Coopetition¶
A strategic relation in which the same actors simultaneously cooperate in one linked activity and compete over value or position in another.
Core Idea¶
Coopetition is a strategic relation in which the same actors simultaneously cooperate and compete in linked activities. They may exchange inputs or contribute to a common technical project while also contesting customers, market position or another scarce stake. The two modes need not occur in the same transaction, and neither cancels the other. Bengtsson and Kock's later formulation explicitly allows horizontal or vertical relations among two or more actors, correcting the overly narrow reading that only direct same-product rivals jointly building one resource qualify.[1][2]
The identity is the dual relation and its activity boundary, not an assured win–win outcome. Formal contracts, shared ownership, knowledge leakage, free-riding and profitable persistence are possible features or risks, not universal structural roles. A component supplier who also rivals its buyer and firms contributing to common open-source software while retaining separate strategies show distinct ways to realize the same pattern.[3][4]
Structural Signature¶
Sig role-phrases: overlapping actors → cooperative arena + contested stake → simultaneous activity boundary → dual strategic relation.
- Overlapping actors: the cooperating and competing parties must be the same identifiable actors, not unrelated pairs aggregated at industry level.[2]
- Cooperative arena: they coordinate, exchange or contribute in a defined activity that each can use. A supplier exchange can qualify without a jointly owned asset.[3]
- Competitive arena: those actors also contest value or position in a linked arena. This rivalrous component is necessary; mere collaboration is not enough.[2][3]
- Concurrent linkage and boundary: the modes coexist and can affect each other, while the activity distinction explains where cooperation occurs and where rivalry remains. An explicit contract or consortium may mark that boundary but is not required.[2][4]
What It Is Not¶
It is not a synonym for any partnership. If a supplier and buyer have no rivalrous stake, their transaction may be beneficial exchange but not coopetition. Two firms that compete but never cooperate likewise fail the dual test. Sequentially ending collaboration and later beginning rivalry is a change of strategy, not a simultaneous dual relation.[2]
Nor does coopetition require all participants to sell an identical product, own a common platform, agree on a formal governance scheme or achieve positive net returns. Bengtsson and Kock's revised definition encompasses horizontal and vertical ties; the PLOS supply-chain model treats component supply and downstream device rivalry as different arenas.[2][3]
Scope of Application¶
In smartphone manufacturing, a vertically integrated firm can supply components to another device maker while competing with it downstream. Kwok and Lee explicitly model Samsung as Apple's supplier and rival; they use bills of materials and a model-specific coopetition factor to compare periods in which component sourcing changed. The reported factor is an output of that study's model, not a universal measure of every dual relation.[3]
In company-hosted open-source AI projects, firms contribute to shared software while retaining distinct strategic interests. A mixed-method study of PyTorch, TensorFlow and Transformers reports strategic, non-strategic and contractual collaboration types and governance asymmetry. More concretely, one interview respondent described Microsoft Azure, AWS and Google Cloud as cloud-market rivals that each maintained integrations with Hugging Face's Transformers project. That is a specific rival-contributor setting, not a claim that every external commit is coopetitive.[4]
Clarity¶
Ask who, where and when for both modes. “Apple and Samsung are partners” suppresses the smartphone rivalry; “they are rivals” suppresses component supply. The category becomes precise when the same actors are shown trading or coordinating in one activity while contesting a linked payoff in another.[3]
The distinction also prevents equating a shared artifact with the entire relation. Open-source code can be a cooperative interface, but project governance, contributor motives and competition among products remain separate. A shared repository alone does not prove rivalrous payoffs among any two named contributors.[4]
Manages Complexity¶
Coopetition reduces a tangled network of transactions to four audit handles: actor overlap, cooperative activity, contested stake and boundary between the activities. This makes it possible to classify a vertical supply tie and a horizontal code-collaboration tie without pretending their governance is identical. The PLOS model further illustrates that changing one sourcing link can alter a measured coopetition factor, while the open-source study shows different collaboration configurations among projects.[3][4]
The reduction does not supply a universal payoff formula. It directs analysis to where coordination helps an actor and where the same actor must protect independent strategic interests. Contract terms, market share and governance power need separate evidence.[2][4]
Abstract Reasoning¶
Let actors \(A\) and \(B\) have a cooperative interaction \(C\) and a rivalrous interaction \(R\) during an overlapping period. A claim of coopetition requires evidence for both \(C(A,B)\) and \(R(A,B)\), plus a meaningful linkage: a component supply choice can affect the downstream product contest, or shared code work can influence separate firms' product strategies. Neither \(C\) alone nor \(R\) alone suffices. The formal notation is an analytic aid, not a published universal equation.[2][3][4]
From that classification, ask how a change in one arena affects the other. Kwok and Lee's model reports a lower Apple coopetition factor after Samsung display supply was removed in one period; that is consistent with the cooperative tie mattering to the dual relation. It does not demonstrate that every supplier substitution causally improves competition or welfare.[3]
Knowledge Transfer¶
The smartphone and software cases share same actors / cooperative arena / rivalrous arena / simultaneous boundary. In one, cooperation is component procurement and rivalry is device sales; in the other, cooperation is code contribution and rivalry concerns distinct strategic positions. The common structure transfers while exact market, asset and governance arrangements do not.[3][4]
That transfer is limited by evidence. The open-source authors observe several types of collaboration and governance asymmetry, not a universal claim that every contributor is a rival. For each pair, establish both modes; do not infer coopetition merely from participation in a project with competitors somewhere in its wider ecosystem.[4]
Examples¶
Apple–Samsung component supply. Mapped back: actors = Apple and Samsung; cooperative arena = Samsung components used in Apple phones; competitive arena = downstream smartphone contest; boundary = procurement versus device sales. The PLOS study models the sourcing relation and notes that replacing Samsung as a display supplier lowered its model's measured Apple coopetition factor between its two periods. That is a study-specific observation, not proof of an invariant degree of coopetition.[3]
Competing cloud providers' Transformers integrations. Mapped back: actors = Microsoft Azure, AWS and Google Cloud, whom a study respondent identifies as rivals in cloud compute; cooperative arena = maintaining integrations with the shared Hugging Face Transformers project; competitive arena = their distinct cloud offerings; boundary = shared software compatibility versus cloud-market rivalry. This is a respondent-reported case in a mixed-method study, not a quantified comparison of cloud-market outcomes.[4]
Structural Tensions¶
Shared contribution versus private differentiation. More contribution can improve a common input but may disclose know-how or strengthen a rival; withholding protects differentiation but can weaken the cooperative benefit. Neither maximal openness nor maximal enclosure is guaranteed to preserve both modes. Diagnostic: Which specific contribution changes the shared outcome, and which payoff remains contested?[2][4]
Coordinated interface versus actor autonomy. A host or supplier relationship can make collaboration efficient, yet concentrated control can increase another actor's dependence. Distributing control may improve autonomy while making collective decisions harder; concentrating it may simplify decisions while narrowing the other actor's options. Diagnostic: Who can change access, sourcing or rules, and how does that affect the concurrent rivalry?[3][4]
Measuring one tie versus inferring the whole strategy. A supplier link or shared commit is observable, but the simultaneous competitive relation requires separate evidence. Counting only transactions risks missing rivalry; labeling every competitor interaction coopetitive risks admitting cases with no actual cooperation. Diagnostic: Can both modes be documented for the same actors in the same period?[3][4]
Structural–Framed Character¶
Evaluative weight. Coopetition is not intrinsically beneficial or harmful; its value depends on the actors' objectives and effects on others. Human-practice dependence. Firms choose supplier, code and pricing strategies, so the relation is enacted through human/organizational practice rather than discovered as a physical law.[2][3]
Institutional origin. Strategy scholarship named and refined the concept, while contracts, markets and open-source governance shape particular realizations. No one institution or contract template defines it. Vocabulary travel. The simultaneous dual relation travels literally between supply chains and software projects, but a metaphorical “coopetition” between molecules has not established strategic actors or rivalrous payoffs.[2][4]
Import versus recognition. In a new sector, recognize the pattern only after identifying actor overlap, cooperative interaction and a linked contested stake. Merely importing the fashionable label onto a partnership or rivalry fails the test. Its character: mixed-framed—a repeatable relational strategy structure whose realization and evaluation depend on actors, markets and governance contexts.[2]
Structural Core vs. Domain Accent¶
Portable skeleton. Live Competition supplies the necessary rivalrous component: the actors contest a scarce position or payoff. The staged relation is composition/presupposes, not strict subsumption of Coopetition under pure Competition. Live Cooperation is a tempting parallel parent, but its specific costly-contribution/social-dilemma signature is not established in every supplier exchange; live Collaboration likewise requires joint work that a vertical procurement tie may lack.[2][3]
Domain-bound mechanism. The same actors deliberately maintain a cooperative interface while competing in another linked strategic arena. Smartphone component sourcing and open-source code contribution are different interfaces, and formal governance is contingent. The core identity is the concurrent activity boundary, not any shared asset or numerical optimum.[3][4]
Why not prime. The broad coexistence of opposed interaction modes may be a future-prime question, but the literal coopetition concept here is a strategy relation among actors with incentives, transactions and competitive positions. A physical system with two counteracting forces is not automatically coopetition. The Competition prime supplies a portable component; the dual business/organizational relation remains domain-specific.[2]
Instantiates / Related Primes¶
This entry presupposes Competition. A coopetitive relation includes a real rivalrous payoff component between its actors.
Relationships to Other Abstractions¶
Current abstraction Coopetition Domain-specific
Parents (1) — more general patterns this builds on
-
Coopetition presupposes Competition Prime
A coopetitive relation includes a real rivalrous payoff component between its actors.The live Competition prime requires rival actors contesting a scarce stake. Coopetition retains that component while the same actors simultaneously coordinate in another linked activity.
Hierarchy path (1) — routes to 1 parentless root
- Coopetition → Competition
Neighborhood in Abstraction Space¶
Coopetition sits in a moderately populated region (56th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.
Family — Competitive Strategy & Market Entry (9 abstractions)
Nearest neighbors
- Multimarket contact — 0.90
- Communicative Constitution of Organizations — 0.85
- Translation (sociology) — 0.85
- Multi-Instrument Coordinated Campaign — 0.85
- Patron–Client Relationship — 0.85
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
Pure cooperation lacks the linked rivalrous stake. Pure competition lacks concurrent cooperative interaction. Strategic alliance is one possible governance form, not the whole identity. Open coopetition is a narrower open-source or openly shared-resource manifestation. Win–win exchange need not involve competitors and therefore is not automatically coopetition.[2][4]
References¶
[1] Maria Bengtsson and Sören Kock, “Coopetition in Business Networks—to Cooperate and Compete Simultaneously,” Industrial Marketing Management 29 (2000), 411–426, publisher abstract. Full paper was not directly inspectable in this pass. registry ↩
[2] Maria Bengtsson and Sören Kock, “Coopetition—Quo vadis? Past accomplishments and future challenges,” Industrial Marketing Management 43 (2014), 180–188, publisher abstract. The article refines the definition beyond strictly horizontal relations; full paper was not directly inspectable in this pass. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o
[3] J. J. M. Kwok and D.-Y. Lee, “Coopetitive Supply Chain Relationship Model: Application to the Smartphone Manufacturing Network,” PLOS ONE 10(7) (2015), e0132844, especially supply-chain relationship model and Figure 4. The case study is model-based and partly hypothetical, not a causal market-outcome study. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p
[4] Cailean Osborne, Farbod Daneshyan, Runzhi He, Hengzhi Ye, Yuxia Zhang and Minghui Zhou, “Characterising Open Source Co-opetition in Company-hosted Open Source Software Projects: The Cases of PyTorch, TensorFlow, and Transformers,” author-posted mixed-method preprint (2024), abstract and §4 (respondent G on competing cloud providers' Transformers integrations). Its cases and governance findings are bounded to the studied projects. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p