SUMMARIES.COM is a concentrated business information service. Every week, subscribers are e-maileda concise summary of a different business book. Each summary is about 8 pages long and contains thestripped-down essential ideas from the entire book in a time-saving format. By investing less than one hourper week in these summaries, subscribers gain a working knowledge of the top business titles. Subscriptionsare available on a monthly or yearly basis. Further information is available at http://www.summaries.com.

INSIDE INTELHow Andy Grove Built the World’s

Most Successful Chip Company

TIM JACKSON

1.

When most people think of the start-up phase of a hightechnology company like Intel, they have a mental picture of alone scientist working feverishly away in a garage late at nightto develop a device everyone else thinks is impossible. Theestablishment of Intel in 1968, however, was nothing like thiscliche.

Intel’s co-founders were Robert Noyce and Gordon Moore.

Robert Noyce was 40 years old in 1968. He had been hired byBell Laboratories in 1956 and had assisted in the developmentof the transistor -- for which he and two other engineers had beenawarded a Nobel prize. Robert Noyce was widely regarded asone of the pre-eminent engineers in his field.

When Noyce had left Bell Laboratories in 1957 to set up FairchildSemiconductors, he came into contact with Gordon Moore whowas hired to head up Fairchild’s research and developmentprogram. Moore, like Noyce, had a PhD in chemical engineering.

By 1968, both men had become disillusioned with how Fairchildwas going, and decided to start their own company. To fund it,Noyce called Arthur Rock, an investment banker based in SanFrancisco. They put together a two page document outlining theirbusiness plan, and made 15 phone calls to people they thoughtmight be interested in investing in the as yet unnamed company.All 15 people agreed to provide funding -- they literally lined up$2.3 million of start-up funding in one afternoon of phone calls.

It helped, of course, that both Noyce and Moore were quitewealthy themselves, and were each contributing around$300,000 themselves to the new company. Nevertheless, mostof the start-up funding was provided solely on the strength of theprofessional reputation of the two founders.

Within a couple of weeks, they had decided on a name for theirnew venture, deciding that "NM Electronics" didn’t sound rightbut "Integrated Electronics" sounded much better. This waseventually shortened to Intel, and the new company was formallyincorporated on July 16, 1968. (Once the company was formed,they then found another company called Intelco was already inexistence, but they were able to buy the right to use the Intelname from Intelco for $15,000).

While the new company’s initial "Business Plan" could havebeen described as sketchy and vague at best, in reality thecompany founders knew exactly what they wanted to developand build -- memory devices for computers using the newsemiconductor devices, built from silicon chips.

‘‘Fabricating silicon chips was the modern world’s answer tomedieval alchemy, the turning of base metals into gold. Excepthere, the raw material was sand, which was turned into acrystalline silicon which arrived at the fab molded into a longsausage, two inches in diameter. The silicon would then besliced into thin "wafers" a fraction of an inch thick. By a series ofsecret, almost magical processes, each wafer would be coatedwith scores of identical miniature circuits, neatly stepped in rowsand columns. Then the wafers would be scored with a diamondcutter, and the individual chips would then be sawn away fromtheir neighbors and wired individually into black ceramicpackages, often with a line of metal pins down each side.’’

-- Tim Jackson

Noyce and Moore found an old Union Carbide plant in MountainView, California, which was set up for manufacturing. Theyleased the building, which was about 17,000 sq. feet in size, and

started ordering manufacturing equipment that would have thecapacity to manufacture around 10 million integrated circuits ayear -- an ambitious goal for a new company without even acircuit design of its own yet.

Noyce and Moore also hired a director of operations to assumeresponsibility for the manufacturing side of the company. Theychose Andrew Grove, a PhD physicist who had nomanufacturing experience whatsoever.

‘‘In 1968, people who met Grove for the first time usually noticedthree things. One was that he was very bright, very good atexplaining things -- particularly semiconductor devices, whosephysical behavior he had written a book about. Another was thathe was very organized, and seemed to know exactly what hewanted and how he was going to achieve it. A third was that hewas very keen to make an impression, to justify his position.Grove knew that Noyce and Moore had taken a risk by givinghim the job of director of operations, and he was determined toprove they had made no mistake.’’

-- Tim Jackson

At around this same time, Noyce and Moore also hired BobGraham, one of Fairchild’s star salesmen, to become Intel’smarketing director.

2.

The first focus for Intel was to develop a new manufacturingprocess for integrated circuits that would be more cost effective.To achieve this, they used an approach Noyce and Moore hadfirst examined while at Fairchild --a process using metal oxideon silicon which was called "silicon gate". Since they weren’tcertain the process could be scaled up to commercial volumeproduction, they also worked on two other approaches --multichip memory modules and the "Schottky bipolar" -- with theobjective that all three approaches would be tried simultaneouslyuntil the best was successfully into production.

While this simultaneous development program may have beencommercially prudent, it also had the effect of generating somehighly charged internal discussions within the company. Most ofthese "discussion" sessions would be conducted with bothparties screaming at each other at the top of their lungs -- andusually in front of other people. The first year of Intel’s historywas full of such stand-up screaming matchups.

By the spring of 1969, the biploar team had developed a 64-bitmemory chip which Honeywell agreed to use in a new computerthe company was developing.

‘‘The chip for Honeywell became the new company’s firstcommercial product. It was a symptom of Intel’s target marketthat the new chip wasn’t even given a name. Instead, it wasreferred to only by a part number, 3101. Intel’s potentialcustomers were engineers inside computer companies, whothough of themselves as rational decision makers choosingbetween one part and another strictly on technical merit, qualityand price. A catchy name wouldn’t have increased sales; on thecontrary, it might have excited suspicion that there was ashortage of engineering talent to cover up for. A simple partnumber, preferably a number that meant something, was theway to go.’’

-- Tim Jackson

The early success of the bipolar team spurred the silicon gateteam onwards with steely determination. They were having

Inside Intel - Page 1

trouble with their manufacturing process, and it was not untilMoore came up with a suggestion that the metal oxide shouldbe ‘‘doped’’ with an impurity to lower its melting point thatproduction of the silicon gate based memory devices couldbegin.

While Intel was perfecting its memory devices, wholesalechanges were taking place in the outside world. Many newcompanies were starting to compete for the same markets Inteltargeted. One of these competitors, Advanced Micro Deviceswith Jerry Sanders as CEO, was even successful in getting BobNoyce to invest in it. (It was characteristic of Noyce’s confidencein Intel’s ability to stay ahead of the pack that he was evenprepared to invest in another company in the same field. Noycedid, however, go back a long way with Sanders, which was themajor reason Noyce chose to invest in AMD).

‘‘We really are the revolutionaries in the world today -- not thekids with the long hair and beards who were wrecking theschools a few years ago.’’

-- Gordon Moore, Fortune Magazine interview, 1973

Ultimately, the silicon gate development team was able to comeup with a new memory cell, called the "dynamic random-accessmemory" or DRAM, which would store 1,024 bits of informationwhile being smaller faster, cheaper to manufacture and requiringless operational power than conventional chips. The new chip,designated the 1103, was released commercially in October1970.

Intel’s sales were $566,000 in 1969, with the majority beingderived from the sale of 3101 chips to Honeywell. In 1970, totalsales revenue increased to $4.2 million, mainly on the back ofsales of the 1103 chip. The company now had more than 100employees, and was rapidly outgrowing the old Union Carbideplant. To provide room to grow, a 26-acre block of orchard landwas acquired further south in Santa Clara, in an area that wasnow beginning to be called "Silicon Valley" in recognition of thenumber of electronics industry businesses which were moving tthe area.

3.

In the process of developing a set of chips for a new desktopcalculator for the Nippon Calculating Machine Corporation, acouple of Intel’s engineers hit on the idea of building aminiaturized general purpose computer instead, which couldthen be programmed to do the arithmetic for the calculator. Sucha device would be simpler to build -- requiring only four chips (acentral processing unit, working data memory chip, read-onlymemory chip and input/output chip) -- than the eight chip setoriginally envisaged.

Although the new device was originally developed for the oneclient, Intel did retain the right to sell the design to othercustomers, so long as they weren’t in the business ofmanufacturing calculators. Ultimately, the new device, nowcalled the microprocessor and given the code number 4004, wasreleased in November 1971.

‘‘Intel’s marketing people looked at the new chip and madepessimistic noises. If it wasn’t going to put the mainframecomputer out of business, what was this new gadget for? Itscreators inside Intel’s research labs were full of ideas. They feltit could add intelligence to a whole range of electrical businessmachines. Until now, building intelligence into such machines

had been prohibitively expensive, because it required designinga special dedicated piece of computer hardware for eachapplication. The 4004 would change all that.’’

-- Tim Jackson

Not surprisingly, the early adopters of the 4004 were quiteobscure companies, whereas the companies who were buyingIntel’s memory chips were America’s leading technologyproviders. In fact, it wasn’t until the next generation of themicroprocessor, an 8-bit chip designated the 8008, was releasedin August 1972 that anyone started to think along the lines ofconsidering the feasibility of using Intel microprocessors insidebusiness computers.

The marketplace success of the Intel memory chips andmicroprocessors only seemed to intensify the passion of theinternal battles. This lead to the resignation of Bob Graham asmarketing director, after he lost out to Andrew Grove. Grahamwas replaced by Ed Gelbach, Texas Instrument’s national salesmanager.

Intel went public in October 1971 at $23.50 per share. Betweenthem, Robert Noyce and Gordon Moore held just over37-percent of the issued shares, with a market value of around$20 million.

4.

While the 1103 memory chip had been successfully introducedinto the marketplace as the world’s first semiconductor memorydevice, Intel was acutely aware of the problems associated withthe device. It would need to be revamped and made easier touse if it was to make the transition from technological curiosityto commercial success.

In addition to redesigning the chip itself, Intel engineers alsostarted building memory sub-systems that used 1103 chips incomplete integrated memory systems. Fortuitously, at aroundthe same time, Intel was approached by two ex-Honeywellengineers, who presented a business plan idea on setting up adivision of Intel to do precisely that. Noyce and Moore wastedno time in agreeing terms, and Intel’s memory systems businesswas born about a month later.

Memory systems were essentially arrays of 1103 chips anddriver circuits. Because the performance of the entire unit ratherthan the individual chip was of paramount importance,substandard 1103 chips were were not suitable for sale could beused in the memory devices. This was an added bonus for Intel,for whom the substandard chips had simply been a necessarypart of the overall manufacturing process.

Another Intel engineer also came up with a new way to use thecompany’s silicon gate technology -- a semiconductor devicethat behaved like read-only memory but that could be easilyprogrammed. The company called the new technology anerasable, programmable read-only memory or EPROM for short.

With these two new technologies at the fore, Intel’s salesrevenues grew rapidly and spectacularly -- from $9 million in1971 to $66 million in 1973 and $134 million the following year.Intel now had 3,100 employees, and was in the process of addinga third fabrication plant in Livermore, California, to its two existingplants at Santa Clara and Mountain View. By 1974, Intel hadbecome the fifth largest maker of integrated circuits in the world.This accomplishment was marked by a slight reshuffle ofresponsibilities, with Robert Noyce becoming chairman, Gordon

Inside Intel - Page 2

Moore president and CEO and Andy Grove being madeexecutive vice president of the company.

A small development team was set up to make a few technicalchanges to the 8008 chip, including the conversion to a newprocess called NMOS. While they were doing that, it was decidedthey should also make a few modifications to the design itself,and change to a 40 pin chip package. Taking all these changestogether, the new chip, the 8080, would run about ten timesfaster than its predecessor, the 8008.

Since the new chip would only cost $20 or $30 to manufacture,Intel was stumped on how it should price the new chip.Eventually, the price was set at $360, as a wry dig at IBM’s bestselling range of mainframes, the IBM 360. The 8080 was firstshipped in April 1974.

While the company’s technical productivity was impressive,Andrew Grove’s management techniques also receivedextensive interest. In essence, the key methods Grove used tomanage innovation within Intel were:

1. He insisted on a rigorous, regular budgeting process.Department heads were required to make detailedprojections of costs and revenues, and then to explain whyany subsequent changes were necessary.

2. Groves also insisted every employee have a written set ofobjectives and key results by which success or failure inachieving the objectives could be measured.

The budgets and the objectives would then form an integral partof regular meetings between the employee and the person towhom he reported, where a written performance review wouldbe delivered and discussed. (At these meetings, performancescategories would be rated as "superior", "exceeds expectations","meets expectations" or "does not meet expectations".)

Other meetings were also held, including an annualmanagement retreat and monthly management review meetingswhere various departments would make presentations on thestrengths, problems and opportunities facing that division.

Grove also instituted a highly unusual company practice --anybody who arrived at work later than 8 am each day wasrequired to sign in. This was considered extremely ridiculous bymost of the employees, particularly since many of themfrequently worked late hours on a regular basis, but Groveinsisted and the policy stayed in place throughout the 1970s andthe 1980s.

Ultimately, the Groves approach to company discipline wouldend up forcing some talented engineers to leave Intel. Thisincluded people like Frederico Faggin and Ralph Ungermannwho, with $1.5 million in start-up funding from ExxonCorporation, went on to form Zilog. (The name "Zilog"represented the last word (z) in integrated (i) logic (log).) Thenew company focused on developing an improved 8080 chip,and even contacted Intel offering to do the work on a contractbasis, but Intel rejected their idea. So Zilog went ahead andabout a year later released the Zilog Z-80 chip -- which waspriced at only $200 and was completely interchangeable with theIntel 8080 chip.

‘‘With a price tag of only $200, the Zilog Z80 began to capturethe imagination of individual hobbyists. People who hadpreviously spent long nights in front of a ham radio were now,thanks to the glories of large-scale circuit integration, tinkeringwith electronics. The Z80 made it possible for the first time forthem to contemplate the revolutionary possibility of building acomputer of their own. Since there were fewer computers in the

world in 1974 than airplanes in the world in 1997, telling yourfriends you were doing this had the same kind of ring then asannouncing that you were assembling a Boeing 767 in your backyard would today.’’

-- Tim Jackson

5.

‘‘Intel was engaged in a perpetual struggle to make its productsnot only more powerful but also smaller, faster and cheaper.Dramatic improvements on these fronts were the best way tograb the attention of customers, and also the best way to keepahead of competitors. But it wasn’t only the chip-designengineers who were under this constant pressure to improve:their work often resulted in a new product that was bigger andmore complicated than its predecessor. The toughest battleswere often fought elsewhere -- in the fabs where thesemiconductors were actually built.’’

-- Tim Jackson

Intel’s fabrication plants were in a constant state of flux aschanges were being made to improve the manufacturingprocesses used. The stakes were very high -- reducing costsmeant chips could be sold cheaper, leading in turn to greatermarket share and greater profitability and so forth.

In addition, as Intel opened more fabs in Oregon, Arizona, Israeland Ireland, consistency became a critical issue. Grovelaunched a campaign called "McIntel" in which he stressed justas a McDonalds burger tastes the same anywhere in the world,Intel chips should be manufactured equally well irrespective ofthe location of the fab.

While the manufacturing side of the company was struggling tokeep up, the microprocessor design team was also underenormous pressure to develop the next generation of theirproduct. One team spent months and hundred of thousands ofdollars developing a new processor, called the 8800, only tohave the project canceled when it became clear the new chipwould be uncompetitive against the products under developmentby Zilog and Motorola.

In the end, the 8800 project was canceled, and a small handfulof engineers were assigned to develop the design of what wouldlater be released as the 8086 chip. They were given just 10weeks to design the new chip -- an impossibly short time frame,which turned out to be effective for two reasons:

1. The time frame forced them to avoid the temptation toincorporate features that would be "nice to have". Only thosefeatures which were "essential to have" could be addressedin such a short time period -- resulting in an end product thatwas lean and efficient.

2. The unreasonable deadline forced them to become creativeand instilled the need for discipline into the design effort. Asa result, the new chip’s compatibility with its predecessor wasnot just a good feature if possible -- it was essential.

Whereas the old 8080 chip had been an 8-bit microprocessor,the new chip, the 286, was a 16-bit microprocessor -- providingmore computing performance and allowing far more memory tobe effectively used. The 286 was released commercially in June1978.

‘‘What Intel had produced -- and what the circuit designers andfab people would within the year turn into a finished chip,manufactured and ready for sale to Intel’s customers -- was

Inside Intel - Page 3

something that purists would view as a kludge: an inelegant,botched design. But it fulfilled brilliantly the task it set out to do.It took users of the old, slow 8-bit 8080 into the age of the muchfaster 16-bit processing. It allowed them to build systems with amegabyte of memory. And most important of all, it allowed themto continue to use all the software they had written for the oldprocessor -- with only very minor modifications.’’

-- Tim Jackson

At around this same time, Jerry Sanders, the CEO of AMD,contacted Bob Noyce to discuss the possibility of entering into a"second sourcing" arrangement with Intel. In the electronicsindustry, many companies will refuse to specify a componentwhich is made by just one company, since that would be handingthat company a license to charge above market rates for thecomponent. Therefore, electronic components are much morelikely to be used in other products if a viable second source existsto supply those components.

It made good commercial sense for Intel to provide AMD withthe technical information it required to manufacture Intel’smicroprocessor chips, so just such an agreement wasnegotiated and entered into. The agreement also provided AMDwith the rights to use Intel’s microcode for future chips, and setout the royalties AMD would be required to pay Intel for theproducts manufactured.

However, the AMD second sourcing agreement covered only the8085 chip. When Intel introduced the 286 microprocessor, AMDassumed a similar type of arrangement would be entered into.Intel, however, declined and AMD ultimately aligned itself as asecond source supplier behind Zilog’s Z8000 chip.

6.

‘‘In 1979 Intel was not used to direct competition. During its firstdecade the company had prospered by using its strength inresearch and development to invent entirely new categories ofproducts, and using its ability to master complex processes towin the race to manufacture those products in commercialvolumes. Intel could claim to be the pioneer behind the firstmemory chip, the first DRAM and EPROM and the firstmicroprocessor -- the four most important inventions of thedecade. But the company had little experience in the disciplinesthat occupy most companies: slugging it out with the competitionon cost, quality, service and delivery time. The result of thisinexperience was that when circumstances took Intel’stehnological lead away, the company began to stumble.’’

-- Tim Jackson

The problem first arose with its DRAM product line -- whereasIntel had developed the first 1K DRAM, by the time the 16Kgeneration was released to the market, Mostek beat Intel to themarket. Similarly, in the Microma watch business, Intel wasstunned when the market price of digital watches fell to $9.95retail -- far below the $30.00 retail price point the company hadanticipated. Intel quickly sold its Microma business.

Other problems also soon became apparent. Despite the factthe company had a high caliber sales department, the fact mostof the senior executives were engineers strongly influenced thecompany’s approach to the sales process. Andy Grove, inparticular, would be called on sometimes to visit importantcustomers. His idea of a social call was to march in, launch intoa point-by-point comparison of features with competitiveofferings, plonk a sample on the desk, tell them what the unit

price would be and then ask how many they wanted. And if thecustomer wasn’t interested or wanted to ask more questions, hewould cut them off and move on to the next call. These types ofapproaches were not always very effective in the real world.

The one thing that underscored Intel during the 1970s, however,was that this was an era of production constraints. In reality, thecompany was rarely able to manufacture enough of any of itsproducts to completely satisfy demand. Therefore, the companydidn’t need to focus on sales too rigorously -- until the Motorola68000 chip started outselling Intel’s 8086 by a significant margin.

Alarmed by this, Intel pulled together a task force who developeda campaign that was called "Operation Crush".

‘‘Are we doing this as an exercise to improve our position in themarketplace and achieve the recognition we deserve? Or are wedoing this to kill Motorola? Because the other guys are so stupidthey won’t make any difference at all. The Z8000 is nothingbecause Zilog doesn’t know what it’s doing. The 16016 is evenmore of a joke because National doesn’t know what it’s doing.There’s only one company competing with us, and that’sMotorola. The 68000 is the competition. We have to kill Motorola,that’s the name of the game. We have to crush the f......ingbastards. We’re gonna roll over Motorola and make sure theydon’t come back again.’’

-- Jim Lally, Intel executive

Since the 68000 chip was actually better than the 8086, Intelfocused on five points of differentiation:

1. Intel’s reputation as the technology leader.

2. A more complete Intel product family.

3. A focused and well trained technical sales force.

4. Better performance at the system level.

5. Better customer service and support.

Operation Crush was a huge success. With Intel’s sales staff andsales engineers out in the field generating enthusiasm, sales ofthe 8086 chip leaped ahead. Motorola’s market share fell to15-percent, while Zilog and National were effectively squeezedout altogether.

7.

Gordon Moore, one of Intel’s co-founders, was famous forarticulating in 1965 what is now known as Moore’s Law: That thepower and complexity of the integrated circuit would doubleevery 18 months.

This law allows planners to forecast when it will make economicsense to put microprocessors into consumer products. It alsohas direct applicability to the computer industry. Yet despite this,for many years, Intel refused to enter the personal computerindustry itself -- focusing instead on simply supplyingmicroprocessors and memory units to third party manufacturers.

The one computer company it had never managed to make asale to was IBM, who always insisted on using its own proprietarytechnology. Intel’s engineer in Florida, however, made a call atIBM’s development facility in Boca Raton, Florida, and wasintroduced to Don Estridge.

Unbeknown to Intel, a secret development program wasunderway at IBM to develop a personal computer. The originaldesign used the Motorola processor, but that was running lateso Don Estridge was putting together a second version of thePC, and he was using an Intel 8088 chip. (This was a variant of

Inside Intel - Page 4

Intel’s 8086 chip which used an 8-bit bus rather than the 16-bitbus used in the 8086).

The first indication IBM might be serious about using the Intelchip came when they demanded Intel arrange a second sourcefor the 8088 at the end of 1980. Intel realized what IBM wasdoing, and also concluded AMD was the only viablesecond-source supplier available.

When AMD was approached, they too realized what IBM was upto and they negotiated a second sourcing arrangement with Intelthat covered not only the 8088 chip but also future chipsdesigned by Intel over the next ten years. The deal was finallyexecuted in February 1982, just six months after the release ofthe IBM personal computer.

As a result of the IBM personal computer, Intel’s sales ofmicroprocessors would increase 1,000-fold, AMD would flourishfor a while, Motorola’s 68000 chip would move to the Apple Mac,while Zilog and National would be squeezed out of themicroprocessor market altogether. With the release of the 80286chip for the IBM PCAT machine in late 1982, Intel continued toincrease its share of the personal computer microprocessormarket.

The second sourcing arrangement entered into between Inteland AMD would later come to form the basis of the longest andmost bitter contractual dispute ever fought in the semiconductorindustry. In essence, Intel used AMD as a second source for the80186 and 80286 generation chips. Every computermanufacturer then naturally assumed AMD would also havesecond source rights for the 80386 chip, and therefore they didnot worry too much about negotiating a definitive second sourceinto their supply agreements. Then, just before the 80386 chipwas released, Intel announced there would be no secondsource. By that time, there was so much market interest andmomentum that the computer manufacturers were forced to usethe 80386 chip anyway. Intel, therefore, could charge a higherprice, because there was no alternative supplier. This strategyadded hundreds of million of dollars in additional revenue forIntel, which impacted directly on the company’s bottom line.

8.

While Intel’s microprocessor business was forging ahead, itsmemory business was in serious trouble. Despite the fact Intelhad an impressive history of being the innovators in memorytechnology, the Japanese were proving to be far better atmanufacturing. In fact, by 1979, Intel had a less than 5-percentmarket share of the memory components market.

While Intel’s senior management had not really come to termswith the dramatic changes taking place in the marketplace, thecompany’s line managers were acutely aware of the problem.As a result, the line managers were actually shifting thecompany’s resources away from memory chips (which sold ataround $10 to $20 each) and into microprocessors (whichtypically sold for several hundred dollars each). Therefore, evenwithout a management decision being explicitly taken, Intel wasfocusing more and more on its microprocessor line of business.

The final straw for the memory business came when Fujitsu beatIntel to the market with a 64K DRAM memory chip by more thanthree years. Being aware of the cost dynamics of the business,Intel realized that by the time they could match Fujitsu, theJapanese company would have marched so far down the costcurve Intel would never be able to catch up. The battle was

effectively over even before Intel could bring its own offering tofruition.

While Intel tried to used a cross product strategy to stay in thememory market -- sending subtle signals the best supplies of thenewest microprocessors would only go to those manufacturerswho used Intel memory chips -- this was only a stallingmaneuver. In addition, the Japanese were totally committed toa fight to the death -- so much so that one company instructedits sales staff to undercut any Intel bid by 10-percent irrespectiveof what Intel bid, and if Intel were to requote, the Japanesecompany should again bid 10% lower than Intel the next timeround.

The Japanese companies were prepared to fight a price warsince memory chip sales represented a much smaller proportionof their total revenues that the business did for Intel, TexasInstruments or National Semiconductor.

‘‘Intel, by 1985, faced a fundamental issue -- should it continueto compete in the DRAM market? The trouble was that almostnobody in the company, from Gordon Moore and Andy Grovedownwards, was able to approach the matter in an objective way.Intel’s association with the DRAM was too long and too close.The company had created the memory-chip market, and thememory-chip market had created the company. To abandonmemories in order to focus on the more profitablemicroprocessor business would be a betrayal of the company’sroots. Yet what were the alternatives?’’

-- Tim Jackson

‘‘Gordon Moore and I were discussing our quandary. Our moodwas downbeat. I looked out the window at the Ferris wheel of theGreat America amusement park revolving in the distance, thenI turned to Gordon and asked, "If we got kicked out and the boardbrought in a new CEO, what do you think he would do?" Gordonanswered without hesitation, "He would get us out of memories".I stared at him, numb, then said, "Why shouldn’t you and I walkout the door, come back and do it ourselves?"’’

-- Andrew Grove

Within two years, Intel had completed the move out of thememory business. If it had made the move five years earlier, thecompany would have been far stronger. As it was, Intel was in adrastic financial position in 1984 -- forcing the company to reduceits head count from 25,400 employees in 1984 to 18,200 in 1986.Intel’s handling of the change of direction may have beengrudgingly undertaken, but at least the company had survived.

However, the mid-1980s were also the scene of one impressiveIntel victory. Andy grove came up with the idea of using copyrightlaw rather than patent law to protect the design of the masksused in Intel’s microprocessor chips. To apply copyright law inthis field, Intel had to get a change made in the U.S. federalcopyright law, extending it to the mask design. After a numberof years work, this initiative was successful when U.S. copyrightlaw was changed in 1984 to include protection for the intellectualproperty embodies in these masks.

In an important court case between Intel and NEC, themicrocode for Intel’s microprocessors was also found to beprotected by copyright law. This was an unexpected bonus forIntel -- it meant the company could pursue the clonemanufacturers with the full force of copyright law behind it. Thiswould later prove to be a highly valued avenue for Intel’s intensecommercial battles.

Inside Intel - Page 5

9.

‘‘We’re pleased to report 1986 is over. It was, without question,the toughest year in Intel’s history, filled with plant closings,layoffs and deep losses.’’

-- Intel’s 1987 Annual Report

In fact, in 1986, Intel’s net revenues had dropped to $1.27 billion,compared to a 1984 peak of $1.6 billion. A $198 million profit in1984 had turned into a $173 million loss by 1986. On the otherside of the ledger, however, the company had maintained itsR&D expenditure, and now had the 386 chip ready to bereleased.

The only problem was that IBM was dragging its heels onreleasing a new generation of personal computers. Fortunatelyfor Intel, however, Compaq was keen to seize the initiative, andin October released its 386 powered computer -- generating ahuge amount of consumer interest. In turn, Intel used thatconsumer demand to price the 386 chip at $150 -- compared tothe $40 per chip Intel received for each 286 chip it sold.

The fact that IBM was now just one of a number of computermanufacturers instead of the clear standard setter had the effect(from Intel’s perspective) of bringing into clearer focus theoperating system monopoly enjoyed by Microsoft. It came downto Intel’s monopoly over the microprocessor versus Microsoft’smonopoly over the operating system. And, to make matters moreinteresting, each side claimed the other was to blame for thesluggish performance of the PC by comparison with the AppleMac.

‘‘With this mistrust between the two companies lying in thebackground, it was no wonder the fading of IBM as a force in thecomputer industry coincided with the beginnings of out-and-outhostility between Microsoft and Intel. From Bill Gates’s point ofview, the PC business was now a zero-sum game between thetwo firms. Every dollar that Intel could extract from the buyer ofa PC would be a dollar out of Microsoft’s pocket -- and viceversa.’’

-- Tim Jackson

The fact that Intel had ended all second sourcing agreementswith the 386 chip meant the company’s margins on the chipswere healthy, and this single factor more than anything else,underwrote Intel’s progress back towards profitability again. Intelstrengthened its competitive advantage in this field, aggressivelygoing after any third party which attempted to carve out a shareof the microprocessor market for itself time an again.

AMD did, in fact, attempt to force Intel to provide it with secondsource manufacturing rights, alleging its previous agreementswith Intel made that provision. A 3-year attempt at arbitrationeven came down slightly on AMD’s side of the discussion.However, by the time Intel had appealed the decision to theCalifornia State Supreme Court, another year had passed. Bythat time, the company was moving on to new technology, andsecond sourcing rights for the 386 chip was of no realcommercial significance any longer.

It wasn’t just the established companies that attracted Intel’sattention either. When ULSI Systems set up business in 1987attempting to develop a 387 math coprocessor chip that wouldwork with Intel’s 386 microprocessor, Intel took legal actionpreventing ULSI from commercially releasing their 387 chip forfive years. By the time the legal action was thrown out, five yearshad passed without ULSI being able to sell its product.

Around this same time, there was also a change in themanagement structure of the company. Despite the fact he hadbeen running the company on a day-to-day basis since itsinception, Andy Grove was still a vice-president. Therefore, inApril 1987, Grove was officially appointed CEO, with GordonMoore retaining his title as chairman of the company.

‘‘In March 1991, after Intel had been promoting quality and costas one of its key corporate objectives for six years, the companyparticipated in a "benchmarking" exercise where it sharedconfidential data from its chip fabrication activities with other bigsemiconductor companies around the world. The resultsshowed that on overall manufacturing costs, the top two tiers ofparticipants consisted of Japanese companies, the Koreanscame next and Intel last, on a par with Taiwan. Intel was alsoslower at building factories than average, and slower to rampproduction of a new product up to the desired level. On each newprocess, Intel took two more years to achieve or approach thesame yields as the competition. When you paused for thought,it wasn’t hard to see why this was. With its dominance of the x86processor market, Intel was making so much money that it didn’tneed to be an efficient producer.’’

-- Tim Jackson

10.

By 1988, Microsoft was aggressively promoting the virtues of anew microprocessor technology known as RISC -- standing for"reduced instruction set computing". The theory was that mostof the instructions for a chip like the 386 aren’t used very often.Therefore, a RISC chip could perform a limited number offunctions exceptionally quickly, but would struggle on others. Onbalance, though, for most tasks, a RISC chip would out performa 386.

The only disadvantage, from Intel’s perspective, was that a RISCchip would be incompatible with any software that had beendeveloped for the x86 generation of processors. This lack of"backward compatibility" was seen as an essential point to theprotection of the Intel franchise. Microsoft, on the other hand,aggressively promoted RISC as the way to move forward.

Undetered by Intel’s concerns, Microsoft developed theWindows NT operating system -- "NT" standing for newtechnology -- as an ideal operating system for RISC processors.Microsoft also formed a 21-member industry group, called ACE-- standing for Advanced Computing Environment -- to promotethe RISC technology as the way forward.

Most Intel staff were quite offended by the formation of the ACEgroup, but surprisingly, Andy Grove was reasonablyphilosophical.

‘‘You guys should think about it differently. They’re not yourwives. They’re in business for their own vested interests, and weshould never assume that their interests match ours. We haveto give them room to be different. Think of Microsoft more like afellow traveler on a train. They guy gets on, and he’s going in thesame direction as you. When you’re in the train, you’re going tobe civil. But you’re not going to give him the keys to your house,and you’re not going to give him your credit cards. You know thatyour paths will diverge at some point. It doesn’t pay to put toomuch trust in him.’’

-- Andy Grove

Inside Intel - Page 6

To illustrate the point, Intel undertook its first consumeradvertising campaign in October 1989. Known as the "Red X"program, this $4 million campaign had advertisements with thenumber "286" with a huge red cross painted over the top, withthe number "386" and the letters "sx" written instead.

It was the first time in Intel’s history the company had everattempted to make the public -- rather than computermanufacturers -- aware of Intel’s products. This campaign wascontroversial at the time -- partly because Intel was using it totest the then new idea that consumers would make theircomputer purchase decisions based on what kind of processorwas inside the machine.

The advantage of this program to Intel was clear cut. If consumerdemand for Intel processors could be built up, Intel could escapethe tyranny of being dictate to by computer manufacturers andtake the initiative by pitching its products directly to theconsumer. That consumer demand would then be an incentivefor manufacturers, irrespective of any other factor.

Of course, Intel was only willing to take this approach becauseno other company manufactured 386 processor chips. Itsmargins, therefore, were higher for the 386 chip than for the 286.Therefore, it make good economic sense for Intel to try and moveas many computer users as possible to the 386 chip.

‘‘Eventually, the significance of the Red X began to dawn on thePC industry: if it increased demand for new computers, then thesales of the entire industry would be stimulated. In the end,therefore, Red X proved to be good news even for the retailerswho had badmouthed it at the outset.’’

-- Tim Jackson

When other processor chip manufacturers like AMD startedusing 386 in the name of their chips, Intel promptly took them tocourt for trademark infringement -- and lost. The judge ruled thenumber 386 had become so prevalent throughout the industrythat it had effectively become a generic name. While thisdecision (coming in March 1991) was too late to have any effecton the 486 chip, by the time the next chip was released by Intelin 1995, it was called the Pentium rather than the 586. And, mostimportantly, Pentium was a trademark belonging to Intel.

Coinciding with the new chip naming convention, Intel alsolaunched its $20 million "Intel Inside" campaign in May 1991.Prior to then, all of the company’s marketing had been focusedon engineers as the target audience. That type of marketing wasperformance based and information intensive. The Intel Insideconsumer marketing, by contrast, stressed the cutting edgenature of Intel technology in 30-second television ads,supplemented by various print ads.

‘‘The paradox behind the campaign was that it was advertisinga range of products that almost no consumers were expected tobuy themselves. Unless you were a technology-savvy computerjock who knew how to open up the box of a PC and upgrade themotherboard, the chance that you would ever buy an Intelproduct was minimal. Instead, the focus was entirely on puttingconsumers in a frame of mind so that when they were nextbuying a computer, it would be the Intel name, and the IntelInside logo, that would provoke the spark of recognition and reliefthat often precedes a buying decision. Intel Inside had itsprecedents: Dolby noise reduction systems, which could not bebought separately but were installed inside cassette decks;Teflon non-stick material, which could only be bought with afrying pan attached; Nutrasweet, the sweetener that came inmany soft drinks. But Intel added a twist of its own. The company

launched a partnership marketing campaign, in which it sharedthe cost of computer companies’ own advertisements if theyincluded the Intel Inside logo in the ad copy with appropriateprominence’’

-- Tim Jackson

Meanwhile, Intel continued to make good progress. This wasreflected in the company’s share price -- which had risen from$23.50 at the time of Intel’s IPO to $4,385 a share in June 1993,taking into account stock splits. That made share options a keystaff motivator -- engineers who had received 1,000 stockoptions at the IPO had shares worth $5 million by 1993 if theyhad held on to them. And Intel had a very wide employee stockoption program -- with more than 5,000 participants by 1984.

As Intel approached its 25th birthday in July 1993, sales werearound $6 billion per year, generating annual profits in excessof $1 billion. The company now began to introduce furtherinnovations, such as "clock speed doubling" -- which increasedthe internal processing speed of the processor -- and a sparesocket on motherboards for "over drive processors" to beinstalled in the future. Both of these initiatives were designed toincrease the number of PC systems sold worldwide, therebyincreasing the sale of Intel chips.

Despite all its market success, Intel continued to pursue anaggressive approach against competitors, using every tool in itslegal arsenal to discourage competitors.

‘‘There are only two kinds of companies: the quick and the dead.’’-- Andrew Grove, Intel Annual Report

With the release of the Pentium chip in 1993, Intel also movedinto additional lines of business that were closely aligned with itstechnical expertise. For example, instead of solely providingprocessor chips, the company also started offering computermanufacturers motherboards into which the chips could beplaced. Within three years, Intel was shipping 25 millionmotherboards per year -- basically eliminating a number of smallcompanies that had traditionally offered this product to computermanufacturers, and thereby attracting the attention of U.S.anti-trust investigators.

In October 1994, a customer (who happened to be from thedepartment of mathematics at Lynchburgh College, Virginia)called Intel’s tech support department to tell them his Pentiumseemed to be making an error when doing some numbercrunching. The Intel staff member who took the call attemptedto brush him off with the explanation it must be some other partof the computer system causing a problem.

The only problem was Intel had actually been aware for aboutfive months of the problem, and that the Pentium chip was flawed-- the result of five wrong entries on an internal data table withinthe chip that contains 2,048 entries. As a result, the chipincorrectly calculated the answer to a division problem onceevery 9 billion calculations.

Bugs in new processor chips were not uncommon occurrences,and they were routinely corrected as the processing run wenton. The difference on this occasion, however, was that Intel wasjust in the process of attempting to transform itself from acomponent company to a global brand. Therefore, it couldn’thandle the problem like an engineer would approach it -- Intelinstead had to think like a consumer goods supplier.

When nobody from Intel called the mathematics professor back,he started sending out e-mail to his colleagues, and postingnotices on the Internet. When Intel found out, they sent a couple

Inside Intel - Page 7

of new Pentiums to the professor, but it was too little too late.When the specialist trade publications started running articles,the broader general media weren’t too far behind. Intel had a fullscale public relations disaster on its hands.

Intel’s board of directors met and decided to provide replacementprocessors for every one that requested one -- at a cost to Intelof $475 million, around half the company’s income for thatquarter.

‘‘The crisis demonstrated Intel’s financial strength compared withthe rest of the manufacturers serving the market. Of allmicroprocessor vendors, it was the only one that could remotelyafford to pay for an error in a product that had shipped to themass market on this scale. The guarantee that Intel eventuallycame up with -- grudgingly, belatedly and against its betterjudgment -- was the strongest proof it could give there wasmeaning behind the phrase, "Intel Inside". That proof mighteasily have been given at one-tenth of the cost, or even less, ifonly Andy Grove had been willing to change his mind a fewweeks earlier. But the point was made all the same.’’

-- Tim Jackson

11.

Even with the Pentium chip flaw behind it, Intel still facesaggressive competitors from all quarters. IBM, Motorola andApple entered into an alliance to develop an entirely new chip,the PowerPC 601, which offered Pentium class processingpower at half the cost. Fortunately for Intel, the new chip failedto live up to its own hype.

Despite all its success, Andy Grove still stresses a "back tobasics" approach for Intel. He has refused to allow workers toget into telecommuting, flexible working hours or any othereasing of discipline. In fact, he recently reintroduced the "late list"sign in process -- under which any employees who are not atwork by 8 am are required to sign a register when they do arriveat work.

The company continues its research and development programsat an impressive pace. Parallel development teams are alreadyworking on the next two generations of new processor chips.Ongoing refinements are being made to the manufacturingprocesses on a regular basis, as the company brings to themarket ever more powerful chips.

The threat of RISC has receded somewhat as Intel hasdemonstrated the performance gains that are available throughfurther enhancements of the x86 architecture.

‘‘After years of focus on keeping interlopers away from its x86monopoly, the recognition was beginning to permeate throughthe company that the key to its future growth would lie less inpushing its x86 competitors further and further into their cornersthan in increasing the growth rate of the entire computer market.’’

-- Tim Jackson

Therefore, Intel has settled many of its long running legal battles,and Intel now tends to take a broader view of its mission -- tomake the PC as attractive a technology as possible. Intel hasclearly focused itself as the technology leader, who will givedirection to the industry as a whole as it moves towards aprosperous future.

In that vein, Intel has mended loads of bridges lately -- includingforming a strategic alliance with Microsoft. There is still tension

in the alliance, and from time to time the company’s do get outof step with each other, but by and large the apparent powerstruggle for the mantle on leadership has been tempered to amore tolerable level.

These changes don’t mean Intel is now a push-over. Far from it.The company still is very aggressive in the way it operates itsbusiness. But the old "us against the world" fortress mentality ofthe old Intel has now been molded into a senior industry partyand technology leader style role -- in which Intel’s most attractiveassets are given full opportunity to bloom.

The arrival of the Internet again threatens the role of Intel in theyears ahead. How the company will respond to the challengesand opportunities generated by the Internet remains to be seen.

Intel’s Merced chip, the successor to the Pentium pro, also offersother challenges for Intel to navigate safely through. Merced isthe first chip to significantly alter the architecture Intel uses forits chips. While the new chip is backwards compatible with theentire x86 processor family, it also introduces a new designaround which future growth in processing power can be built.Future generations of Intel’s chips will further build on andenhance the new architecture.

‘‘Andy Grove’s legacy that he leaves behind him is visiblethroughout the world, in hundreds of millions of personalcomputers powered by Intel processors. Intel may be second toMicrosoft in its power in the industry, but it is considerably moreprofitable -- and also more profitable than its customers puttogether. As well as delivering spectacular returns toshareholders, Intel has also done well by its employees. Thecompany does no publish figures of how many millionaires it hascreated, but the doubling of Intel’s stock price that began inmid-1966 has meant that scores of middle managers inside thecompany, and possibly hundreds, have made more than enoughmoney to retire on. The decision to extend the company’s stockoption scheme to all its employees in 1997 came after aspectacular year in which Intel paid out profit-sharing andretirement contributions of $820 million to its staff.’’

-- Tim Jackson

‘‘Despite these risks, only brave souls will be ready to bet againstIntel. With its dominant position in the industry, its massiveinvestments in manufacturing capacity and a managementsystem that helps it respond to change faster than manycompanies of its size, Intel doesn’t look too vulnerable.’’

-- Tim Jackson

© Copyright 1999. All Rights Reserved.

Summaries.Com

Inside Intel - Page 8