'But the networking business took 8-10 years to build up.' Craig correctly counters, 'Yes, but if it was managed better, we could have grown it more quickly.'"

Craig Barrett addressed the patience issue head-on:

"For the people involved with the new ventures, I'm sure they see us as being impatient. What I tell them is, show me how you're going to be number 1 or number 2, and how you're going to build a viable business. If they can't do this, then we're not going to be patient. There are examples, such as chip sets and networking, where we've been very patient. My philosophy is to deal out patience in small doses. But I believe we've demonstrated to the company that we have a degree of patience. The damn environment changes so fast, that we have to adjust priorities and resource requirements as the environment dictates."

Another common theme that emerged was that having executive sponsorship for an opportunity was an important criterion for getting funded. According to Jim Johnson,

"At Intel, if you have an idea, there are 20,000 people who can say no, but just a few that can say yes."

Speaking about the funding for the project with Mattel, Michael Bruck said,

"If the idea was not so compelling, it probably would not have worked... I've seen good ideas get shot down at Intel. This project has involved high-level involvement at Intel, and we've received buy-in from the people at the top. At Intel, lots of people have veto power, but few have full approval power... I've been surprised at how easy the funding has come about."

A popular view within the organization was that Intel's top executives did not always apply the same level of financial discipline to their own projects. Representing a common theme, one senior manager said,

"One weakness within Intel is that new business activities get decided by executive prerogative."

According to another manager, speaking of the Digital Imaging group,

"When they broke up the ProShare division (which had been strongly endorsed by Andy Grove) and moved it to the Create & Share division, they had to establish a culture of fiscal discipline because they had had an entitlement mentality. They had to do a similar thing with the silicon people because this had been Craig Barrett's pet project."

Compensation & Incentive Systems. At Intel, the evaluation system worked as follows: goals and objectives for the coming year were established in December. Performance was evaluated the following December based on the achievement of these objectives. Employee bonuses, which were typically a very significant portion of overall compensation, were based on these evaluations. This created a major challenge for new ventures, where reasonable objectives were difficult to anticipate. According to Jim Johnson,

"In these new ventures, things change about every 30 days. Yet the targets are frozen 12 months in advance. Our group has been reorganized since then!"

Dennis Carter elaborated,

"Unfortunately, [objectives] are often hopelessly outdated after 12 months. Things just change too much. In the case of a start-up effort, it's even more difficult... The system penalizes you for the downside, but does not reward you for the upside associated with these changes. A mechanism for relief might be useful."

Another challenge was how to provide appropriate incentives to the new venture groups. Compared to typical start-ups, new corporate ventures had relatively low upside. According to Gerry Parker,

"The EB payouts were bad for my groups... Right now, it's all downside. This makes things not terribly easy."

According to Jim Johnson,

"Having done several of these new ventures... Being part of Intel is a huge advantage when dealing with external organizations, but it can also be a huge disadvantage internally... Within Intel, there's no way to have real equity, autonomy, or ownership."

Addressing the compensation issues, CFO Andy Bryant said,

"(The new businesses) say, we want the benefit like a start up would have. Fine, then I'm going to take away the Intel stock options. I see it as a risk-reward issue. If you want to be rewarded like a start up then you need to have the same motivation as another start up... The problems that we're having are often related to the expectations of the people who do it. They say we want to go off and be independent and have the upside of a startup. OK, so we take away your Intel stock options. The translation when they go back to their groups is that "they're not working with us. I had a great idea, but they wouldn't fund it." But VCs only fund 20% of the plans. So if we do the same, the 80% that don't get funded complain and say we're not being supportive. Meanwhile we've funded the other 20% and they're busy getting implemented."

One of Intel's six core cultural values was risk taking. The philosophy was that with a leading edge technology strategy, Intel would always compete in an environment of constant change. Attempting to maintain the status quo in an environment of change was viewed by Intel as a recipe for failure. Therefore, risk taking was encouraged. However, as Epoch III began to unfold, a common view was developing in the organization that Intel did not particularly support risk taking any more. According to Dennis Carter,

"It is a commonly held belief among middle managers at Intel that it is difficult to take risks,... that Intel doesn't reward risk takers who fail, only those who have succeeded. This is certainly the perception, and possibly also the reality."

Building General Management Competencies. During the first epoch, Intel's technical sales and field engineering teams played an essential role in applying the company's products and finding new applications for them. During the second epoch, the application -- the PC -- was clear, unchanging, and relatively homogeneous. This reduced the importance of inbound marketing, as well as Intel's capabilities in this area. Dennis Carter explained,

"During the 70's, there was a very good process for inbound marketing. Our sales force was in excellent touch with our OEM customers. During the second epoch, our sales force stayed in tune with the OEMs, but our marketing expanded to the much larger audience of PC consumers. We have limited marketplace feedback for things outside of processors and many of our internal processes withered. For the 3rd epoch, it will be critical to redevelop these internal processes... Product planning is often done too quickly by people too removed from the market... Intel is very good at execution, and Intel management processes lead to the development of excellent intuition in familiar areas, but in new product areas we are bad at judging what will play in the market... Today the link with the market is missing... We need a closed loop into the product planning process."

Claude Leglise added,

"The technology is the easy part. We know how to get that. It's not so hard and we know how to do it. The part that is frightening is what's the application."

Gerry Parker said:

"At Intel, we have a propensity to focus on one thing, build it big, and then say 'here it is'. We have the mentality of 'build it and they will come'... We're now making some real progress on being more customer responsive and my green teams are spending time at every staff meeting on customer status and finding customers."

One challenge facing Intel was developing corporate entrepreneurs who could turn good ideas into good businesses. According to Claude Leglise,

"We have labs in Oregon that are concocting ideas all day long. We have a resource pool for ideas. We've made equity investments in a couple hundred small companies. It's a matter of sifting through and seeing what makes sense, and taking the products to customers."

Dennis Carter added,

"There are lots of ideas and new projects sprouting up everywhere within Intel, but the real question is how we can turn them into a business. We have the capacity to do so, but is the infrastructure there to support it?... We have clever ideas coming from places like IAL and the Content Group, but little experience in making a business out of any of them. We need to develop the intuition for doing this. And we need to develop people with the entrepreneurial intuition."

Another important issue was developing general managers who could lead and manage these ventures. According to Craig Barrett:

"[During the second epoch] we became much more verticalized behind IA and related businesses. Now we're much more broad, with networking, Parker's businesses, and so forth. This requires less top down management and more P&L and line management. "

Andy Grove added, in a discussion with Intel's top management,

"There's a lack of a merchant mentality inside Intel. We're not good at it. We need to get experience, and then build on that experience."

 

Looking Ahead

In spite of Intel's efforts to create new businesses, Chairman Andy Grove was not satisfied. Asked about the new businesses at Intel, Grove said in early 1999,

"We've been very consistent and uniform. All of the new efforts, like all of the other efforts before them, have all been failures... Nobody can fault us for not trying. We've learned a lot, but we've not yet put those learnings into the creation of a third business."

Grove then brought Sean Maloney, senior vice president of sales, into the discussion:

Grove (to Maloney): "How have the new businesses gone?"

Maloney: "We've succeeded at changing the attitudes in the company from unrealistic expectations and an unsupportive management structure to more realistic expectations and a more supportive management structure. But at this point we have no results."

Grove: "I rest my case."

Then referring to Intel's vision of a billion connected computers:

Grove: (to Maloney) "What percentage of ESM (Executive Staff Management) would you say gets it?"

Maloney: "I would say a majority of ESM now gets it."

Grove, shaking his head no: "If by a majority you mean 51%, then I agree with you."

Grove himself felt compelled to become more educated in this area, and he had begun meeting with a wide range of successful Internet companies to better understand their businesses and financial models. Speaking to his senior management team in mid-1999, Grove encouraged them to expand their expertise as well, telling them:

"There is lots of expertise and knowledge here. Most of us are engineers by training and one thing we know is how to learn. You need to go out in your own space and learn everything you can about this new environment."

At the same time, CEO Craig Barrett was spending a significant percentage of his time travelling internationally as part of his efforts to further stimulate international growth.

In leading Intel into the new millennium, Barrett seemed to relish the challenges facing he and his company. Barrett said,

"I've tried to spread a couple of messages around here. First is that there's a different vision. We have more business opportunities than just the microprocessor. Microprocessors by themselves will not be the growth engine that it's been in the past... I try to challenge the new ventures to grow their businesses so that they're relevant, and I challenge the people in the microprocessor business to be successful enough so that the new ventures won't matter... The creosote bush is still here. It hasn't gone away, and I hope it doesn't! We need to grow some other bases, but it would be inappropriate to say we have changed 100%."

Chairman Emeritus Gordon Moore described some of the challenges facing Barrett:

"Craig is a strong hands-on guy, like Andy, but even more so. He keeps himself in the loop a lot more. It's pretty amazing that he can do this, and it's because he has so damn much energy... Eventually Craig will need to find someone to complement himself... He has done a tremendous job, moving our focus from the microprocessor to the Internet building blocks. The key issue here is getting critical mass in an area... Jack Welch at GE says that you have to be number one or number two in an industry, and this is especially true in technology. Hangers on tend to not be very good. At the same time, we can't neglect the microprocessor business."

 

Exhibit 1: Key Events in Intel's History

1968

  • Intel founded by Robert Noyce (CEO) and Gordon Moore (EVP)

1969

  • Intel introduces its first product, 3101 64-bit Schotky bipolar RAM
  • Intel introduces the 1101, the world's first MOS static RAM

1970

  • Recession forces company to lay-off employees for the first time
  • 1103, world's first commercially successful DRAM, introduced

1971

  • Intel moves into its own building in Santa Clara
  • Late List instituted
  • 1702, world's first EPROM, introduced
  • Company goes public at $23.50 per share, raising $6.8 million
  • 4004, world's first microprocessor, introduced

1972

  • 2102 1kB static RAM, company's first NMOS, introduced
  • Company enters digital watch business, acquiring Microma
  • 8008, first 8-bit microprocessor, introduced

1973

  • Intellec 4-40 microprocessor development tool introduced

1974

  • 8080 industry standard 8-bit microprocessor introduced

1975

  • Bob Noyce elected chairman of the Board
  • Gordon Moore elected president and CEO

1976

  • 2147 static RAM, first HMOS product, introduced
  • 8085 8-bit microprocessor introduced

1977

  • Intel enters bubble memory business

1978

  • Intel phases out of the Microma digital watch business
  • 8086 industry-standard 16-bit microprocessor introduced

1979

  • Gordon Moore elected chairman of the Board
  • Bob Noyce elected vice chairman
  • Andy Grove elected president and COO
  • 8088 industry standard 8-bit microprocessor introduced
  • Bob Noyce awarded National Medal of Science by President Jimmy Carter

1980

  • Intel introduces the first math coprocessor, the 8087

1981

  • iAPX 432 microprocessor introduced
  • Intel's sabbatical program begins in the US
  • IBM announces its first PC, based on Intel's 8088 microprocessor
  • Intel's "125% solution" launched

1982

  • 80186/80188 16 bit embedded processors introduced
  • 80286 16-bit microprocessor introduced
  • First LAN coprocessor, the 82586, introduced
  • Wall Street Transcript names Gordon Moore the outstanding CEO for the semiconductor industry the second year in a row
  • IBM announces plans to purchase 12% of Intel for $250 million

1983

  • Intel imposes pay freeze/salary cuts in face of poor business conditions

1984

  • Gordon Moore and Bob Noyce named to IEEE Hall of Fame
  • IBM announces PC AT based on Intel's 80286 processor

1985

  • Intel announces first layoffs in 10 years
  • Intel/AMD/National Semiconductor file joint anti-dumping petition with US government against Japanese EPROM manufacturers
  • Intel decides to quit the DRAM business
  • Intel 386 CPU introduced
  • Bob Noyce inducted into National Inventors Hall of Fame

1986

  • Court rules that microcode can be copyrighted and that Intel's copyright is valid, in response to lawsuit brought by NEC

1987

  • Andy Grove elected CEO
  • Bob Noyce receives National Medal of Technology from President Ronald Reagan
  • Intel launches Knockout campaign
  • IBM sells the last of its shares of Intel stock

1988

  • Employee cash bonus announced
  • Late List ends

1989

  • i860 processor introduced
  • Intel 486 processor introduced
  • i960 CA processor introduced for embedded applications

1990

  • Craig Barrett becomes executive vice president of Intel
  • Bob Noyce dies
  • Gordon Moore receives the National Medal of Technology from President George Bush

1991

  • Intel announces it will exit EPROMs in favor of flash memory
  • Intel launches Intel Inside® program

1992

  • Intel announces OverDrive processors
  • Court rules that AMD does not have the right to copy any Intel microcode

1993

  • Craig Barrett named COO
  • Intel introduces Pentium processor

1994

  • Intel introduces ProShare videoconferencing systems
  • A mathematics professor discovers Pentium processor bug, leading to the Pentium processor flaw crisis

1995

  • Intel introduces Pentium Pro processor

1996

  • Oracle, Sun, and IBM announce the network computer

1997

  • Intel introduces Pentium processor with MMX technology
  • Intel introduces Pentium II processor
  • Sub-$1000 PC market gains momentum
  • Craig Barrett commissions a course on "Growing the Business" for the top 100 managers in the company
  • Andy Grove selected as Time Magazine's "Man of the Year"

1998

  • Intel Celeron processor introduced
  • Intel introduces Pentium II Xeon processor
  • Craig Barrett elected CEO
  • Andy Grove elected chairman of the Board
  • Gordon Moore elected chairman emeritus

1999

  • Intel introduces Pentium III and Pentium III Xeon processors
  • Intel acquires Level One technologies for $2.2 billion
  • Intel announces plans to launch Intel Data Services

Exhibit 2: Intel's Evolution From Memory Company to Microprocessor Company

Percentage of Intel's Revenues

 

 

 

Exhibit 3: The Relative Importance of Different Distinctive Competencies, Memory Versus Microprocessors

Memory Products

Microprocessor Products

 

 

 

 

Exhibit 4: Market Segment Share of I-A Microprocessors

By Units

 

Exhibit 5: Intel Financial Highlights, Revenue

Net Revenue, 1968-1986 ($Millions)

Net Revenue, 1986-1998 ($ Billions)

 

Exhibit 6: Intel Financial Highlights, Profits

Net Profit, 1971-1986 ($Millions)

Net Profit, 1987-1998 ($Billions)

Exhibit 7: Intel Financial Highlights, R&D

R&D Expenditures ($M), 1971-1986

R&D Expenditures ($B), 1987-1998

 

Exhibit 8: Intel Financial Highlights, Plant Property & Equipment

PP&E Expenditures ($M), 1971-1986

PP&E Expenditures ($B), 1987-1998

 

 

Exhibit 9: Intel Market Capitalization

($ Millions, 1973-1987)

($ Billions, 1987-1999)

Exhibit 10: The Evolution of Intelís Microprocessors

Processor

Date of Release

Clock Speed (MHz)

# Transistors (millions)

Typical Uses

4004

Nov 1971

0.108

0.002300

Busicom calculator

8008

Apr 1972

0.200

0.003500

Dumb terminals, bottling machines

8080

Apr 1974

2.0

0.006000

Traffic light controller, Altair PC

8085

Mar 1976

5.0

0.006500

Toledo scale

8086

Jun 1978

5.0-10.0

0.029000

Portable computing

8088

Jun 1979

5.0-8.0

0.029000

IBM PCs and clones

80286

Feb 1982

6.0-12.0

0.134000

PCs

386 DX

Oct 1985

16-33

0.275000

Desktop computing

386 SX

Jun 1988

16-33

0.275000

Entry level desktop & portables PCs

386 SL

Oct 1990

20-25

0.855000

Portable PCs

486 DX

Apr 1989

25-50

1.2

Desktop computing and servers

486 SX

Apr 1991

16-33

1.185

Entry level desktops

486 SL

Nov 1992

20-33

1.4

Notebook PCs

Pentium processor

Mar 1993 to

Jun 1996

60-200

3.1-3.3

Desktops, notebooks, and servers

Pentium Pro processor

Nov 1995

150-200

5.5

High-end desktops, workstations and servers

Pentium processor with MMX tech.

Jan 1997

166-233

4.5

High performance desktops & servers

Mobile Pentium proc. with MMX tech.

Sep 1997

200-300

4.5

Mobile PCs and mini-notebooks

Pentium II processor

May 1997

233-450

7.5

High end desktops, workstations, and servers

Mobile Pentium II processor

Apr 1998

233-400

7.5-27.4

Mobile PCs

Celeron processor

Apr 1998

266-466

7.5-19.0

Low cost PCs

Mobile Celeron processor

Jan 1999

266-400

18.9

Low cost mobile PCs

Pentium II Xeon processor

Jun 1998

400-450

7.5

Midrange & higher workstations & servers

Pentium III processor

Feb 1999

450-550

9.5

High end desktops, workstations, and servers

Pentium III Xeon processor

Mar 1999

500-550

9.5

Business PCs, 2-, 4-, and 8-way servers and workstations

 

 

Exhibit 11: Intel Organization Chart, 1999

Exhibit 12: Intel Officers, 1999

Executive Officers:

Gordon E. Moore, Chairman Emeritus of the Board

Andrew S. Grove, Chairman of the Board

Craig R. Barrett, President and Chief Executive Officer

Corporate Officers:

Michael A. Aymar, Vice President and General Manager, Internet Data Services

Andy D. Bryant, Senior Vice President and Chief Financial Officer

Louis J. Burns, Vice President and General Manager, Platform Components Group

Dennis L. Carter, Vice President and Director, Strategic Marketing

Sunlin Chou, Vice President and General Manager, Technology and Manufacturing Group

Mark A. Christensen, Vice President and General Manager, Network Communications Group

F. Thomas Dunlap, Jr., Vice President, General Counsel and Secretary

Kirby A. Dyess, Vice President and Director, New Business Development

Carlene M. Ellis, Vice President and Director, Education

Patrick P. Gelsinger, Vice President and General Manager, Desktop Products Group

Hans G. Geyer, Vice President and General Manager, Flash Products Division

D. Craig Kinnie, Vice President and Director, Intel Architecture Labs

Sean M. Maloney, Senior Vice President and Director, Sales and Marketing Group

John H. F. Miner, Vice President and General Manager, Enterprise Server Group

Patricia Murray, Vice President and Director, Human Resources

Stephen P. Nachtsheim, Vice President and Dir. of Operations, Corporate Business Development

Paul S. Otellini, Executive Vice President and GM, Intel Architecture Business Group

Gerhard H. Parker, Executive Vice President and General Manager, New Business Group

Ronald J. Smith, Vice President and General Manager, Computing Enhancement Group

Stephen L. Smith, Vice President and General Manager, IA-64 Processor Division

Arvind Sodhani, Vice President and Treasurer

Michael R. Splinter, Senior Vice President and GM, Technology and Manufacturing Group

Leslie L. Vadasz, Senior Vice President and Director, Corporate Business Development

Ronald J. Whittier, Senior Vice President and General Manager, Intel Content Services

Albert Y.C. Yu, Senior Vice President and General Manager, Microprocessor Products Group

 

Exhibit 13: Intel Revenue by Product Category, 1996

 

Microprocessors

$14.9 B

Chipsets

$1.4 B

Motherboards

$1.2-2.0 B

Networking

$1.0-2.0 B

Flash Memory

$950 M

Embedded controllers

$530 M

Video conferencing

$33.5 M

 

Exhibit 14: Desktop PC Market, Unit Shipments

(000s of Units, By Price Range)

 

 

Exhibit 15: Average Sales Price, Intel Microprocessors

 

 

 

Exhibit 16: Market Segment Share of I-A Microprocessors

I-A Desk-bound, By Units (000s)

I-A Mobile, By Units (000s)

Exhibit 16: Market Segment Share of I-A Microprocessors (cont.)

I-A Servers & Workstations, By Units (000s)

I-A Upgrades, By Units (000s)

Exhibit 17: Intel Microprocessor Pricing

(1000-Piece List Price, $)