You are on page 1of 34

Strategy as Vector and the Inertia of Coevolutionary Lock-in

Author(s): Robert A. Burgelman


Source: Administrative Science Quarterly, Vol. 47, No. 2 (Jun., 2002), pp. 325-357
Published by: Johnson Graduate School of Management, Cornell University
Stable URL: http://www.jstor.org/stable/3094808
Accessed: 10/12/2010 05:33

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=cjohn.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Johnson Graduate School of Management, Cornell University is collaborating with JSTOR to digitize, preserve
and extend access to Administrative Science Quarterly.

http://www.jstor.org
Strategy as Vector and To examine the consequences of a period of extraordi-
the Inertia of nary success for the long-term adaptive capability of a
firm's strategy-making process, this comparative longitu-
Coevolutionary Lock-in dinal study of Andy Grove's tenure as Intel Corporation's
chief executive officer (CEO)documents how he moved
Robert A. Burgelman Intel's strategy-making process from an internal-ecology
StanfordUniversity model to the classical rational-actor model during
1987-1998. His creation of a highly successful strategy
vector pursued through an extremely focused induced-
strategy process led to coevolutionary lock-in with the
personal computer market segment, in which Intel's strat-
egy making became increasingly tied to its existing prod-
uct market. Intracompany analysis of four new business
development cases highlights the inertial consequences
of coevolutionary lock-in. The paper examines implica-
tions of coevolutionary lock-in in terms of its effect on
balancing induced and autonomous strategy processes
and exploitation and exploration in organizational
learning.?
There is a vast literatureascribingthe success of a company
to the vision, strategy, and leadershipapproachof its chief
executive officer (CEO).Some of these accounts put the
CEOat center stage (e.g., Welch, 2001); others put him or
her more modestly in the background(e.g., Collins,2001).
Organizationaland strategic management researchers, how-
ever, have long highlightedthe difficultiesleaders encounter
in aligningorganizationalaction in the pursuitof strategic
intent (e.g., Mintzberg,Ahlstrand,and Lampel, 1998). Recent
work in organizationalecology (e.g., Barnettand Hansen,
1996), the behavioraltheory of the firm (e.g., Levinthaland
March,1993), and neo-institutionaltheory (e.g., Zuckerman,
2000) continues to illuminatethe external and internallimita-
tions facing top management. Yet we still understandlittle
about why some firms have periods of extraordinarysuccess,
what the role of the CEOis in heraldingand leadingthe orga-
nizationthroughsuch periods, and what the consequences
are of such periods for strategy makingthereafter.While
organizationalresearchers are mostly concerned with ordi-
narystates and expect regression toward the mean to wash
out fluctuationsover time, periods of extraordinarysuccess
?2002 by Johnson GraduateSchool, have potentiallyimportantconsequences for the strategy-
CornellUniversity.
0001-8392/02/4702-0325/$3.00. makingprocess as a long-termadaptiveorganizationalcapa-
bility,that is, spanning multiplegenerations of CEOs.
The researchfor this paperwas critically Longitudinalfield-based research on strategy makingat Intel
dependenton my case writingand teach- CorporationduringAndy Grove'stenure as CEOoffered the
ing collaborationwith AndrewS. Grove, opportunityto study a periodof extraordinarycorporatesuc-
Intel'sformerCEOand currentchairman,
since 1988. The generous collaborationof cess and its consequences for the company's strategy-mak-
IntelCorporation and its managersin this ing process. Intelseemed a particularlyinteresting research
researchis much appreciated.James G.
Marchand EzraW. Zuckermanoffered site because it is one of the most importantfirms of the digi-
many useful comments on an earlier tal age (Gilder,1989; Isaacson, 1997), and its evolution high-
draft.I am especiallyindebtedto Jim
Marchfor drawingattentionto coevolu- lights the fundamentaltechnologicaland economic forces
tionarylock-inas a phenomenonof orga- that characterizedigitalindustries(e.g., Arthur,1987). The
nizationaladaptation.The ideas of this research could be used to compare Grove'sstrategy-making
paperwere significantlysharpenedas the
resultof challengingcomments and approachto that of his predecessor (GordonMoore)and suc-
queries of the ASO reviewersand of cessor (CraigBarrett)and thus could examine his efficacy as
ChristineOliver,its editor.Thanksalso to
the managingeditor,LindaJohanson,for CEOwithin the context of Intelas an evolving system over
helpfuleditorialsuggestions. time.
325/Administrative Science Quarterly, 47 (2002): 325-357
Andy Grovesucceeded GordonMoore as CEOin 1987 at the
time that Intelwas recoveringfrom defeat in its original
semiconductormemory business and refocusingon its
microprocessorbusiness (Burgelman,1994). He held the
position untilearly 1998. Between 1987 and 1998, Intel
became the clear winner with its microprocessorsin the per-
sonal computer (PC)marketsegment. Intel'srevenues grew
from $1.9 billionto $25.1 billion-an increase of 29.4 percent
per annum-and net income grew from $248 millionto $6.9
billion-an increase of 39.5 percent per annum. In 1998,
however, Intel'sgrowth in the core business slowed down
significantly.Also, it had become clear that new business
development was relativelyunsuccessful duringGrove's
tenure as CEO.In 1997, CraigBarrett,then Intel'schief oper-
ating officer (COO),observed that Intel'score microprocessor
business had begun to resemble a creosote bush, a desert
plantthat poisons the groundaroundit, preventingother
plants from growing nearby.The creosote bush metaphor
raised potentiallyinterestingquestions about the strategic
consequences of Intel'sabilityto dominate in the PC market
segment. It drew attentionto the phenomenon of coevolu-
tionarylock-in:a positive feedback process that increasingly
ties the previoussuccess of a company's strategy to that of
its existing product-marketenvironment,thereby makingit
difficultto change strategic direction.Despite the attention
given to winner-take-allcompetition in digitalindustries(e.g.,
Arthur,1987) and the role of inertiain organizationaland
industryevolution (e.g., Hannanand Freeman, 1977, 1984),
researchers have paid littleattentionto how coevolutionary
lock-incomes about and may become a significantsource of
strategic inertia.This study addresses this gap. It seeks to
shed lighton the role of the CEOin creatinga strategy-
makingprocess that leads to coevolutionarylock-inand what
its implicationsare for organizationaladaptation.
Grovedescribed his approachas "vectoring"Intel'sstrategy-
makingprocess. Vector-a quantityhavingdirectionand mag-
nitude, denoted by a line drawnfrom its originalto its final
position (OxfordEnglishDictionary)-seems an apt metaphor
to describe his efforts to align strategy and action. By creat-
ing a strategy vector, Grovewas able to drive Intelin the
intended directionwith a total force equal to all the forces at
its disposition.The paper examines the long-termadaptive
implicationsof Grove'sstrategic leadershipapproach,which
seemed to approximatethe classical rational-actormodel
(Allisonand Zelikow,1999; Bendorand Hammond,1992),
and contrasts it with that of his predecessor.
COEVOLUTIONARY LOCK-ININ FIRMEVOLUTION
Informedby evolutionaryorganizationtheory (e.g., Aldrich,
1999; Baum and McKelvey,1999), earlierresearch on Intel
before Grovebecame CEOsuggested that effective strategy
makingmay be as much about creatingan environmentin
which middle management makes strategic decisions as it is
about strategy makingin the classical sense and that the role
of top management might be to recognize transitionsrather
than to initiatethem (Burgelman,1994). These findingswere
consistent with an internalecology model of strategy making,
which was conceptualizedin terms of inducedand
326/ASQ, June 2002
Coevolutionary Lock-in

autonomous strategy processes (Burgelman,1991). Induced


strategy exploits initiativesthat are withinthe scope of a
company's currentstrategy and that extend it furtherin its
currentproduct-marketenvironment.Autonomous strategy
exploits initiativesthat emerge throughexplorationoutside of
the scope of the currentstrategy and that providethe basis
for entering into new product-marketenvironments. Intel's
strategy makingbefore Grove became CEOresembled an
internal-ecologymodel in which induced (memory-related)
and autonomous (microprocessor-related) initiativescompet-
ed for the company's scarce resources based on their suc-
cess in the externalcompetitive environment.This paper
documents how Grove'ssuccessful strategy vector created a
highlyfocused induced-strategyprocess, which moved Intel's
strategy makingaway from the internal-ecologymodel and
closer to the rational-actormodel. It shows how positive envi-
ronmentalfeedback associated with the successful strategy
vector caused coevolutionarylock-inand how this can illumi-
nate time-paced evolution (Gersick,1994; Brown and Eisen-
hardt,1997) and the dynamics of competitive intensity (Bar-
nett, 1997).
Strategic Inertia of Coevolutionary Lock-in
This paper'sdetailed ethnographicdata also document new
sources of strategic inertiathat may be the unintendedcon-
sequence of coevolutionarylock-in.Systemic sources of iner-
tia associated with coevolutionarylock-inprovideadditional
insight into structuralinertia(Hannanand Freeman, 1984).
They help elucidate the dynamics of the evolving relativeeffi-
ciency of internalselection (Miller,1999; Lovas and Ghoshal,
2000) and externalselection (Sorenson, 2000), as a compa-
ny's product-marketenvironmentmatures, and of the rate
and directionof innovationrelativeto environmentalevolution
as firms grow large (Sorensen and Stuart,2000). Study of the
psychologicalsources of inertiaassociated with coevolution-
ary lock-incan be used to assess Prahaladand Bettis's (1986)
contention that executives become ingrainedwith beliefs
about causes and effects that may not hold after the environ-
ment changes. And they help sort out Audia,Locke, and
Smith's (2000) argumentthat success tends to increase deci-
sion makers' feelings of self-efficacy from that of Millerand
Chen (1994), who suggest that it causes complacency,
understoodas driftingwithout furtherattempts at improve-
ment. These psychologicalsources of strategic inertiadraw
attentionto the potentiallimitationsof evolution guided by
the strategic intent of the CEO(Lovasand Ghoshal,2000).
Most importantfor the purposes of this paper,the various
sources of strategic inertiaassociated with coevolutionary
lock-inhave implicationsfor maintaininga balance between
induced and autonomous strategy processes and between
exploitationand explorationin organizationallearning.They
help connect these ideas, which are rooted in evolutionary
organizationtheory (Burgelman,1991; March,1991), with
related ideas of the modern economic theory of the firm
(Rotembergand Saloner,1994, 2000).
327/ASQ, June 2002
RESEARCHMETHOD
The research reportedin this paper is partof a longitudinal
multistage, nested case study design (e.g., Yin,1984;
Leonard-Barton,1990) focused on majorperiods of Intel's
history(Burgelman,2002). These include Epoch I: Intelthe
memory company (1968-1985); Epoch II:Intelthe micro-
processor company (1985-1998); and Epoch III:Intelthe
Internetbuilding-blockcompany (beyond 1998). These three
epochs correspondroughlyto the tenure of GordonMoore,
Andy Grove,and CraigBarrettas Intel'sCEOs.
Data Collection
Interview data. Forthis paper,which focuses on Intel's
Epoch II,I used data from 63 informants,collected mostly
throughinterviews I and/ora research associate conducted
and throughinformalinteractions.Informalinteractionssome-
times involveda research associate. Others took place in the
strategic long-rangeplanningsessions I observed, executive
education sessions I taught for senior Intelexecutives, and
workingwith Intelstaff in preparingfor executive education
sessions. I also had access to transcriptsof interviews con-
ducted and tape-recordedby Intelconsultants. The list of
these informantsand their position in the organizationis pro-
vided in table 1. Managersfrom differentlevels, different
functionalgroups, and differentbusinesses were involved.
Throughoutthe research period, I used informaldiscussions
with many currentand former Intelemployees to corroborate
data obtainedfrom the formalinterviews. Most interviews
lasted between one and two hours and focused on key
events, people, and issues. Keyevents involved,for instance,
the introductionof successive generations of microproces-
sors. Key people were individualsor groups from different
functionalareas or differenthierarchicallevels who made crit-
ical decisions or made proposalsthat, while not necessarily
implemented, triggeredhigh-levelreconsiderationof strategic
issues. Key issues included,for instance, how to allocate
resources to differentbusinesses, how to resolve internal
competition between differentmicroprocessorarchitectures,
and how to enter into new businesses. Most interviews
were not tape-recorded(exceptions are listed in table 1,
below), but the interviewersmade extensive notes. Manyof
the interviews were done together with research associates.
Transcriptsof the research associates' notes showed agree-
ment on the substantive content of the interviews.This pro-
vided some confidence that the data were validand reliable.
Archival data. Archivaldata, such as documents describing
the company's history,annualreports,and reportsto financial
analysts, were obtainedfrom Intel.Additionalarchivaldata
were obtainedfrom outside sources, such as industrypubli-
cations and financialanalysts' reportsand business press arti-
cles about Inteland the semiconductorand computer indus-
tries. The archivaldata could be juxtaposedto the interview
data to check for potentialsystematic biases in retrospective
accounts of past strategy.
Case teaching as a data source. The interviewand archival
data were used to write several case studies about the role
of strategy makingin Intel'sevolutionduringthe periodthat
328/ASQ, June 2002
Coevolutionary Lock-in

Table1
Informants Providing Data Concerning Epoch II(1988-1998)
Informal
Name and most relevant job during Epoch II Interview Interaction
1. GordonMoore,chairman X X
2. AndyGrove,CEO X X
3. CraigBarrett,COO X X
4. GerryParker,executive VP,Technologyand Mfg. Group X X
5. PaulOtellini,executive VP,IntelArchitectureBusiness Group X X
6. FrankGill,executive VP,IntelProductsGroup,gen. mgr.Networking X X
7. Les Vadasz,senior VP,CorporateBusiness DevelopmentGroup X X
8. AlbertYu,senior VP,MicroprocessorProductsGroup X X
9. RonWhittier,senior VP,IntelArchitectureLabs,ContentGroup X X
10. AndyBryant,senior VP and CFO X X
11. Sean Maloney,senior VP,Sales and MarketingGroup X
12. Dennis Carter,VP,CorporateMarketingGroup X X
13. Ron Smith,VP,gen. mgr.Chipsets X X
14. PatrickGelsinger,VP,gen. mgr.ProShare X X
15. MikeAymar,VP,Desktop ProductsGroup,Hood River X X
16. MarkChristensen,VP,gen. mgr.Networking(late 1990s) X X
17. John Miner,VP,EnterpriseServerGroup X
18. Hans Geyer,VP,gen. mgr.FlashProductsDivision X
19. Patty Murray,VP,HumanResources X
20. HaroldHughes, VP and CFOmid-1990s X X
21. John Davies, VP,ConsumerMarketingDesktop Prod.Grp.,Hood River X
22. AvramMiller,VP,CorporateDevelopmentGroup,Hood River X
23. Jim Johnson, gen. mgr.PC EnhancementOrganization(late 1980s) X X
24. ClaudeLeglise, MarketingDirectori860 (late 1980s) X X
25. Steve McGeady,gen. mgr. Home Media Lab(mid-1990s) X X
26. Scott Darling,gen. mgr.Busin.Com. Prod.Grp.,ProShare(late 1990s) X X
27. SandraMorris,managerIntelProd.Grp.(mid-1990s) X X
28. TomYan,mgr.development OEMProd.and Syst. Div.,Hood River X
29. DickPashley,gen. mgr.FlashMemoryDivision(early1990s) X X
30. WarrenEvans, Business Process Network,Planning X X
31. Renee James, technicalassistant to AndyGrove(mid-1990s) X X
32. KatherineYetts, technicalassistant to CraigBarrett(mid-1990s) X X
33. MichaelBruck,programmanagerContentGroup X X
34. VinDham,programmanagerPentiumprocessor (early1990s) X X
35. RichardWirt,directorSoftware, IAL X X
36. Les Kohn,technicalmanager,i860 processor (late 1980s) X
37. Bruce McCormick,manager,Flash(mid-1980s) X
38. SallyFundakowski,manager,CMG(early1990s) X
39. Tom Macdonald,marketingdirectorfor 386 and 486 processors X
40. Jim Yasso, mgr.in Desktop Prd.Grp.and Microp.Prd.Grp.(mid-1990s) X
41. Don Whiteside, gen. mgr., DigitalImagingand Video Division X
43. LoriWigle, strat. mkting.dir.DigitalImagingand Video Division X
43. TomWillis,managerin CorporateBusiness DevelopmentGroup X
44. Dave Williams,directorHome MediaLab X
45. Dave Cobbley,directorHome Media Lab X
46. Rob Siegel, programmanagerHood River X X
47. Ganesh Moorthy,mgr.,Applianceand Comp. Div.(Deskt. Prod.Grp.) X
48. KrishBandura,engineer, Hood River X
49. Roy Coppinger,productmgr.OEMProd.and Syst. Div.,Hood River X
50. EricMentzer,marketingmanager,Chipsets X* X
51. AndyWilhelm,technicalmanager,Chipsets X*
52. Andy Beran,finance manager,Chipsets X*
53. Tom Bruegel,finance manager,Networking(mid-1990s) X*
54. Dan Sweeney, marketingprogrammgr.,Networking(mid-1990s) X*
55. Steve Cassell, engineeringmgr.,Networking(early1990s) X*
56. KirbyDyess, marketingmgr., PC Enhancemt.Org. (late 1980s) X* X
57. Susan Studd, humanres. mgr., PC Enhancemt.Org. (late 1980s) X* X
58. GerryGreve, marketingdirectorProShare(mid-1990s) X*
60. LauraFinney,finance managerProShare(mid-1990s) X*
61. TaymoorArshi,engineeringmanager,ProShare(mid-1990s) X*
62. MarkOlson, productmarketingmanagerMicroproc.Prod.Grp. X
63. John Sutherland,manager,Systems ManagementDivision X
* These interviewswere tape recordedby Intelconsultants,an transcriptsof the raw recordedinterviewdata were
made availableto this author.

329/ASQ, June 2002


Grovewas CEO(Coganand Burgelman,1991; Steere and
Burgelman,1993a, 1993b; Fine and Burgelman,1997; Bam-
ford and Burgelman,1997a, 1997b; Bamfordand Burgelman,
1998; Suzukiand Burgelman,1998; Burgelman,Carter,and
Bamford,1999). Lengthydiscussions with the research asso-
ciates involvedin writingthese cases providedme with an
opportunityat each writingto check whether they thought
my interpretationof the data was consistent with theirs, pro-
vidingan additionalcheck on internalvalidity(e.g., Dyck and
Starke, 1999). Grovetaught these cases in StanfordBusiness
School's Master'sof Business Administration(MBA)program
throughoutthe research period.This yielded richadditional
data as he reflected on Intel'sstrategic situationin class. It
provideda window into the mind of the CEOas strategic
thinkerthat has rarelybeen matched in previousstudies.
Multilevel Comparative Analyses
I adopted the methodology of groundedtheorizing(Glaser
and Strauss, 1967) to analyzethe field data. While grounded
theorizingrequirescare not to use data simplyas illustrations
of preconceived theoreticalideas, analysis is only possible
within a theoreticalperspective. With this in mind, I used
three interrelatedconceptual frameworksgenerated through
groundedtheorizingin earlierwork. Together,these frame-
works form an evolutionaryresearch lens to performa multi-
level comparativeanalysis of Intel'sstrategy makingduring
Andy Grove'stenure as CEO.At the company level, the
analysis is comparativewith respect to time. I examined
Intel'sstrategy makingduringEpoch IIwith a framework
includinginducedand autonomous strategy processes
(Burgelman,1991) and compared it with Epoch I. At the com-
pany-environmentinterface level, the analysis is also compar-
ative with respect to time. I examined the coevolutionof
Intel'sstrategy with the PC industryduringEpoch II,leading
to lock-in,with a frameworkof internaland externalforces
drivingcompany evolution (Burgelman,1994) and compared
it with Epoch I. The forces taken into account in this frame-
work includethe basis of competitive advantage in the indus-
try,the'firm'sdistinctivecompetencies, its officialcorporate
strategy, its strategic actions, and its internalselection envi-
ronment.At the intracompanylevel, the analysis compares
new business development efforts duringEpoch II.The
process model of internalcorporateventuring(Burgelman,
1983), which identifiesthe interlockingkey activitiesof multi-
ple levels of management involvedin internalnew business
development, helped in examiningthe behavioraldetails of
the development of four cases in the context of Intel's
strategy-makingprocess.
Strengths and Limitations of the Research
By concentratingon one firmand trackingone CEOthrough-
out his tenure, I had access to sources with intimateknowl-
edge of the details of the company's strategy making.It also
allowed me to become familiarwith "the manager'stemporal
and contextualframe of reference" (Vande Ven, 1992: 181).
Because I had virtuallyunlimitedresearch access to the com-
pany throughoutthe twelve-year research period, I was able
to obtain inputfrom differentlevels of management, which
330/ASQ, June 2002
Coevolutionary Lock-in

provideda basis for triangulationand made it possible to


maintainan appropriatelevel of distance and neutrality,while
capitalizingon the teaching collaborationwith Andy Grove.
Nevertheless, the research has several limitations.First,it
focused on a single high-techcompany run by one of the
foundingteam members. Also, duringGrove'stenure as
CEO,the PC industryexpanded enormously,and fortuitous
circumstances contributedto giving Intelthe opportunityto
become a drivingforce. Finally,duringthe study, I kept track
of the evolving fortunes of Intel'scompetitors, but it would
have been fruitfulto study these other organizationssystem-
aticallyif time and access had permittedit.
LOCK-INOF STRATEGYAND
COEVOLUTIONARY
ENVIRONMENT
Grove's Strategy Vector
DuringEpoch II,GordonMoore remainedas chairmanand
CraigBarrettserved as chief operatingofficer (COO).Looking
back in 1999, Andy Grove pointed out that "At no point in
Intel'shistoryhas it been a solo show. It's never been only
one person leadingthe organization.Ourtraditionis some-
what of a shared power structure."Nevertheless, many
insiders confirmedthat Andy Grove drove strategy making
duringEpoch II.Table2 provides a chronologyof selected
key instances throughoutEpoch II,when it was clear that
Grove made the difference in how Inteltook strategic action
in the core microprocessorbusiness.
The data presented in table 2 show that Grove's role in dri-
ving Intel'sstrategy makingrelied more on strategic recogni-
tion than on foresight. Intelhad been luckyto invent the
microprocessorand even more luckyto obtain the design
win for the IBMPC. But it was ex post facto strategic recog-
nitionof the importanceof these fortuitousevents that set
Intelon its highlysuccessful course. An article in the New
YorkTimesin 1988 pointed out that it was "irksometo com-
petitors . . . that there is a fair amount of luck involved in all
of this [Intel'ssuccess]." Respondingto this, Andy Grove
was quoted as saying, "Thereis such a thing as luck and
then you grab it and exploit it" (Pollack,1988). Grove some-
times also called it "earned luck"(Schlender,1989). Table2
indicates that the abilityto get the organizationto follow up
on the mandates that he imposed based on his strategic
recognitionwas another defining characteristicof Grove's
leadership.ContrastingGrove'sstrengths to those of co-
founder RobertNoyce and his own, GordonMoore said,
"Andyis a true manager.He is 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
stronglydata driven."
Focusing Intel on the microprocessor business. Table2
indicates that toward the end of Epoch I, then-COOGrove
recognized that Intel'sfuture lay in microprocessors rather
than memory products.To make sure that the organization
would be committed to the new microprocessor-focused
strategy when he became CEO,Grove made majorchanges
in Intel'ssenior management. He recalled:
331/ASQ, June 2002
Table2
Company Level of Analysis: Andy Grove's Impact on Intel's Strategy Making during Epoch II*
Selected key instances Strategic recognition Strategic action
Transitionto Epoch II:Focusing Intel on microprocessors as chief operating officer (mid-1980s)
Ed Gelbach (sales VP and director)t: COO Andy Grove: "I stayed quiet Groveremoves Carstenas GM Com-
"In board meetings the question of because I didn'tknow what to do, ini- ponents Divisionin summer 1985.
DRAMwould often come up. I would tially."
supportthem from a marketperspec- Grove moves Sunlin Chou and the
tive, and Gordon[Moore]would sup- COO Andy Grove: "It's not always DRAM Technology Development
port them because they were our clear why you do certainthings. You Groupto microprocessors.
technology driver.Andy [Grove]kept do a lot of things instinctively,without
quiet on the subject." knowingwhy you're doing it. I knew Grove goes to Oregon in October
we had to get out of DRAMsand put 1985 and tells the organization:"Wel-
Jack Carsten (GMComponents Divi- all our brighteston microprocessors. come to the mainstreamof Intel."
sion)t: "Grove said: 'Don't worry
about the memory business, it is not COOAndy Grove:"I recall going to
importantto our future."' see Gordon(Moore)and asking him
what a new managementwould do if
Anothersenior executivet: we were replaced. The answer was
"Grove has been preaching:'Make clear: Get out of DRAMs.So, I sug-
the tough decisions! Don'tdo tomor- gested to Gordonthat we go through
row something because you did it the revolvingdoor,come back in, and
today."' just do it ourselves."
Resolving the battle between i860 (RISC)and x86 (CISC)microprocessors within Intel (1991)
Dennis Carter(VPCorporateMarket- Andy Grove in February1991: "The Grovedid not allowthe plannedintro-
ing): "In the end, Andy [Grovel strategy process reflects the compa- ductionof both486c and486r proces-
resolved the debate. He essentially ny's culture. You can look at it posi- sors that would have signaled a
did a compromise that favored tively or negatively.Positively,it looks plannedtransitionpath from CISCto
CISC." like a Darwinianprocess: we let the RISC.The i860 business was to con-
best ideas win; we adaptby ruthless- tinue by that name andwas soon halt-
ly exitingbusiness; we provideauton- ed in early 1991.
omy, and top management is the ref-
eree who waits to see who wins and Andy Grove in November 1992: "It
then rearticulatesthe strategy; we was a confusing periodfor Intel....
match evolving skills with evolving The i860 was a very successful rene-
opportunities.Negatively,it looks like gade product that could have
we have no strategy; we have no destroyed the virtuouscircle enjoyed
staying power, we are reactive, try by the IntelArchitecture.... Intelwas
and move somewhere else if we fail; helping RISC by legitimizing
we lackfocus." it. ..."
Identifyingthe magnitude of capital investment as Intel's new differentiator(1993)
DirectobservationduringSLRP1993: Andy Grove: Pointing to the great Grovewas willingto makethese large
In his kick-off presentation, Grove uncertainty associated with these bets. Duringthe remaining4 years of
identified Intel's successive key capital investments, Grove posited his tenure as CEO, Intel invested
strategic differentiators throughout that they would provide Intel with a $13.5 billion in plant, property,and
its evolution:Silicontechnology com- new competitive advantage. He equipment.
petence (1970s), design competence asked, rhetorically,"Who is going to
(mid-1980s), intellectual property invest $5 billionon speculation?" In 1997, CraigBarrettsaid, "It'sa risk
(late 1980s), and brand preference to go out and spend billionsof dollars
(early1990s). He then suggested that on these manufacturingplants. But if
the increasinglylarge capital invest- we didn't,we couldn'tpossibly reap
ments necessary for next-generation the benefits. We're going down the
processors had become the new dif- road at 150 miles per hour,and we
ferentiatorfor the next several years. know there's a brickwall someplace,
but the worst thingwe can do is stop
too soon and let someone else pass
us" (Reinhardt,Sager, and Burrows,
1997: 71).

Continued

332/ASQ, June 2002


Coevolutionary Lock-in

Table2 (Continued)

Resolving conflict around "IntelInside" between Corporate Marketingand Intel Products Group
DirectobservationduringSLRP1993: Andy Grove: "This is a lame state- Grove decides: "Dennis [Carter]and
At the end of SLRPthe objectives as ment. And yet it is the inflection Frank[Gill]must rephrasethis. It must
stated in 1992 were revisitedin light point-[similar] to what happened be words that will affect hundredsof
of the discussion during the 1993 with the transitionfrom memories to people that work for them and are
SLRP.The thirdobjectivein 1992 was microprocessors. This involves a fightingover it. The new words [must
"Manage the Intel and Intel Inside dialectic. It is a move from a single make sure] we get creditfor what we
brandsfor significantreturnand long- space to a dual one. This dualityis all do for our lend-user] customers:
term advantage."Grovefelt that this over the place. It is a continuationof ease-of-use, richness, upgradability;
objectivehadto be restated in lightof the change from OEM to a distribu- and who our customers could be."
the intense conflicts that had broken tion channel."
into the open between CM (Dennis
Carter)and IPG(FrankGill)duringthe
SLRP1993 discussions.
Supporting Intel's motherboard business in the face of organizational resistance (mid-1990s)
HaroldHughes (formerCFO):"Andy Andy Grove:"Ihave been rabidabout Grovesupportedthe development of
was always brilliantat identifying fourthings in my careerat Intel:moth- the motherboardbusiness in spite of
threatsto our business. Forexample, erboards, Intel Inside, chipsets, and strong oppositionof the microproces-
on the motherboardsbusiness, Andy videoconferencing." sor division,whose OEM customers
and I clashed. I said that we were complained vigorously about Intel's
never going to make any money on verticalintegrationstrategy,and in the
motherboards. But they did push face of reservationson the partof the
adoptionof our microprocessors.Our CFO.
motherboardbusiness allowedthe lit-
tle [OEMs]to stay competitive."
Supporting the chipset business to drive industry adoption of Intel technology (mid-1990s)
Several executives pointed out that After the chipset business became Grovethen began to view the chipset
Andy Grove initiallydid not support very successful, AndyGrovechanged business as an importanttool for sup-
the development of the chipset busi- his mindabout chipsets as a strategic portingthe corporate strategy. Andy
ness based on the new Peripheral business for Intel. Bryant(CFO)said, "At a time when
Component Interconnect (PCI) bus motherboardpricing was extremely
technology but, rather, wanted to competitive,the motherboarddivision
introducethe new technology as an decided not to use Intel'schipsets be-
enabling technology into the PC cause they were morecostlythanthird-
industry with a consortium-based party alternatives-even though they
effort. providedsuperior performance..
Groveruledthatthe long-terminterests
of the company required moving
advancedtechnologyinto the market-
place and that we should forgo short-
termreturnsforthe long-termbenefits."
Driving Intel to meet the threat of the growth of the low-end of the PC market segment (1997)
Direct observation during SLRP in Andy Groveduringhis SLRPkickoff: Grove articulated a new mandate,
September 1997: Grove was very "We say we have a top-to-bottom requiringthe assignment of a large
concerned about recent develop- strategy. But we don't act top-to-bot- number of engineers to the task of
ments in the PC marketsegment. He tom, because Intel has low-end pho- developing a microprocessorspecifi-
felt that Intel'stop managementwas bia..... But the low end is not going callyfor the low-end marketsegment
failing to see the strategic implica- away. . . . The data about desktop
tions of the rapidgrowth in demand sales at the retail,reseller,and direct Inaboutsix monthsthe team developed
for below-$1,000 PCs. level all show a downward trend in a new productcalledthe Celeronproces-
price:$500 in about a year! I have not sor, which made it possiblefor Intelto
seen that before, And the volumes at regain marketsegment share against
the low end are up. So, the good AMDinthe low end by early1999.
news about segment zero is that we
have it on our road map. The bad In early 1999, Paul Otelliniobserved,
news is that we don't have an engi- "We've made a lot of progress on the
neered product." low end. One year ago in the sub-
$1,000 market our share was about
38 percent. We then lost some
ground, but we have regained share,
so we're at about 38 percent again."
Continued

333/ASQ, June 2002


Table2 (Continued)

Looking back: Grove's influence on the PC industry during Epoch II


GordonMoore (chairmanemeritus)in
1999: "When he became CEO, he
reallyjumped on the opportunityto
organize the industry. I wasn't so
inclined to do this. He likes public
exposure more than I did, and he has
a strongerfeeling aboutwhere he fits
in. ... Andy has had a tremendous
impacton what's going on outside."
* Abbreviationsand terms used in this table are as follows: DRAM= dynamicrandomaccess memory;RISC= reduced
instructionset computing;CISC= complex instructionset computing;and SLRP= strategic long-rangeplanning.The
motherboardis the main integratedcircuitboardin a PC;it containsthe microprocessor,the memory,and other sup-
port chips. A chipset is the set of supportchips for the microprocessor,for example, a chip that controls computer
graphics.Bus refers to the set of electricalconnections between a microprocessorand the other chips on an integrat-
ed circuitboard.The speed of communicationallowed by the bus affects PC performance.The PCIbus architecture
increased speed significantlyover the previousbus standard.
t Executiveinterviewedfor Epoch I Study,not listed in table 1.

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 man-
agers 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.
Intel's new corporate strategy reflected key lessons that top
management had learned from the DRAM (dynamic random
access memory) exit. In the context of a case discussion in
an MBA class in the early 1990s, Grove said:

We learned that we had to get around the companies that had sub-
jugated us in DRAM. We learned that high market share was critical
for success and that to get market share we had to be willing to
invest in manufacturing capacity. Such investments involve big bets
because they have to be made in advance of actual demand. We
learned that commodity businesses are unattractive, so we didn't
want to license out our intellectual property anymore.

General-purpose microprocessors were a disruptive technolo-


gy (Christensen and Bower, 1996). Microprocessor develop-
ment was subject to Moore's Law, which posits that comput-
ing power doubles every 18 months and is available at the
same price. Andy Grove was among the first to recognize
that, in contrast to the vertically integrated mainframe and
minicomputer industries, the PC industry followed a "horizon-
tal" model in which a component manufacturer's products
needed to be able to work with other component manufac-
turers' products (Grove, 1993, 1996). Grove's "vertical" and
"horizontal" were a precursor to what economists call
"closed" and "open" models of industry organization (Farrell,
Monroe, and Saloner, 1998: 144). Success in the horizontal
PC industry was governed by increasing returns to adoption,
334/ASQ, June 2002
Coevolutionary Lock-in

a new economic force (e.g., Arthur,1987) that was initially


not well understood by most industryparticipants.Increasing
returnsto adoption meant that a technological platform,like
Intel'sx86 microprocessors,became increasinglyvaluablethe
more people were using it. Achievinga high installedbase
was key to creatinga virtuouscircle. While economies of
scale and economies of learningwere importantdetermi-
nants of the relativesuccess of different industryparticipants
competing withinthe same microprocessorarchitecture,
increasingreturnsto adoptionstrongly affected competition
between differentarchitectures.
Resolving the internal battle between CISCand RISC.The
x86 architecturewas based on complex instructionset com-
puting(CISC).Duringthe mid-1980s, however, Intel's
autonomousstrategy process generated the development of a
microprocessor(the i860) based on a new architecturecalled
reduced instructionset computing(RISC).Internalchampions
of the i860 had been able to generate supportfrom worksta-
tion originalequipmentmanufacturers(OEMs),which were
new customers for Intel.During1989-90, the autonomous and
somewhat surreptitiousdevelopment of the i860 and its initial
marketsuccess looked likea potentiallyadaptivevariation
(Burgelman,1991). But the new microprocessorsoon created
significantconfusion insidethe companythat reflected exter-
nal confusionabout the importanceof the RISCarchitecture
for the futuredevelopmentof the PC.The internalconfusion
manifested itself in the emergence of two warringcamps
within Intel'smicroprocessordevelopment group (MPG).Each
camp had its externalsupporters.AndyGrovesaid that
Microsoftsupportedthe i860. Compaq,however, stronglysup-
portedthe x86 architecture.Accordingto Grove,withina short
periodof time, the RISCcamp had been able to claimabout 50
percent of the microprocessordevelopment resources
because there was no clear corporatestrategy regardingRISC
(personalcommunication).Some within Intelproposedto cre-
ate a transitionpathfrom the x86 architectureto the RISC
architectureby bringingout two versions of the i486, one
called i486c and the other i486r,but this proposalraninto
strong resistance from Dennis Carter,Intel'ssenior marketing
executive duringmost of Epoch II,who feared that it would
undermineIntel'sbrandidentity.Inpartmotivatedby the nega-
tive consequences that a similarbattle between CISCand
RISCwas havingwithinrivalMotorola(Tredennick,1991),
Groveeventuallyresolved the situation.Table2 quotes Dennis
Carteron how Grovedecided the issue. It also reportsGrove's
growingconcerns about Intel'sstrategy-makingprocess. The
episode strengthened his determinationto fullyexploit Intel's
favorablestrategicpositionwith the x86 architecture.He said,
"Thecommitmentto the x86 architecturevectorizedevery-
body at Intelin the same direction."
Effectively driving strategy making in the core business.
The significanceof the rise and fall of the i860 microproces-
sor lies primarilyin the effect it had on Grove'sefforts to fur-
ther strengthen Intel'sinduced strategy process. Table2
shows that Grove had come to the conclusion that Intel's
Darwinianstrategy process was perhaps a guise for lack of a
clear strategy. His efforts to vectorize everybody at Intel in
335/ASQ, June 2002
the same directionin 1991 created an inducedstrategy
process superblysuited for exploitingthe richopportunities
in the PC marketsegment of the microprocessorindustry.
Several entries in table 2 describe how Grovedrove Intel's
strategy makingin the core business duringthe remainderof
Epoch II.He showed keen insight in the successive strategic
differentiatorsthat had formed the basis of Intel'scompeti-
tive advantage in the past and emphasized the importanceof
large capitalinvestments for competitive advantagefor the
remainderof the 1990s. He forced senior executives to
resolve the frictionsthat were emerging between corporate
marketing'sconcerns about protectingthe Intelbrandand the
needs of businesses outside the core microprocessorbusi-
ness. He forced the motherboardbusiness to adopt Intel's
more advanced but also more expensive PCIchipset technol-
ogy in the face of resistance of both the motherboardman-
agers and the finance organization.Towardthe end of Epoch
II,Groveforced the microprocessorbusiness to face up to
the dangerous threat posed by the rapidlygrowing low end
of the PC market.He recognizedthat Intel's"low-endpho-
bia" was preventingit from meeting the challenge posed by
this majorenvironmentalshift and directed Intelto engage in
a crash effort to develop the Celeronprocessor to meet it.
Finally,as GordonMoore observed, Grove'sstrategy vector
gave Intelthe opportunityto drive its externalenvironment,
that is, the development of the PC marketsegment.
Intel's Narrow Business Strategy
Alreadyin 1989, then-chairmanGordonMoore had observed
that CEOAndy Grove had significantlynarrowedIntel's
strategic focus, but he also predictedthat the growth poten-
tial of the microprocessorbusiness would not make that a
problemin the next twelve years (Burgelman,1991). Looking
back in 1998 and comparingIntel'sstrategy duringEpoch II
and Epoch I, Grovesaid, "Themost significantthing was the
transformationof the company from a broadlypositioned,
across-the-boardsemiconductorsupplierthat did OKto a
highlyfocused, highlytuned producerof microprocessors,
which did better than OK"(Kawamotoand Galante,1998).
Manysenior executives confirmedthat Groveforced a dis-
tinct shift in the strategy-makingprocess toward a narrow
business strategy focused on microprocessorsfor the PC
marketsegment. Table3 providesevidence of this shift. The
views expressed in table 3 touch on variousaspects of the
strategic leadershipapproachGroveused to focus Intel's
induced strategy process narrowlyon the microprocessor
business. They includesetting clear objectives and establish-
ing a structuralcontext (Bower, 1970; Burgelman,1983),
includingstrategic planning,organizationstructure,and
resource allocation,to align strategy and action.
Unambiguous strategic objectives. Intel'sstrategic focus
became ingrainedin the strategy-makingprocess throughthe
setting of clear and consistent objectives. Intel'snumber-one
objective was to strengthen the position of Intelmicroproces-
sors in the evolving computer industry.A relatedobjective
was to "make the PC it," which became somewhat of a rally-
ing cry. Groveviewed the PC as the idealtool for computing
as well as for communications,and even for entertainment.
336/ASQ, June 2002
Coevolutionary Lock-in

Table3
Company Level of Analysis: Views on Intel's Narrow Business Strategy
during Epoch II
GordonMoore (1989):
"Overtime ... Intelhas narrowedand narrowedits technologicalinterests.
Andy[Grovelhas been instrumentalin this.... We can do variationson
present businesses very well. But doing something new is more difficult."
GerryParker(1989):
"We could now manufactureeverythingin one and one-halfplants.That's
obscene. Youneed a broadproductbase-EPROM [electricallyprogramma-
ble read-onlymemory]is a natural...."
Les Vadasz(1988):
"Thesystem [strategiclong-rangeplanning]is now [inthe late 1980s] more
top-down.A high-levelgroupsets the corporatestrategy,and business
units operate withinthat focus. Business units must focus on a few things
and do them right ... Some managerscomplainthat their 'sandbox'is too
well defined."
A senior executive (1995):
"Intelmay be too focused too soon. We have narrowedour range of experi-
mentationtoo fast from 360 degrees to 180 and then to 90. The code
words are:Youdon't have a business plan;your strategy is vague."
"We must narrowdown from a 360 degree scan to 20, but even so we still
have 20 things to do. Andy[Grove],however, wants a 'laser shot.'"
FrankGill(1997):
"In1994-95, Andy[Grove]would tell me 'Frank,I make a billiondollarsin
profitper quarterand you make a billiondollarsin revenue per year.This is
all distraction,so focus on Job 1.'"
Anothersenior executive (1998):
"... a lot... is drivenfrom Job 1, because every six months we have a
SLRP[strategiclong-rangeplanningmeeting].Andy[Grove]stands up and
says ... here is a problem.And everyone says ... we can go do wonderful
things to solve that problem."
CraigBarrett(1999):
"[Duringthe second epoch] we became much more verticalizedbehind IA
and relatedbusinesses. Now we are more broad.... This requiresless top
down managementand more P&Land line management."
A thirdsenior executive (1999):
"Barrettis very differentfrom Grove.First,he's encouragingnew ideas....
Andy wouldn't have let that happen. Craig made it happen .... Second is
behavior.Ifyou have a good idea, overwhelm it with resources:What do
you need? Do what it takes. Come backwith a prize.... That'sa different
style."
A fourthsenior executive (1999):
"ButI am more concernedabout Andy[Grovelbecause of his singularfocus.
Andysays that PCs are becoming a commodity.So, we must focus on
servers and not let Sun [Microsystems]capturethis. It is like going backto
the old days."
"Barrettat some pointwill be expected to set the corporatestrategy;and if
he doesn't, Andy [Grove]will."

Intel also made a distinction between "Job 1" and "Job 2."
Job 1 encompassed everything that had to do with making
the Intel architecture more successful. Job 2 involved the
development of new businesses around the core business.
CEO-driven strategic planning. Grove said that he had used
changes in the company's strategic long-range planning
process (SLRP) to redefine the content of the new corporate
strategy and get the organization to execute it:
In 1987, we blew up the SLRP process. Formerly it had been a very
bottom-up process, but there was no strategic framework. Each of
the different groups was supposed to come up with the strategy for

337/ASQ, June 2002


their group, and then we would try to piece them together like a jig-
saw 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 technical assistant at the time,
Dennis Carter... I became very directive in prescribing the strate-
gic 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.

Typically, Grove's SLRP kickoff speech was followed by a


two-hour presentation in which he addressed Intel's strategic
challenges, presented his vision of what was happening in
the industry, and identified high-level trends. The remainder
of the three-day meeting involved presentations by Intel's
senior executives concerning specific issues and topics. They
worked across product and functional groups to put their pre-
sentation together, with the help of a staff member. These
executives had been given their assignment without knowing
in advance what Grove was going to present. Dennis Carter
pointed out that this was viewed as a tough assignment,
dreaded by some, and that instances of strategic dissonance
surfaced immediately.
Centralized organization structure. During Grove's tenure
as CEO, Intel's organization structure became highly central-
ized. In the words of one senior executive, "Intel was orga-
nized around funneling things up to Gordon, Andy, and
Craig." Intel was structured as a matrix, with various corpo-

Table4
Company-environment-interface Level of Analysis: Highlights of Coevolution of Intel's Narrow Business Strategy

PC market segment

'81-84 '85 '86 '87 '88 '89 '90 '91 '92


IBMintroducesPC/XT/AT Largeinstalledbase for Compaqfirstwith 386 PC; Clone PC manufacturers
with an Intelchip and IBMPC/XT/AT; Compaq IBMfollows. are gainingshare.
Microsoftoperatingsys- has emerged as a viable
tem. competitor;Inteland
Microsoftare the fortu-
itous beneficiariesof a
"virtuouscircle."
Intel strategy
8088 and 80286 chips for Decision to be sole source Courtbattlewith AMD Intelis sole source for
IBMPCs;cross-license for new 80386 processor about intellectualproperty i486 processor; "Intel
other chip manufacturers; and to maintainproduct rightsfor x86 microproces- Inside"end-usermarket-
Intelinitiallynot fully leadership;"RedX" end- sors preventsAMDfrom ing campaign;Inteldevel-
aware of importanceof user marketingcampaign. enteringthe 386 market ops an ecosystem; i860
PC for its future. for 4 years; same for 486. battle resolved;Intelcre-
ates IALto enable the PC
industry.
Revenue
($Billion) 1.4 1.3 1.9 2.9 3.1 3.9 4.8 5.8
Profits($B) 0.0 -0.2 0.2 0.5 0.4 0.7 0.8 1.1
Cap. Invest. ($B) 0.24 0.15 0.3 0.5 0.4 0.7 0.9 1.2
R&Dexp. ($B) 0.19 0.23 0.3 0.3 0.4 0.5 0.6 0.8

338/ASQ, June 2002


Coevolutionary Lock-in

rate functions on one side and various product groups on the


other. Each product group carried profit and loss responsibili-
ty for its respective market, but no product group controlled
all of the functional resources needed to execute its strategy.
The functional groups were responsible for supporting the
product groups and for cultivating necessary expertise across
the organization. The functional groups were highly stable so
as to develop capabilities, while the product groups were
constantly redefined in order to match the evolving product-
market environment. Given the importance of microproces-
sors in Intel's new corporate strategy, and the relentless pace
with which new product generations needed to be devel-
oped, manufactured, and marketed, coordination among all
the groups was critical.
Tightly managed resource allocation. The resource alloca-
tion process strongly favored Intel's core microprocessor
business. As one executive observed in 1999:

Virtuallyevery single quarter,the requests outweigh the willingness


to spend. We would end up ZBB-ing[zero-basedbudgeting]the
lower ROIprojects.The largerROIprojectswere almost always
relatedto the mainstreamCPU [microprocessor]business. There-
fore, if you were not partof the mainstreambusiness, you needed
to be very spiritedand very perseverantto driveyour projects
throughthat process every quarter.I knew they were great busi-
nesses by any other metric,just not comparedto the microproces-
sor business. ... Ifyou were in a non-corebusiness, it was tough.

Complementary Strategic Thrusts


Comparing Epoch IIwith Epoch I, Craig Barrett said in 1999,
"We became the industry driving force." Table 4 identifies

and PC MarketSegment, Epoch II

'93 '94 '95 '96 '97 '98


Commoditizationof PCs: Internetemerges: threat Growthin demand for AMD[AdvancedMicro
intense marginpressure of the networkcomputer below-$1,000 PC is a real Devices] gains marketseg-
for PC OEMs;threatfrom (NC). threatto Intel;NCthreat ment share on the low
IBM-Apple-Motorola RISC does not materialize. end.
alliancedoes not material-
ize.

Intelis sole source for IntelintroducesPentium IntelintroducesPentium IntelintroducesCeleron


new Pentiumprocessor; Profor workstationswith with MMXand laterin the processor to combat AMD
Intelverticallyintegrates Windows NT;AMDlitiga- year PentiumII. on the low end; PentiumII
into motherboardsand tion for 386 and 486 set- Xeon for workstations/
chipsets, which are deci- tied; Microsoftpressures servers.
sively helpfulin Pentium Intelto stop its nativesig-
launch;Pentiumflaw crisis nal processing (NSP)pro-
and resolution. ject.

8.8 11.5 16.2 20.8 25.1 26.3


2.3 2.3 3.6 5.2 6.9 6.1
1.9 2.4 3.6 3.0 4.5 4.0
1.0 1.1 1.3 1.8 2.3 2.7

339/ASQ, June 2002


key dynamics of the PC marketsegment between the early
1980s and 1998. It also identifies several complementary
strategic thrusts, brieflydiscussed below, that made it possi-
ble for Intelto drivethe PC marketsegment. These comple-
mentarythrusts did not reflect a comprehensive ex ante for-
mulated strategic planto take controlof the PC market
segment. Rather,Grove'ssuccessful narrowbusiness strate-
gy set in motion a positive feedback process that extended
the numberand magnitudeof strategic responsibilitiesthat
Intelneeded to take on to sustain its positionas driverof the
PC marketsegment. These, in turn, reinforcedthe induced
strategy process.
Sole-source supplier. The installedbase of x86 microproces-
sors created by IBM'ssuccess in the PC marketsegment
(with Intel's8088 and 80286 microprocessors)had signifi-
cantly and fortuitouslyshifted bargainingpower in Intel's
favor.Understandingthe implicationsof increasingreturnsto
adoptionoffered Intelthe opportunityto become sole-source
supplierof microprocessorsfor the PC marketsegment as of
the 80386 microprocessorgeneration. Nevertheless, this was
a bold move given IBM'sstill very powerfulposition in the
industry.Lookingback, Grovesaid, "Whatgood is the 386 if
IBM doesn't adopt it? ... We were chewing our nails until
1986, when Compaqadopted the 386. IBMadopted it the
next year." Intelwas able to keep rivalAMDtied up in the
courts over intellectualpropertyrightsdisputes, which
allowed it to remainthe sole source for the 386 processor for
four years. The 386 microprocessorwas succeeded by the
i486, which was introducedin April1989. It again took four
years (untilthe summer of 1993) before AMDwas able to
launchits first 486-compatibleprocessors.
Investing in manufacturing. One of the imperativesassoci-
ated with the sole-source strategy was that Intelneeded to
become a world-classmanufacturer.Table4 shows the large
and rapidlyincreasingcapitalinvestments Intelmade during
Epoch II.Intel'snew manufacturingprowess depended on a
new distinctivecompetence: close integrationof the Micro-
processor Group'schip designs and process technology and
manufacturingcompetencies withinthe Technologyand Man-
ufacturingGroup.Intelbecame renowned for its abilityto
optimize the manufacturingprocess of a new chip design and
then to rollout that process to Intel'sother plants using the
"copy exact" principle.
Pacing the race through product leadership. Table4 shows
the rapidpace of productintroductionsbetween 1993 and
early 1998: Pentium(1993), Pentium Pro (1995), Pentium
MMX(1997), Pentium 11(1997), and the Celeron(1998)
processors. This time-drivenproductintroductionstrategy,
however, reflected deep intuitionfor the feasible pace of
development of the PC industry.Inan MBAclass in fall 1994,
Andy Grove revealed that he had learnedfrom studyingthe
data that the peak-to-peakproductionacross microprocessor
generations for 386 and 486 microprocessorshad been about
three years and would be the same for the Pentiumproces-
sor. Based on this, Groveassumed that the next generation
microprocessor,the P6, would follow the same adoption
340/ASQ, June 2002
Coevolutionary Lock-in

cycle, which informedthe timing of Intel'snext majorcapital


investment decisions.
Building brand with end users. InApril1990, Intel launched
its first "IntelInside"campaign.Aimed directlyat end users,
ratherthan Intel'straditionalPC OEMcustomers, the cam-
paign sought to influence customers to ask for Intelmicro-
processors specificallywhen they purchaseda PC. Major
OEMs such as Compaqand IBMinitiallyrefused to partici-
pate in some elements because they felt that Intel Inside
decreased their abilityto differentiatetheir products from the
competition, but eventuallyall of them carriedthe Intel Inside
logo on their products, in part,because Intelengaged in mas-
sive co-marketingcampaigns with the OEMs. From 1990 to
1993, Intelinvested more than $500 millionin end-user mar-
keting campaigns. Paradoxically, the Pentiumflaw crisis of
November-December1994, which accordingto Grove,
"shook Intelto its core," in some ways indicatedthe power-
ful impact of Intel'sbrandingstrategy on end users.
Introducing industry-enabling technologies. Increased
competitionamong a growing numberof PC OEMs created
intense pressure on their profitmargins. Combinedwith its
successful sole-source strategy, this gave Intelthe abilityto
appropriatea large partof the availableprofits in the PC mar-
ket segment. This created a positive feedback loop, which
increasinglyshifted the center of industryinfluence from the
PC OEMs to Intel(andto Microsoft)duringthe 1990s. Only
the largest PC OEMcustomers could affordto do much
research and development (R&D).Other OEMs became
increasinglydependent on Intelfor technological innovation.
Intelcreated the IntelArchitectureLabs (IAL)for the purpose
of developing new technologies that would remove techno-
logicalbottlenecks preventingPCs from takingfull advantage
of the increased processing power of new-generation micro-
processors. These technologies were offered to the OEM
customers for free or for nominalroyaltypayments.
Cultivating an ecosystem of complementors. The most
importantcomplementaryproductfor Intel'smicroprocessors
was Microsoft'sWindows operatingsystem software. Andy
Grove described the relationshipbetween Microsoftand Intel
as "two companies joined at the hip."While constantly vying
for perceived leadershipof the PC industryand jealously
guardingtheir own spheres of influence (software for
Microsoftand hardwarefor Intel),most of the time the two
companies were able to maintaintheir symbiotic relationship
throughoutGrove'stenure as CEO. Intelalso invested in cre-
ating internalsupportgroups to help other independent soft-
ware vendors develop applicationsrequiringhigh processor
power to stimulate demand for its next generation proces-
sors. Intelprovidedits partnerswith advance information
about its next microprocessordesigns and support products.
Forward integration into chipsets and motherboards.
Intel'schipsets and motherboardsmade it possible to lever-
age its strong strategic position in microprocessors by
enabling OEMcustomers, who did not have the resources to
develop these system-level products,to introducePCs with
Intel'slatest microprocessors.This is turnwas helpfulin
341/ASQ, June 2002
reducingits dependency on the stronger OEMs, in case the
latterwere reluctantto stay with Intel'sroadmap for devel-
oping next-generationmicroprocessors.This actuallyhap-
pened when some majorOEMs initiallydecided to wait to
introducePentium-processor-basedPCs, and Intelenabled
PackardBell and Dell to take the lead.
Successful Coevolution Turns into Inertia
DuringEpoch II,in contrast to Epoch I, Intel'sdistinctivecom-
petencies continuedto evolve with the basis of competition
in the PC marketsegment of the microprocessorindustry,
and the officialstrategy clearlydrove strategic action, lever-
aging both positionand distinctivecompetence. This gave
the company great momentum between 1987 and 1997,
which is reflected in revenue growth and profitgrowth (table
4). In late 1998, Intel'sstock marketvaluationsurpassed
$200 billionfor the first time.
Lock-in. Intel'snarrowbusiness strategy tied its success
increasinglyto that of the PC marketsegment. By 1993, 486
microprocessorsaccounted for 75 percent of the company's
revenues of $8.8 billionand 85 percent of its $2.3 billionin
net profit.By 1998, 80 percent of Intel's$26.3 billionin rev-
enues and just about all of its $6.1 billionin net profitscame
from microprocessors.Signalingthe company'sextreme
dependence on the prospects of its product-marketenviron-
ment, revenues grew only 5 percent, and net income
declined 13 percent during1998, in partas a result of the
unexpectedly rapidrelativegrowth of the low end of the PC
marketsegment. Table4 shows the increasinglylarge capital
and R&Dinvestments that needed to be made to keep dri-
ving the coevolutionaryprocess. Also, Intel'sdependence on
the OEMcustomers as a distributionchannel for its micro-
processor products made forwardintegrationinto systems
productsdifficult.Intel'sstrong interdependencewith
Microsoftimpeded strategic initiativesin the software area.
In one widely noted case-Intel's Native Signal Processing
(NSP)initiativeto augment the microprocessor'svideo capa-
bility(table4)-Grove admittedthat Intel "caved" in the face
of Microsoft'sdispleasure (Schlender,1996).
Inertia. By 1997, Intel'sroad map for the development of
next generations of microprocessorsdetermined its long-
term development trajectory,which was not easily changed.
While Intelhad put mechanisms in place that allowed very
fast response to short-termcontingencies affecting the road
map, Dennis Carterexplainedthat the abilityto make quick
adjustments, paradoxically,reinforcedthe company'sstrate-
gic focus and the lock-inwith the PC marketsegment. The
successful crash effort to develop the Celeronprocessor,
however, signaled that while Intel'slock-inwith the PC mar-
ket segment remainedstrong, the lock-inof the PC market
segment with Intelwas perhaps loosening. Also, toward the
end of 1996, Andy Grovewas beginningto worryabout the
effect Intel'sstrong influencewith its OEMcustomers was
havingon its strategy-makingprocess. Inan MBAclass dis-
cussion in fall 1996, Grovesaid, "Thereis a hiddendanger of
Intelbecoming very good at this. It is that we become good
at one thing only."
342/ASQ, June 2002
Coevolutionary Lock-in

COEVOLUTIONARY LOCK-INAND STRATEGIC INERTIA


Reduced Capacity for New Business Development
By 1997, then COO CraigBarrettdid not believe that Intel
could sustain its historicalgrowth rate and profitabilitysolely
with microprocessors. Barrettrealizedthat Intel'sintense
focus on microprocessors had made it difficultfor new ven-
tures to thrive inside Intel(hence, his use of the creosote
bush metaphormentioned earlier).Differentgroups in the
company continued to explore a multitudeof new business
ideas (Burgelman,Carter,and Bamford,1999), but Intel's
autonomous strategy process had become less able to
exploit new business opportunities.Dennis Carternoted that
outbound marketing(deliveringa technology to the market)
dominated inboundmarketing(findingnew market needs
that could be met by technology). FrankGill,an executive
vice president in charge of Intel'snew business development
duringmost of Epoch II,pointed out that Intel'smatrixorgani-
zation did not providemanagers with much opportunityto
learnto make trade-offsamong variousfunctionalconsidera-
tions. This impeded the development of new generations of
general managers able to develop new businesses. Also,
business-level general managers must resolve the initial
ambiguityabout the correct strategy of a new business, but
in the corporatecontext this is not sufficient. To continue to
obtaincorporatesupport,the process of strategic context
determinationmust be activated,which helps linkthe new
business strategy to the corporatestrategy. This explorative,
iterativeprocess involves multiplelevels of management in
buildinga new strategic thrust for the corporation(Burgel-
man, 1983).
Duringfall 1999, Andy Grove reflected on the slowing down
of growth in the core microprocessorbusiness and his
efforts to develop new businesses duringEpoch II:"Theold
CEOknew that this was coming. He tried like hell to develop
new business opportunities,but they almost all turned into
[dirt]."Publicdata support Grove'scontention that he knew
relativelyearlyon that Intelwould have to transformitself
again. Alreadyin 1993, he had said:
Ourpeople have navigatedsuccessfully throughone transformation,
so perhaps it won't be as hardto sign them up for anotherone. But
success can trapyou. The more successful we are as a micro-
processor company,the more difficultit will be to become some-
thing else. To take advantageof some opportunitiesI see ahead,
we're going to have to transformourselves again. The time to do it
is while our business is still strong. (Grove,1993: 60)
While Grove recognized the need for strategic renewal, diffi-
culties in developing new businesses duringEpoch IIsuggest
that he and Intelwere subject to sources of strategic inertia
associated with coevolutionarylock-in.Table5 identifies
these two sources of strategic inertia.The ProSharecase
shows that the CEO'sactive involvement in drivingnew busi-
ness development is likelyto impose the logic of the suc-
cessful core business in an area in which it may not apply,
thereby impedingdevelopment of an appropriatebusiness
strategy and simultaneouslyinducingescalation of commit-
ment. The Hood Rivercase shows that even if the CEOis
343/ASQ, June 2002
not actively involved,he or she may cast a shadow of influ-
ence that also impedes the development of an appropriate
strategy for the new business, even though not inducing
escalation of commitment. The chipset case shows some of
the majordifficultiesa new business must overcome to get
corporatesupport if the CEOinitiallyviews it as an enabler

Table5
Company Level of Analysis: Coevolutionary Lock-inand Sources of Strategic Inertiaduring Epoch II
Views from below Intel's strategic intent Strategic action
If strategic, apply logic of core business strategy
Intel's strategy for videoconferencing (ProShare)
PatrickGelsinger(GMProShare): Grove'sintentwas to make video- Groveassigned PatrickGelsinger,in
"ProSharewas viewed as a horizon- conferencingan integralcapabilityof charge at the time of the next-gener-
tal capability-that was Andy's the PC.Tothis end, he favoreda ation microprocessordevelopment,
[Grove]wish." frontalassault on the entire PC mar- to ProShare.Grove:"MovingPat off
ket segment, ratherthan targeting of P6, a producton which the future
"We could have acted on the vertical verticalsegments first. of our companytrulydepends, to run
marketssix months sooner if Andy this new initiativewas a very contro-
had not had such a strong opinion." versialstep. But in manyway this is
the test of it."
AnotherProShareexecutive: "There
wasn't a debate about it, there wasn't Grovecontinuedto be deeply
even a discussion.... Andy had involvedin the strategicdecision
alreadytrainedthe organization, makinguntil1996, when he asked
meaning Intel,that periodicallyhe FrankGillto scale down the effort,
gets all these flashes of an idea." which involvedsome 700 people at
the time.
FrankGill(seniorexecutive): "Itwas
not being out of the loop so much as Grovein 1999: "Weassumed that
not being sure.... [I thought] just because it could be done techni-
maybe the throwingof massive callythere would be high demand. I
resource at it would work. I didn't was an enthusiasticuser and sup-
know for sure and Andyand Pat porter,but I've stopped using it....
were quite confident." If we were to do it over again,our
approachwould be not so much like
the Normandyinvasion,but more of
a vertical focus.... We brought a
style and conceptualapproachto an
area where it did not work."
Intel's strategy for bringing the PC into the living room (Hood River)
Rob Siegel (projectmanager)and his Andy Grove:"'ThePC is it,' Grove Siegel and his team continuedtheir
team identifiedthe target applica- declares. 'Thatsums up Intel'sbusi- efforts throughthe fallof 1996. But
tions and uses for the Hood River ness planand rallyingcry.''Some they raninto fundingproblemswhen
product.The design called for the thinkthe informationsuperhighway the idea of a "networkcomputer"
use of Intel's233 MHzPentiumII will come throughtheirTV,'Grove (NC)gained some trackingunderthe
processor,the highest performance proclaimed... .'[But]the information impulseof Oracle'sLarryEllison,and
CPUat the time. tool of the futureis on yourdesk, the Desktop ProductGroup(DPG)
not in your livingroom"'(Burstein reallocatedresources to meet the
By August 1996, Siegel: "... we and Kline, 1995: 24). perceivedthreatto the core busi-
had accomplisheda lot. We had ness. Siegel was able to get funding
Microsoftdoing what we wanted reinstated,but the marketfor Hood
them to do, and we had established Riverdid not develop as planned.
an impressivecustomer list. Inaddi-
tion, the ProductLineBusiness Plan Inearly 1997, MikeAymar(GMof
presentationwent well. We received DPG)haltedthe venture.
the highest rating,and AndyGrove
came up with the phrase, 'Hijackthe we expected the
Aymar:"Originally
TV,'which became our rallyingcry." venture to ... generate demand for
another1 millionPCs per year. But
marketprojectionswere for various
vendorsworldwideto ship only in
the tens of thousandsof units in '97
and '98 .... This is insufficient."

Continued
344/ASQ, June 2002
Coevolutionary Lock-in

Table 5 (Continued)

Views from below Intel'sstrategic intent Strategic action


If non-strategic, pay as you go
PCIchipsets as a new business
Andy Beran(financemgr): "We Ron Smith (GMchipset business): Senior MicroprocessorGroupexecu-
never would have gotten into the Regardinghis intentto develop the tives supportedSmith'sefforts to
business if we had to fight for inter- chipset business based on Intel's develop the chipset business. Smith
nal capacity.... It always would new PCItechnology: "AndyGrove was able to use the new PCItech-
have looked likea lower [return]to told me that we had no damn busi- nology to wrest controlof chipsets
the processors." ness doing PCI .... That was early away from PC OEMsand make the
on. He and I had a heated discussion chipsets an importanttool for sup-
Berannoted that top management about it. ... He basically said some- portingthe launchof Intel'snew
let them keep the cash they generat- thing to the effect of who do we Pentiumprocessor. Havingsucceed-
ed with old productsto fund the thinkwe are, a chip companythink- ed in the face of corporateambiva-
development of chipset business: ing we are going to drivean I/Obus lence, Grovewrote Smith a note
"Atthe pointwhere that wasn't standard?" saying, "AndI said it couldn'tbe
enough, we were alreadysuccessful done." Fromthen on, Groveviewed
enough to keep going." chipset business as strategically
importantfor the core business.
RandyWilhelm(technicalmgr.):
"Therewas some doubt, I think,in
certainpartsof Intelthat we were
able to push a bus standard,where-
as in the past we had always had
key OEMspushingthe bus stan-
dard."

EricMentzer(marketingmgr.):"They
said, we don't believe you guys are
going to be successful, so we don't
want you going into those
accounts.... The processor division
was out tellingthe field sales force
and the customers, don't use this;
use the low-riskthing."

Networking as a new business


FrankGill(GMNetworking):"First,in Grove:"Therewas a time when I Grovefunded both opportunities,but
the early 1990s, there was Andy could have flippeda switch between he said, "Muchmore fundingwas
Grove'sabilityto get everybodyto videoconferencingand networking." going to videoconferencing."
focus on job 1.... Any other activity
was viewed as a distraction. .... A "Ihave been rabidabout four things Grovedid not allow much time for
second factor was that .... Since all in my careerat Intel:motherboards, discussion of the networkingbusi-
the planningactivityinvolvingAndy IntelInside,chipsets, and videocon- ness duringthe strategic long-range
focused on job 1, he did not have ferencing.What if I had been equally planningsessions of the early-to-
sufficientinsightor knowledgeto rabidabout networking?Intelcould mid-1990s.
meaningfullycontributeto our net- be a very differentcompany."
workingand connectivitybusiness- As of 1997, FrankGill:"Mark[Chris-
es." Reflectingon strategicdiscussion tensen] clearlygot networkingbetter
concerningthe networkingbusiness connected within Intel.He came up
"In1994-95, Andywould tell me, with FrankGill,Grovesaid, "Iam not with the fast Ethernet'big pipes
'Frank,I make a billiondollarsin happywith statements that are need big processors' notionand
profitper quarterand you make a bil- somewhat right,but mostly wrong. buildingremote management hooks
liondollarsin revenue per year.This Maybe I am too good for my own into the networkcards. He also put
is all distraction,so focus on Job 1."' good. I weed out all the weeds, but more focus on OEMcustomers
also some of the potentialseeds .... where Intelhad channel power."
MarkChristensen(Gill'ssuccessor): Barrettis more comfortablewith
"Forthe first six years, from 1991 to leavingstrategy a bit more murky, After 1997, networkingwas viewed
1997, it was basically'pay your own undefined." as partof the corporatestrategy,
way' for growth. Ifyou didn'tgrow, leadingto a majoracquisitionand full
you had the threatof getting down- corporatesupportfor growingthe
sized. Muchof the fundingwas business.
being funneled into programsthat
would help microprocessorgrowth-
Job 1."

345/ASQ, June 2002


only of the core business. The networkingcase shows that
these difficultiesare exacerbated if the linkwith the core
business cannot be easily established, thereby limitingits
growth fundingto the resources that it can generate on its
own. Table6 summarizesthe comparativeanalysis of these
four cases using the process model of internalcorporateven-
turing(Burgelman,1983). The process model identifies key
interlockingactivities of differentlevels of management (cor-
porate, middle,and venture) in the core (definitionand impe-
tus) and overlaying(structuralcontext and strategic context)
subprocesses of venture development.
Strategic Inertia I: If Strategic, Apply Logic of Core
Strategy
The ProShare case. The ProShareventure's purpose was to
make videoconferencinga standardPC capability,which
would help create additionaldemand for microprocessor
power. Grove'sstrategic intent determinedthe strategic con-
text for the venture from the start (table 5). His support
shielded the venture from the strong selection pressures of
the structuralcontext, in particularIntel'srigorousfinancial
reviews. Grovegot deeply involvedin monitoringthe defini-
tion of the venture strategy and in authorizingfundingof its
development (table6). Ina fall 1999 discussion with an MBA
class, Grove mentioned that Intelhad spent about $750 mil-
lion on the unsuccessful venture. His insistence on applying
the horizontal,frontalassault strategy of the microprocessor
business to ProSharereduced the degrees of freedom of the
executives in charge of the new business development effort
(table 5). Pat Gelsinger'stask was to delivera technology to
the marketin the same way that Inteldeliverednext-genera-
tion microprocessorsto the market.Technicaland need-linking
efforts were limitedin their effectiveness, discipline-instilling
product-championing efforts were not requiredto secure
resources internally,and the effectiveness of strategic forcing
efforts to secure a fast-growingbeachhead in the market
was limited(table6). FrankGill,the senior executive posi-
tioned between Groveand Gelsingerwas left-or rather,as
he put it, "ableto stay"-out of the loop. With Groveper-
formingthe role of Gillin the strategic context determination
process, the discipline-instilling organizationalchampioning
efforts-requiring Gillto convince peers, as well as top man-
agement, that the continuationof the videoconferencingven-
ture was in the long-terminterest of the corporation-were
not required(table6). Finally,as a consequence of the early
and sustained supportfrom the CEO,the opportunitycosts
associated with ProSharewere not considered until1996,
when Gillwas asked to scale the venture down (table 5).
The Hood River case. The Hood Riverventure's purpose
was to bringthe PC into the livingroom as an electronic
entertainmentdevice. Hood Riverwas started as a seed pro-
ject with initialfundingfrom Intel'sCorporateBusiness
Development group in February1996. The venture's strategy
was influencedfrom the start by Grove'spubliclystated
strategic intent that the "PCis it," which was taken to heart
by Rob Siegel, the projectleader (table 5). This drove the
technical and need-linkingefforts in the Hood Riverproduct
definition.Since there was no directand forcefulsupport
346/ASQ, June 2002
Coevolutionary Lock-in

Table6
Intra-companyLevel of Analysis: Comparative Process Model Analysis of Four New Business Development
Cases during Epoch II

Leadership activities by
management level and
subprocess* ProShare Hood River Chipsets
Corporatemanagement level:
Definition:Monitoring Fromthe start Fromthe start Flyunderradar
Impetus:Authorizing Fromthe start Erratic Pay as you go
Strategiccontext: Rationalizing Premature Didn'tget to Lagging(linkto
Pentium)
Structuralcontext: Structuring Suspended Stronginfluence Stronginfluence
Selecting (linksto structuraland
strategic contexts) Suspended Stronginfluence Strong influence
Middlemanagement level:
Definition:Coaching Limited Limited Strong
Impetus:Strategicbuilding Didn'tget to Didn'tget to Not necessary
Organizationalchampioning(links
impetus and strategiccontext) Not necessary Didn'tget to Strong
Strategiccontext: Delineating Premature Didn'tget to Strong
Structuralcontext: Negotiating Not necessary Ineffective Strong
Venturemanagement level:
Definition:Technicaland need linking Limitedeffectiveness Ineffective Effective
Productchampioning(linksdefinition
and impetus) Not necessary Ineffective Effective
Impetus:Strategicforcing Limitedeffectiveness Ineffective Effective
Strategiccontext:e.g., bootlegging Littleroom Not possible AnticipatePentium
Structuralcontext: Questioning Littleroom Ineffective Workaround
Leadership activities by
management level and Networking
subprocess* Until 1997 Changes after 1997
Corporatemanagementlevel:
Definition:Monitoring Littleinterest Strong
Impetus:Authorizing Pay as you go Strong
Strategiccontext: Rationalizing Lagging Linkto core
Structuralcontext:Structuring Stronginfluence Adjusted
Selecting (linksto structuraland
strategiccontexts) Stronginfluence Adjusted
Middlemanagementlevel:
Definition:Coaching Strong
Impetus:Strategicbuilding Limited Strong
Organizational championing(links
impetus and strategiccontext) Give up Strong
Strategiccontext: Delineating Limited Strong
Structuralcontext: Negotiating Defensive Strong
Venturemanagementlevel:
Definition:Technicaland need linking Effective
Productchampioning(linksdefinition
and impetus) Effective
Impetus:Strategicforcing Effective
Strategiccontext:e.g., bootlegging Limited
Structuralcontext:Questioning Workaround
*Source: Burgelman(1983).

from the CEO for this project, the selective effects of the
structural context were very strong (table 6). This was evi-
dent when funding was temporarily cut off without warning
in December 1996 to harness resources in the face of the
perceived threat of the "network computer" (NC) to Intel's
core business. Ineffective technical and need-linking activities
made it difficult to collaborate with the consumer electronics
OEMs, who had a very different view of the market and the
347/ASQ, June 2002
technology required.Siegel triedto pursue product-cham-
pioningactivities, but, as a relativenewcomer, he could not
exert influence in the networkof resource-controllingrelation-
ships of Intel'smatrix.Strategicforcingnever got started, as
no consumer electronics OEMs or PC OEMswere willingto
adopt the Hood Riverproductconcept (table6). As a result,
MichaelAymar,the middle-levelexecutive, had no foundation
to buildon and could not continue to ask top management
for support. He stopped fundingHood Riverin 1997 (table5).

Strategic Inertia II:If Non-strategic, Pay as You Go


The PCIchipset case. IntelArchitectureLabsdeveloped the
PCIbus technology in the early 1990s. Top management's ini-
tial intentionwas to organizea consortiumto bringPCIto the
PC industryas an enablingtechnology for the core micro-
processor business, as the previousbus standardwas too
slow to take advantageof increased processing power.
Determinationof the strategic context of the PCIchipset ven-
ture was laggingbecause Grovewas opposed to the idea of
turningPCIchipsets into a business (table 5). Ron Smith nev-
ertheless decided to pursue PCIchipsets as a new business.
He tried to "flyunderthe radar"to protect the venture from
close top management scrutinyto builda viable business
foundation(table6). He assembled a team of experienced
functionalmanagers who were well connected with the rest
of the corporationand could access resources that would
otherwise not be available.These managers engaged in care-
ful technicaland need linkingto define Intel'schipset oppor-
tunity.Realizingthey would not be able to secure scarce
manufacturingcapacityinternallyagainst the more profitable
microprocessors,their productchampioningefforts took the
form of contractingwith outside manufacturers.Smith con-
vinced his team that winning inside requiredwinningoutside
throughsuccessful strategic forcing (table6). Eachyear, the
venture delivered more than it had promised,which gave
senior executives such as AlbertYu, PaulOtellini,and Craig
Barretta reason for supportingit in the face of Andy Grove's
doubts. The chipset venture's potentialwas sufficientlylarge
that no additionalbusiness opportunitiesneeded to be found
to reach criticalmass. Ron Smith did not have to engage in
strategic building,which requiresthe agglomerationof addi-
tional business opportunitiesthroughinternaltransferof pro-
jects and/orthroughcarefullytargeted acquisitions,and could
focus on coaching the venture team. Also, Smith had antici-
pated that the PCIchipset would be importantfor facilitating
the launchof the Pentiumprocessor in 1993 and had
instructedthe team to design the chipset accordingly.
Smith's predictionturnedout to be correct,which facilitated
the determinationof the strategic context lateron (table6).
Eventually,Groveconcluded that the chipset venture was an
importantbusiness for Intel.His retroactiverationalization
concluded the process of strategic context determination.
Fromthen on, it had his full support.
The networking business case. Inthe early 1990s, Frank
Gill'scharterwas to develop new businesses for Intel,but
because of the enormous growth of the core business, Andy
Grove began to view these efforts as a distraction.Grovefelt
that Gillwas too much focused on the success of the net-
348/ASQ, June 2002
Coevolutionary Lock-in

workingbusiness and not enough on that of the core busi-


ness. Grovealso felt uncomfortablewith the lack of clarityof
the networkingstrategy (table 5). Gillpointed out that Grove
had been totallyfocused on the core microprocessorbusi-
ness and that the strategic long-rangeplanningprocess
(SLRP)spent littletime on businesses that were not consid-
ered strategic. Top management looked at networkingas an
industryenabler,ratherthan as a new business. Gillgave up
on the organizationalchampioningefforts in the face of peer
resistance and top management's recalcitranceand focused
on short-termfinancialperformanceto protect the business
(table6). This created a vicious circle. Unsuccessful organiza-
tional championinglimitedthe amount of corporateresources
made availablefor the networkingbusiness, which limited
the scope of the strategic buildingactivities that Gillcould
engage in: large acquisitionswere simply not permitted.And
this, in turn, limitedthe growth of the business in the fast-
growing industryto what could be achieved with the strate-
gic forcing activities based on the internallydeveloped prod-
ucts. Fortunately,these were the result of effective technical
and need linkingand experienced strong marketacceptance.
The effectiveness of these activities was at least in partthe
result of Gill'sstrong coaching of his team and successful
shieldingof the networkingbusiness from top management.
Only in 1997, when a new general managerwas able to
show the importanceof networkingfor the microprocessor
business and for Intel'sfuture growth, was its strategic con-
text determined, and it received full top management
support.
DISCUSSIONAND CONCLUSIONS
Coevolutionary Lock-in
This study shows how Andy Grovewas able to take advan-
tage of the fortuitouscircumstances Intelfaced in its micro-
processor business with the rapidascendance of the IBMPC
and to turngood luck into a strategy vector. He made Intel
focus on a narrowbusiness strategy and established an
induced strategy process that tightlyaligned strategy and
action and producedextraordinarysuccess. His deep under-
standing of the forces that gave rise to the strategy vector
also gave him great confidence in dealing with several crises
that challenged it. This study, however, also reveals the com-
plex reciprocalcausation between Grove'sstrategic intent
and the structures and processes that he put in place and
how the very success of the strategy vector resulted in the
emergence of coevolutionarylock-inand impeded new busi-
ness development. AlthoughGrovewas a master of strategy,
who knew that Intelwould have to transformitself again
eventually,he and Intelwere subject to inescapable evolu-
tionarydilemmas associated with the dynamics of coevolu-
tionarylock-in.
Intel'ssuccess as the sole source of the highest value com-
ponent of PCs made it increasinglyable to appropriatethe
availablerents in the PC marketsegment. But this asymme-
try created a positive feedback loop, requiringIntelto make
more and more of the investments necessary to enable
adoptionof next-generationmicroprocessors.These comple-
349/ASQ, June 2002
mentarystrategic thrusts helped Intelto controlits external
environment(Pfefferand Salancik,1978), but they also
increasinglytied its strategic directionand economic fortunes
to the evolutionof the PC marketsegment. Coevolutionary
lock-inengendered by strategic dominance entangled Intelin
a system of relationshipsthat reduced its freedom of action,
a paradoxwell articulatedby J. G. March:"Youcan have
autonomy or you can have power but you cannot have both.
Power depends on linkagesand linkagesdestroy autonomy"
(personalcommunication).
As a drivingforce of the PC marketsegment, Intelwas able
to influencethe pace of industrychange. Such time-paced
strategy is a powerfulalternativeto event-paced strategy
(Gersick,1994; Brownand Eisenhardt,1997). It allows a
company to dictate the pace of strategic change that other
players-customers, competitors, suppliers,and complemen-
tors-must adhere to. Intel'stime-paced strategy, however,
did not simply try to impose its strategic intent on the prod-
uct-marketenvironmentunilaterally.Grovehad learnedthat
there was a naturaladoptioncycle in the PC marketseg-
ment, with a periodof about three years between the maxi-
mum ramp-upfor differentmicroprocessorgenerations. He
also knew that Intelcould not expect to change that much.
At the same time, havingput in place the competencies and
support infrastructureto deliver new generations of micro-
processors to the PC marketsegment, there was a strong
internaldriveto do so. Intel'stime-paced strategy thus rein-
forced the lock-inwith the PC marketsegment. Also, Intel
was able to win the two defining battles in the microproces-
sor marketsegment-against other IntelArchitecturesuppli-
ers and against the RISCarchitecture-that Grovehad identi-
fied in late 1993. But Intel'scompetitive intensityincreasingly
specialized the company's competitive repertoirefor the PC
marketsegment (Barnett,1997), furtherreinforcingcoevolu-
tionarylock-in.
Intel'sintroductionof the Celeronprocessor in 1998 to
counter AMD at the low end of the PC marketsegment testi-
fies to the company's relentless competitive intensity.The
need for a crash effort to introducethe Celeronprocessor,
however, also suggests that Intel,while continuingto inno-
vate at a high rate with its Pentiumprocessor productfamily,
had begun to produce innovationsthat were less in tune with
evolving environmentaldemands (Sorensen and Stuart,
2000). Intelseemed to have difficultiesrecognizingthat the
importanceof the externalselection environmentrelativeto
the internalselection environmentwas increasingtoward the
end of Epoch II(Sorenson, 2000). Intel'sdifficultiesin this
respect seem consistent with the observationthat in suc-
cessful organizationsthere will be a naturaltendency for
internalselective-retentive processes to dominate external
ones (Miller,1999: 94). Coevolutionarylock-inmay thus be an
extension and furtherelucidationof the sources of structural
inertia(Hannanand Freeman, 1984).
Extraordinary success associated with coevolutionarylock-in
heightened Andy Grove'sconfidence in the logic of the core
business strategy (e.g., Prahaladand Bettis, 1986). But
Grove'sdirect involvementin ProSharemade it difficultfor
350/ASQ, June 2002
Coevolutionary Lock-in

the middle-levelexecutive in charge to develop a strategy


that was appropriatefor the new business and to act in
accordance with an objective analysis of the situation.
Grove'sapproachin the ProSharecase supports Audia,
Locke, and Smith's (2000) suggestion that success may
increase a decision maker'sfeelings of self-efficacy. It con-
firms that the inertiaof success is often best understood in
terms of the strength of the decision maker'sbeliefs in the
validityof the currentstrategy, ratherthan in terms of com-
placency or driftingwithout furtherattempts at improvement
(Millerand Chen, 1994). It also supports Miller's(1994) find-
ing that decision-makingstyles tend to be more extreme dur-
ing periods following success than duringperiods following
poor or mediocre performance.Grove eventuallycame to
realizethis, but his strong involvementearly on, before major
marketand technical uncertaintieshad been reduced, led to
escalation of commitment and prevented scaling down or
timely exit from the failingbusiness. This raises important
questions about the limitationsof using top management's
strategic intent as a means for guided evolution (Lovasand
Ghoshal,2000). WithinIntel'sinduced-strategyprocess, guid-
ed evolutionworked fine: many new projects related to the
strategic intent expressed in the microprocessorroad map
were useful variationsthat were effectively selected and
retained.When Grovetried to use strategic intent to shape
new variationsthat were not commensurate with the logic of
the core business, however, the result was misguided evolu-
tion.
Much of Intel'sR&Dinvestments went into technologies that
complemented the microprocessorand thereby offered
opportunitiesto launchnew businesses, but the company
rarelyattempted to do so. One reason for this was that any
technology advance that enriched the PC marketsegment
was likelyto create more demand for microprocessors,
which had very high margins.Thus, it was generally more
valuablein the short runto give away technology and quickly
disseminate it in the market,ratherthan try to builda busi-
ness aroundit. This suggests the powerful effect that finan-
cial strategy and capitalmarketconsiderationsmay have on
product-marketstrategy. It also indicates, however, another
strong structuralinertialconsequence of coevolutionarylock-
in. As Intel'sextraordinarylucrativecore business continued
to grow very fast in the mid-1990s, Grove began to consider
non-core business development as a distraction.Consequent-
ly, it was increasinglydifficultfor non-core new businesses to
command top management attentionand corporate
resources. This was exacerbated by Intel'sstructuralcontext,
which facilitatedexecution of the core business strategy but
was less able to deal with non-core new business develop-
ment: strategic planningwas almost exclusively focused on
the core business. New general management talent was not
easily developed in Intel'smatrixorganization.Resource allo-
cation favored the core business, and new businesses were
constantly in danger of experiencing randomshocks when
criticalresources were taken away to cope with a perceived
threat to the core business. The measurement and reward
system was unforgivingfor deviations from objectives, even
though new business strategies requiresuch flexibility.While
351/ASQ, June 2002
many new ideas continuedto emerge, the structuralcontext
made it difficultto activate the process of strategic context
determinationnecessary to linkthe new business to the cor-
porate strategy.
Implications for Theory
The causes and consequences of coevolutionarylock-insug-
gest that this little-noticedprocess might help illuminate
some of the inescapabledilemmas in the naturaldynamics of
organizationaladaptation.It also helps connect ideas about
the internalecology of strategy making,the modern econom-
ic theory of the firm,and an evolutionaryperspective on orga-
nizationallearning.
Organizational adaptation. Previousfindingsbased on a
study of Intel'stransformationduringEpoch I (Burgelman,
1991, 1994) supportedthe propositionthat companies that
are successful over long periods of time maintaintop-driven
strategic intent, throughthe inducedstrategy process, while
simultaneouslymaintainingbottom-up-drivenstrategic renew-
al, throughthe autonomous strategy process. Recent efforts
by scholars to formalizeparts of the inducedand
autonomous strategy processes frameworkseem to support
this proposition.In Rotembergand Saloner's(2000) mathe-
maticalmodel, the firm employs a visionaryCEOwho is con-
sistently biased in favorof certainprojects but who leaves
the door open for pursuingsufficientlygood opportunities
outside the existing vision. They have shown that this may
offer greater profit-maximizing possibilitiesthan committing
to a narrowbusiness strategy (Rotembergand Saloner,
1994). They showed the importantrole played by objective
middle managers supportingpromisingprojectsoutside the
CEO'svision. Importantly,they also showed that the CEO
must not interferewith the autonomy of middle managers in
allocatingresources to autonomous projects.
The study of Grove'stenure as CEOinitiallycast doubt on
the importanceof the autonomous strategy process. Like
other great leaders, Grovewas able to recognize the unique
opportunitiesfacing Inteland to mobilizehis organizationto
exploit them by creatingan extremely focused induced strat-
egy process. If the growth of the PC marketsegment had
continued unabated,Intel'sinduced strategy process would
probablyhave sufficed to secure continuedadaptation,there-
by reducingfurtherthe relevance of the autonomous strategy
process. This would have underminedthe validityof the
internalecology perspective on strategy making.Towardthe
end of Epoch II,however, it became clear that Intel'sfuture
growth would also depend on new business development
and that the strategies for new businesses might have to be
defined by general managers who were closer to the front
line. Inertialconsequences of coevolutionarylock-in,howev-
er, had significantlyreduced the effectiveness of Intel's
autonomous strategy process. Figure1 providesa schematic
representationof the paper'score theoreticalidea: a compa-
ny's relentless and successful pursuitof a narrowbusiness
strategy throughthe inducedstrategy process may produce
coevolutionarylock-inand reduce the effectiveness of the
352/ASQ, June 2002
Coevolutionary Lock-in

Figure 1. Effects of a strategy vector on the internal ecology of strategy making.

Emerging New
Product-market
Environments

Strategic InertiaII

Autonomous st ic
Strategic ----. . ---
Action

4 Narrow Current
I"""^
Strategic Inertia Business Product-
Environment
Induced Tight
Strategic - Structural
Action Context

autonomous strategy process, which weakens a company's


long-termadaptation.
The heavy lines in figure 1 indicatethe reinforcementof
Intel'sinduced strategy process, the creation of the strategy
vector, and the coevolutionarylock-inwith the PC market
segment that it engendered. Figure1 also shows the impact
of the sources of inertiaassociated with coevolutionarylock-
in on the autonomous strategy process. Some initiativesthat
needed to be pursued throughthe autonomous strategy
process were erroneouslysubjected to the logic of the
induced strategy process (StrategicInertiaI);others faced
Intel'sreduced abilityto activate strategic context determina-
tion processes (StrategicInertiaII).
Figure1 illuminatesinescapable evolutionarydilemmas aris-
ing in the naturaldynamics of organizationaladaptation.
Grove'sstrategic leadershipapproximatedthe classical ratio-
nal-actormodel in pursuingIntel'senormous opportunityin
the PC marketsegment, but at the cost of reducing Intel's
capabilityto develop new businesses. Was this a mistake?
This study suggests that objective necessities arisingfrom
the coevolutionarylock-inof the induced strategy process
and the product-marketenvironmentwere a majorcause of
the relativeneglect of the autonomous strategy process. The
resource requirementsof pursuingthe microprocessorbusi-
ness, especially top management time and attention, did not
leave much room for alternativepursuits.And the short-term
opportunitycosts of pursuingthe microprocessorbusiness
were perceived as low. Also, it seems quite possible that
Andy Grove passed on the CEObaton to CraigBarrettin
early 1998 when he realizedthat a new, less singularly
focused strategic leadershipapproachwas necessary and
there was still time to rebuildIntel'snew business develop-
353/ASQ, June 2002
ment capability.Alternatively,might an effort to maintainthe
internalecology of strategy makinghave severely hampered
the firm?Does optimallong-termadaptationfollow a punctu-
ated equilibriumpattern(e.g., Tushmanand Romanelli,1985),
perhaps involvinga series of discrete periods, each focused
on maximallyexploitingthe availableopportunities,rather
than a more continuous evolutionaryprocess of balancing
exploitationof availableopportunitiesat a given time with
preparingthe groundfor future growth opportunities?This
study cannot definitivelyanswer these alternativequestions.
Its findingssuggest, however, that without majoracquisi-
tions, the likelihoodof moving instantaneouslyand discontin-
uously from one period'sopportunityfrontierto that of anoth-
er is low. Forinstance, it took more than ten years for
microprocessorsto become Intel'snew core business. In
2002, Intelmanagement realizes that large new businesses
do not emerge fullyformed out of the blue. Recognizingthe
possibilityof alternativedevelopmentalpaths, this paper's
identificationof coevolutionarylock-innevertheless casts
new lighton the role of strategy makingas a long-termadap-
tive organizationalcapability.This advance of administrative
science providescompany leaders responsiblefor designing
the strategy-makingprocess with a conceptualframeworkfor
consideringmore explicitlyand sooner the trade-offsinvolved
in balancinginducedand autonomous strategic processes
and exploitationand explorationin organizationallearning.
Strategy and learning. This study's findingsraise the ques-
tion of whether inducedand autonomous strategy processes
are fundamentallyat odds with one anotheror can be effec-
tively pursued simultaneously.Maintainingthe simultaneityof
induced (variationreducing)and autonomous (variation
increasing)strategy processes may involvedifficultiessimilar
to maintaininga balance between exploitationand exploration
processes in organizationallearning(March,1991). Both
processes compete for limitedresources, and company lead-
ers necessarily make trade-offsbetween them. Giventhe
extraordinaryopportunitiesIntelfaced in the core business,
focusing on learningthat increased its mean performance
ratherthan on learningthat could increase the varianceof
performanceseemed rational(March,1991: 82). Also,
Grove'sabilityto vectorize everybodyat Intelin the same
directionled to quickconvergence of individualbeliefs
(strategicinitiatives)and the organizationalcode (the corpo-
rate strategy) (March,1991: 75). Intelexperienced turnover
because the lowest 10 percent of individualperformerswere
systematicallyreplaced, but this also ensured the rapid
socializationof new employees to Intel'sorganizationalcode
because they were keen to understandIntel'sperformance
expectations, which were clearlytied to implementingthe
core strategy. Overall,Intel'sinduced-strategyprocess during
Grove'stenure as CEOfavored organizationallearningthat
was maximallyconcerned with exploitation.
Explorationinvolves experimentation(March,1991) and is
viewed here throughthe lens of the autonomous strategy
process, which dissects explorationinto autonomous strate-
gic initiativesand the process of strategic context determina-
tion. The strategic context determinationprocess, which
354/ASQ, June 2002
Coevolutionary Lock-in

depends criticallyon the general management abilitiesof


middle-levelexecutives, helps companies turn exploration
efforts into new exploitationopportunities.The distinction
between exploratoryinitiativesand the strategic context
determinationprocess helps explainthe mixed recordof new
business development duringIntel'sEpoch II.In spite of
Grove'sefforts to vectorize everybody in the same direction,
numerous autonomous strategic initiativescontinued to
emerge, indicatingcontinued attempts at exploration.The
decrease in Intel'scapacityto activate strategic context
determinationprocesses, however, prevented the company
from exploitingthe more viable autonomous initiatives.
Strategiccontext determinationprocesses thus appearto be
the crucialnexus between explorationand exploitationand
key to balancinginduced and autonomous strategy processes
effectively. Strategiccontext determinationprocesses com-
plement a company's structuralcontext in importantways.
They make it possible to suspend the selective effects of the
structuralcontext, which almost unavoidablytends to
become fine-tunedfor supportingtop management's current
strategic intent. And they serve to create links between
autonomous strategic action and the company's strategy,
thereby amending it. The capacityto activate and successful-
ly complete such processes can be viewed as a measure of
the intelligence of the company's internalselection environ-
ment and may be at the very heart of strategy makingas an
adaptiveorganizationalcapability.
This study's main contributionsconcern the naturaldynamics
of organizationaladaptation.An evolutionaryperspective on
strategy makinghelps bridgeand extend related ideas about
the benefits and potentialopportunitycosts of narrowbusi-
ness strategies in the modern economic theory of the firm
and ideas about exploitationand explorationin theory about
organizationallearning.Fine-graineddetail of a strategy-mak-
ing process approximatingthe classical rational-actormodel
suggests that the pursuitof focus and efficiency may also
become the potentialenemy of effective explorationand
strategic renewal. Strong positive environmentalfeedback
strengthens the relativeimportanceof the internalselection
environmentbut also causes coevolutionarylock-in,which is
a double-edged sword: strategic dominance begets depen-
dence. The relativedominance of the internalselection envi-
ronment may last a long time, more than ten years in the
case of Intel'sEpoch 11,but eventually,cumulativechanges in
the externalselection environmentare likelyto reduce its
efficiency. Coevolutionarylock-inexacerbates tendencies
toward structuralinertiain novel and potentiallyinsidious
ways because it affects the balance between induced and
autonomous strategy processes and a company's abilityto
develop new businesses and, hence, the long-termadaptive
capabilityof its strategy-makingprocess.
Conclusionsfrom a single case study warranthealthy cau-
tion, but by examininga case of extraordinarily successful
CEO-drivenstrategy makingthat approximatedthe classical
rational-actormodel, this paper provides furthersupportfor
the internal-ecologymodel of strategy makingas an adaptive
organizationalcapability.There is little doubt that companies
355/ASQ, June 2002
that find themselves in the fortuitouscircumstances that
Intelfaced in the PC marketsegment after its defeat in the
DRAMbusiness can greatlybenefit from a leader with an
exceptional abilityto capitalizeon them. Yet the benefits of
the rational-actormodel must be tempered by the realization
that in dynamicenvironments,even in digitalindustrieschar-
acterized by winner-take-allcompetition,the relativestrength
of the company's strategy vector will eventuallydecline,
because the forces that make periods of extraordinarysuc-
cess possible are unlikelyto last forever.The inertialconse-
quences of coevolutionarylock-in,however, are likelyto
lingeron if company leaders do not address them. An organi-
zation's long-termadaptation,spanning multiplegenerations
of CEOs, may therefore criticallydepend on maintainingthe
strategic renewal capabilityof its internalecology of strategy
making.
REFERENCES

Aldrich, H. Barnett,W. P.,and M. T. Hansen 2002 Strategyis Destiny:How


1999 OrganizationsEvolving.Lon- 1996 "Thered queen in organiza- Strategy-Making Shapes a
don: Sage. tionalevolution."Strategic Company'sFuture.New York:
Allison, G., and P Zelikow ManagementJournal,Sum- Free Press.
mer Special Issue, 17:
1999 Essence of Decision:Explain- 139-157. Burgelman, R. A., D. L. Carter,and
ing the CubanMissile Crisis, R. S. Bamford
2d ed. New York:Addison Baum, J. A. C., and B. McKelvey 1999 "IntelCorporation: The evolu-
Wesley Longman. (eds.) tion of an adaptiveorganiza-
1999 Variationsin OrganizationSci- tion." StanfordBusiness
Arthur,B. ence: In Honorof DonaldT. School Case SM-65.
1987 "Competingtechnologies:An
overview." In G. Dosi (ed.), Campbell.ThousandOaks,
CA:Sage. Burstein, D., and D. Kline
TechnicalChangeand Eco- 1995 "Inthe square-offbetween
nomic Theory:590-607. New Bendor, J., and T. H. Hammond TVand computer,the smart
York:ColumbiaUniversity 1992 "RethinkingAllison'smod- money mightbe on the boob
Press. els." AmericanPoliticalSci- tube." Los Angeles Times,
ence Review,86: 301-322. October29.
Audia, P G., E. A. Locke, and K. G.
Smith Bower, J. L. Christensen, C. M., and J. L.
2000 "Theparadoxof success: An 1970 Managingthe ResourceAllo- Bower
archivaland a laboratorystudy cation Process. Boston: Har- 1996 "Customerpower, strategic
of strategic persistence fol- vardBusiness School Press. investment,and the failureof
lowing radicalenvironmental leadingfirms."StrategicMan-
change." Academy of Man- Brown, S. L., and K. M.
Eisenhardt agement Journal,17:
agement Journal,43: 1997 "Theartof continuous
197-218.
837-853.
change: Linkingcomplexity Cogan, G. W., and R. A.
Bamford, Raymond S., and theory and time-pacedevolu- Burgelman
Robert A. Burgelman tion in relentlesslyshifting 1991 "IntelCorporation(C):Strate-
1997a"IntelCorporation: The Hood organizations."Administrative gies for the 1990s." Stanford
RiverProject(A)."Stanford Science Quarterly,42: 1-34. Business School Case PS-BP-
Business School Case SM- 256C.
49A. Burgelman, R. A.
1997b"IntelCorporation: The Hood 1983 "Aprocess model of internal Collins, J. C.
RiverProject(B)."Stanford corporateventuringin the 2001 Goodto Great.New York:
Business School Case SM- diversifiedmajorfirm." HarperBusiness.
49B. AdministrativeScience Quar-
terly,28: 223-244. Dyck, B., and F.A. Starke
1998 "Intel'sstrategicposition in 1999 "Theformationof breakaway
the familyroom, 1998." Stan- 1991 "Intraorganizational ecology
of strategy makingand orga- organizations:Observations
ford Business School Case and a process model."
SM-50. nizationaladaptation:Theory
and field research."Organiza- AdministrativeScience Quar-
Barnett,W. P tion Science, 2: 239-262. terly,44: 792-822.
1997 "Thedynamicsof competitive 1994 "Fadingmemories:A process
intensity."AdministrativeSci- theory of strategicbusiness
ence Quarterly,42: 128-160. exit in dynamicenviron-
ments." AdministrativeSci-
ence Quarterly,39: 24-56.

356/ASQ, June 2002


Coevolutionary Lock-in

Farrell,J., H. K. Monroe, and G. March,J. G. Sorenson, O.


Saloner 1991 "Exploration and exploitation 2000 "Lettingthe marketworkfor
1998 "Theverticalorganizationof in organizationallearning." you: An evolutionaryperspec-
industry:Systems competi- OrganizationScience, 1: tive on productstrategy."
tion versus component com- 71-87. StrategicManagementJour-
petition."Journalof Econom- nal, 21: 577-592.
ics and Management Miller,D.
1994 "Whathappens after suc- S0rensen, J. B., and T. Stuart
Strategy,7: 143-182. cess: The perilsof excel- 2000 "Aging,obsolescence and
Fine, K. M., and R. A. Burgelman lence." Journalof Manage- organizationalinnovation."
1997 "IntelCorporation(F):Going ment Studies, 31: 85-102. AdministrativeScience Quar-
beyond success in 1997." 1999 "Selectionprocesses inside terly,45: 81-112.
StanfordBusiness School organizations:The self-rein-
Case S-BP-256F. Steere, D., and R. A. Burgelman
forcingconsequences of suc- 1993a"IntelCorporation(D):Micro-
cess." InJ. A. C. Baumand
Gersick, C. J. G. B. McKelvey(eds.), Variations processors at the cross-
1994 "Pacingstrategicchange: The in OrganizationScience: In roads."StanfordBusiness
case of a new venture." Honorof DonaldT. Campbell: School Case S-BP-256D.
Academyof Management 93-109. ThousandOaks, CA: 1993b"IntelCorporation(E):New
Journal,37: 9-45. directionsfor the 1990s."
Sage. StanfordBusiness School
Gilder,G. Miller,D., and M.-J. Chen Case S-BP-256E.
1989 Microcosm:The Quantum 1994 "Sourcesand consequences
Revolutionin Economicsand of competitiveinertia:A study Suzuki, 0., and R. A. Burgelman
Technology.New York:Simon of the U.S. airlineindustry." 1998 "ThePC-baseddesktop
and Schuster. AdministrativeScience Quar- videoconferencingsystems
terly,39: 1-23. industryin 1998." Stanford
Glaser, B. G., and A. L. Strauss Business School Case SM-51.
1967 The Discoveryof Grounded
Mintzberg,H., B. Ahlstrand, and Tredennick,N.
Theory.Chicago:Aldine. J. Lampel
1998 StrategySafari.New York: 1991 "1991:The year of the
Grove, A. S. RISC."Microprocessor
1993 "How Intelmakes spending Free Press.
Report,February6: 16.
pay off." Fortune,February Pfeffer,J., and G. R. Salancik
22: 57-61. 1978 The ExternalControlof Orga- Tushman, M. E., and E. Romanelli
1996 Onlythe ParanoidSurvive. nizations.New York:Harper& 1985 "Organization evolution:A
New York:Doubleday. Row. metamorphosismodel of con-
vergence and reorientation."
Hannan, M. T., and J. Freeman Pollack,A. In B. M. Staw and L. L. Cum-
1977 "Thepopulationecology of 1988 "An'awesome' Intelcorners mings (eds.), Researchin
organizations."American its market."New YorkTimes, Organizational Behavior,7:
Journalof Sociology,83: 171-222. Greenwich,CT:JAI
929-984. April3.
Press.
1984 "Structuralinertiaand organi- Prahalad, C. K., and R. A. Bettis
zationalchange." American 1986 "Thedominantlogic:A new Van de Ven, A. H.
SociologicalReview,49: linkagebetween diversityand 1992 "Suggestions for studying
149-164. performance."StrategicMan- strategy process: A research
agement Journal,7: 485-501. note." StrategicManagement
Isaacson, W. Journal,SummerSpecial
1997 "Themicrochipis the Reinhardt,A., I. Sager, and P. Issue, 13: 169-188.
dynamoof a new economy Burrows
drivenby the passion of 1997 "CanAndyGrovekeep profits Welch, J., with J. A. Byrne
Intel'sAndrewGrove."Time, up in an era of cheap PCs?" 2001 Jack. New York:WarnerBusi-
December 29: 46-51. Business Week, December ness.
22.
Kawamoto, D., and S. Galante Yin, R. K.
1998 "Thelegacy of AndyGrove." Rotemberg, J. J., and G. Saloner 1984 Case Study Research.Applied
CNET,March26. 1994 "Thebenefits of narrowbusi- Social Research Methods
ness strategies." American Series, 5. BeverlyHills,CA:
Leonard-Barton,D. EconomicReview, 84: Sage.
1990 "Adualmethodologyfor case 1330-1349.
studies: Synergisticuse of a 2000 "Visionaries,managers,and Zuckerman,E. W.
longitudinalsingle site with 2000 "Focusingthe corporateprod-
strategicdirection."RAND uct: Securitiesanalysts and
replicatedmultiplesites." Journalof Economics,31:
OrganizationScience, 1: Adminis-
de-diversification."
693-716. trativeScience Quarterly,45:
248-266.
Schlender, B. R. 591-619.
Levinthal,D., and J. G. March 1989 "Intelproduces a chip pack-
1993 "Themyopiaof learning."
ing huge power and wide
StrategicManagementJour- ambitions."WallStreet Jour-
nal,WinterSpecial Issue, 14: nal, February28.
95-112. 1996 "Aconversationwith the
Lovas, B., and S. Ghoshal lordsof Wintel."Fortune,
2000 "Strategyas guided evolu- July 8.
tion." StrategicManagement
Journal,21: 875-896.

357/ASQ, June 2002