Competition Law and Predictive Ecosystem Concentration Concerns

1. Introduction

Predictive ecosystem concentration refers to a situation in which an undertaking gains or strengthens market power because its ecosystem can collect, combine, analyse, and use large quantities of data to predict consumer behaviour, competitor conduct, demand, pricing, switching patterns, or future market developments.

The concept is particularly important in digital platforms, artificial intelligence, cloud computing, online advertising, fintech, e-commerce, app stores, digital identity systems, mobility platforms, healthcare technology, and data-driven business models.

Traditional competition law generally asks whether an undertaking has market power in a defined relevant market and whether its conduct restricts competition. Predictive ecosystem concentration adds another dimension:

The competitive concern may arise not merely from present market share, but from the ability of an integrated ecosystem to predict market behaviour and use those predictions across interconnected markets.

For example, a platform operating a search engine, advertising service, payment system, cloud infrastructure and consumer marketplace may possess information that allows it to predict demand and competitor strategies. If that informational advantage is combined with control over important infrastructure or distribution channels, the ecosystem may create substantial entry and expansion barriers.

Predictive ecosystem concentration is therefore not a separate statutory offence in most jurisdictions. It is an analytical framework for identifying potential applications of existing competition-law doctrines, particularly abuse of dominance, exclusionary conduct, information exchange, tying, leveraging, self-preferencing, refusal to supply, discriminatory access, and merger control.

2. Meaning of Ecosystem Concentration

An ecosystem exists where several products, services, technologies, data resources, infrastructure layers, or user relationships are interconnected.

A simplified ecosystem can be represented as:

Data → Prediction → Decision → Distribution → User Behaviour → More Data

This creates a potentially self-reinforcing cycle.

For example:

A platform collects consumer data.

Artificial intelligence analyses the data.

The platform predicts consumer demand.

The platform changes rankings, advertising, prices, or recommendations.

Consumer behaviour changes in response.

The platform collects additional behavioural data.

The improved dataset increases the quality of future predictions.

The resulting advantage may become difficult for competitors to replicate.

3. What Makes the Concentration "Predictive"?

Traditional concentration analysis often focuses on:

market shares;

turnover;

assets;

capacity;

ownership;

pricing power;

barriers to entry.

Predictive ecosystem concentration additionally considers:

quantity and quality of data;

predictive analytics capabilities;

machine-learning infrastructure;

access to real-time behavioural information;

network effects;

switching costs;

interoperability;

control over distribution;

algorithmic optimisation;

ecosystem-wide cross-use of information;

ability to anticipate competitor strategies.

Thus, an undertaking with a relatively modest share in one market could potentially possess substantial strategic advantages because it controls an important information or prediction layer connecting several markets.

4. Competition-Law Foundations

A. Article 101 TFEU

Article 101 addresses:

agreements;

decisions by associations of undertakings;

concerted practices;

information exchanges;

cartel coordination.

Predictive technologies create new possibilities for coordination.

Competitors may use:

common pricing algorithms;

common data providers;

algorithmic forecasting systems;

industry-wide predictive platforms;

shared benchmarking systems.

The competition concern is particularly significant where information exchange reduces strategic uncertainty.

B. Article 102 TFEU

Article 102 addresses abuse of a dominant position.

Predictive ecosystem concentration can potentially generate concerns involving:

self-preferencing;

tying;

bundling;

discriminatory access;

refusal to supply;

interoperability restrictions;

exclusive arrangements;

leveraging;

predatory pricing;

margin squeeze;

exploitation of data advantages.

The central question is not whether predictive analytics itself is unlawful. Rather, the question is whether a dominant undertaking uses its predictive advantage in a manner capable of restricting effective competition.

C. Indian Competition Act, 2002

In India, the principal provisions are:

Section 3

Addresses agreements having an appreciable adverse effect on competition.

Particularly relevant conduct includes:

price fixing;

market allocation;

bid rigging;

information exchange;

vertical restraints.

Section 4

Addresses abuse of dominant position.

Potentially relevant forms include:

unfair or discriminatory conditions;

unfair pricing;

limiting production or technical development;

denial of market access;

tying;

leveraging;

exclusionary conduct.

Sections 5 and 6

Merger-control provisions become relevant when acquisitions consolidate:

data;

predictive technologies;

AI capabilities;

complementary platforms;

cloud infrastructure;

advertising ecosystems;

emerging competitors.

5. Major Predictive Ecosystem Concentration Concerns

5.1 Data Concentration

Data can become an important competitive input.

A large ecosystem may obtain information from:

search queries;

transactions;

advertisements;

location;

payments;

consumer reviews;

browsing behaviour;

connected devices;

cloud services.

The competitive advantage may arise from the combination of datasets rather than from any individual dataset.

5.2 Cross-Market Data Combination

Suppose an undertaking operates:

an online marketplace;

payment services;

advertising;

logistics;

cloud infrastructure.

It may combine information from each business.

The resulting predictive capability could allow it to understand:

purchasing patterns;

supplier performance;

consumer demand;

competitor activity;

price sensitivity;

inventory conditions.

The competition concern becomes stronger if rivals cannot obtain equivalent information.

6. Predictive Advantages and Entry Barriers

Predictive ecosystem concentration may create several barriers to entry.

Data barrier

New entrants lack historical datasets.

Learning barrier

AI systems may require extensive historical information to develop accurate models.

Network barrier

More users generate more information, which improves predictions and attracts additional users.

Distribution barrier

The incumbent may control important access points.

Switching barrier

Consumers or businesses may lose accumulated data, reputation, history, or personalised services when switching.

Infrastructure barrier

The ecosystem may control cloud, computing, identity, payment, or authentication infrastructure.

Consequently, the relevant competitive advantage may be dynamic rather than static.

7. Predictive Self-Preferencing

An ecosystem may use predictive information to favour its own downstream services.

For example, a platform might predict:

which sellers are likely to grow;

which products will become popular;

which competitors are gaining users;

which customers are likely to switch.

It could then use those predictions to modify:

rankings;

recommendations;

advertising;

commissions;

access conditions;

visibility.

This raises questions similar to those considered in digital self-preferencing cases.

8. Predictive Foreclosure

A dominant ecosystem might identify emerging competitors before they become significant.

Predictive analytics could reveal:

rapidly growing startups;

unusual changes in consumer behaviour;

new technology adoption;

competitor customer acquisition;

supplier switching;

future demand patterns.

The ecosystem could potentially respond through:

acquisitions;

exclusionary pricing;

contractual restrictions;

interoperability limitations;

discriminatory ranking;

bundling.

This is particularly relevant to potential competition and nascent-competitor theories of harm.

9. Predictive Pricing and Algorithmic Coordination

Predictive systems can analyse competitors' prices and market conditions.

There are two different situations.

Independent algorithmic optimisation

Each undertaking independently uses an algorithm to optimise its prices.

This is not automatically unlawful.

Coordinated algorithmic pricing

Algorithms may facilitate communication, monitoring, implementation or enforcement of an agreement or concerted practice.

Competition law therefore distinguishes:

independent intelligent behaviour from algorithmically facilitated coordination.

10. Information Exchange

Predictive ecosystems can transform ordinary information into highly valuable strategic information.

For example:

future prices;

expected capacity;

inventory forecasts;

demand predictions;

customer switching probabilities;

production forecasts.

Exchange of sufficiently strategic information between competitors may reduce uncertainty and facilitate coordination.

The legal significance depends upon factors such as:

nature of information;

age;

level of aggregation;

frequency;

market structure;

purpose;

recipients;

degree of strategic sensitivity.

11. Case Law

11.1 T-Mobile Netherlands BV v Raad van bestuur van de Nederlandse Mededingingsautoriteit

Case: C-8/08

This case concerned an exchange of competitively sensitive information among mobile telecommunications operators.

The Court of Justice recognised that information exchange capable of reducing uncertainty concerning competitors' future conduct can constitute a concerted practice.

Relevance to predictive ecosystems

The case is important because predictive ecosystems may possess extremely detailed information about future market behaviour.

A competition-risk framework should therefore distinguish:

historical information;

aggregated information;

commercially sensitive information;

future-oriented information.

The more effectively information reduces strategic uncertainty, the greater the competition-law concern.

11.2 Dole Food Company Inc. v European Commission

Case: C-286/13 P

The case concerned exchanges of information in the banana market.

The Court examined whether communications between competitors could reduce uncertainty concerning important competitive parameters.

Predictive ecosystem significance

The case demonstrates that competition law can attach importance to information that helps competitors anticipate market conditions.

Modern ecosystems can perform similar functions through:

predictive analytics;

automated benchmarking;

demand forecasts;

algorithmic pricing systems.

Thus, predictive capacity can increase the competitive sensitivity of information.

11.3 Eturas UAB and Others

Case: C-74/14

The case involved an electronic booking platform through which a communication concerning discount limits was transmitted to participating travel agencies.

The Court considered whether undertakings could be held responsible for participating in a concerted practice where information was communicated through a common electronic system.

Relevance

Eturas is particularly relevant to platform ecosystems because coordination need not occur through traditional face-to-face meetings.

A digital platform can become a mechanism through which:

rules are communicated;

pricing parameters are implemented;

competitors receive information;

commercial behaviour becomes aligned.

Predictive ecosystem analysis should therefore examine the platform's technical architecture as well as formal contracts.

11.4 AC-Treuhand AG v European Commission

Case: C-194/14 P

The case concerned the liability of a consultancy that facilitated cartel activity despite not itself operating at the relevant market level.

The Court confirmed that an undertaking can fall within Article 101 where it intentionally contributes to the implementation of an anticompetitive arrangement.

Relevance

Predictive ecosystems may involve:

data analytics providers;

algorithm providers;

cloud providers;

benchmarking firms;

industry platforms.

The fact that an entity is not itself a direct competitor does not automatically eliminate competition-law risk where it intentionally facilitates anticompetitive coordination.

11.5 Google Shopping

Case: Google Search (Shopping), T-612/17

The European Commission's Google Shopping decision concerned the treatment of Google's comparison-shopping service within its general search results.

The General Court upheld the Commission's finding concerning Google's conduct and its effects on competition in comparison shopping, subject to the precise reasoning of the judgment.

Predictive ecosystem significance

The case is relevant to ecosystems because a platform controlling an important gateway can potentially use information and ranking mechanisms to favour its own downstream service.

Predictive ranking systems could amplify this concern where the platform can forecast:

user preferences;

conversion probabilities;

seller performance;

competitor growth.

The important legal issue remains whether the conduct constitutes an abuse and produces or is capable of producing exclusionary effects, not merely whether the platform uses predictive technology.

11.6 Microsoft Corp. v Commission

Case: T-201/04

The Microsoft case involved several forms of conduct, including interoperability restrictions and tying.

The General Court largely upheld the Commission's findings concerning Microsoft's dominant position and exclusionary conduct.

Relevance to predictive ecosystems

Predictive ecosystem concentration can coexist with control over technical interoperability.

A powerful ecosystem may combine:

data + prediction + infrastructure + interoperability control.

Where competitors cannot effectively interact with the dominant ecosystem, predictive advantages may become significantly more difficult to overcome.

11.7 Intel Corp. v Commission

Case: C-413/14 P

The Intel litigation concerned rebates offered by a dominant undertaking and the assessment of their potential exclusionary effects.

The Court required closer examination of the economic circumstances where the undertaking submits evidence that the conduct is not capable of producing the alleged foreclosure effects.

Predictive ecosystem relevance

Predictive ecosystems may employ highly sophisticated commercial incentives.

For example:

personalised discounts;

targeted rebates;

customer-specific offers;

predictive retention incentives.

Competition analysis therefore should not stop at identifying the existence of a discount. It may require assessment of how the mechanism affects competitors' ability to compete.

11.8 AKZO Chemie BV v Commission

Case: C-62/86

AKZO is a foundational predatory-pricing case.

The Court developed important principles concerning pricing below relevant cost benchmarks by a dominant undertaking.

Predictive ecosystem relevance

A dominant ecosystem may use predictive analytics to identify:

customers vulnerable to switching;

geographic markets with new entrants;

products requiring aggressive pricing;

competitors' financial weaknesses.

Predictive targeting could therefore make exclusionary pricing more sophisticated.

The technology itself does not make the pricing unlawful; the legal analysis continues to depend upon dominance, pricing conditions, competitive effects and the applicable legal test.

11.9 France Télécom SA v Commission

Case: C-202/07 P

The case concerned predatory pricing by Wanadoo, a subsidiary of France Télécom.

The Court upheld the relevant EU-law approach to predatory pricing and rejected the proposition that proof of actual recoupment must necessarily be established as an independent requirement.

Relevance

Predictive systems can make below-cost pricing highly targeted.

A platform may predict:

where entry is occurring;

which customers matter most to rivals;

which geographic areas are strategically important.

The case illustrates why sophisticated pricing analytics can become relevant to exclusionary-pricing analysis.

11.10 Bronner v Mediaprint

Case: C-7/97

The Court established a strict approach to refusal-to-supply claims involving access to an infrastructure controlled by a dominant undertaking.

Predictive ecosystem relevance

Consider a dominant digital ecosystem controlling:

identity infrastructure;

authentication;

cloud interfaces;

data access;

payment infrastructure;

application programming interfaces.

If competitors depend upon such infrastructure, predictive ecosystem concentration may overlap with traditional essential-facility and refusal-to-deal questions.

Bronner demonstrates that dominance alone does not automatically create a general duty to provide access.

12. Predictive Mergers and Ecosystem Expansion

Predictive concentration is particularly significant in merger control.

A conventional merger assessment might consider:

"What market share will the merged undertaking possess?"

A predictive ecosystem assessment asks additional questions:

What datasets will be combined?

Will the transaction improve predictive accuracy?

Will rivals lose access to important behavioural information?

Will the acquisition remove an emerging competitor?

Will the merged firm control complementary technologies?

Will AI capabilities become vertically integrated?

Will the transaction strengthen network effects?

Will interoperability decrease?

Will the ecosystem become more difficult to challenge?

This is closely connected to modern concerns about nascent competition and innovation competition.

13. Killer Acquisitions and Predictive Information

An ecosystem may acquire a small company whose current market share is insignificant.

Traditional market-share analysis may therefore underestimate the importance of the transaction.

However, the target may possess:

unique datasets;

superior predictive models;

innovative AI technology;

specialised user communities;

emerging distribution channels.

The competitive significance of the target may lie in its future trajectory, rather than its present turnover.

14. Ecosystem Lock-In

Predictive ecosystem concentration can generate strong switching costs.

Users may accumulate:

transaction histories;

recommendations;

preferences;

digital identities;

reputational information;

personalised models;

business relationships.

Leaving the ecosystem may mean losing these accumulated benefits.

Consequently, competition may become increasingly difficult even when nominal switching is technically possible.

15. Network Effects and Predictive Feedback Loops

Predictive ecosystems may exhibit a feedback mechanism:

More users → More data → Better predictions → Better service → More users

This can produce indirect network effects.

Once the feedback loop becomes sufficiently strong, an incumbent may enjoy a cumulative advantage.

Competition law should therefore examine whether:

rivals can obtain comparable data;

consumers can switch;

data can be ported;

APIs are interoperable;

competitors can achieve sufficient scale;

the incumbent can leverage one market into another.

16. Predictive Ecosystem Concentration and Self-Preferencing

A dominant ecosystem may use information generated by independent businesses operating on its platform.

For example, a marketplace could observe:

sales volumes;

consumer preferences;

product margins;

inventory;

conversion rates.

It may then use this information to improve its own competing products.

The competitive concern becomes stronger where the platform:

operates the infrastructure;

receives commercially sensitive information from participants;

competes downstream;

uses the information to advantage its own downstream business.

This creates a potential conflict between the ecosystem's role as infrastructure provider and competitor.

17. Predictive Ecosystem Concentration and Data Portability

Data portability can affect competition by reducing switching costs.

Competition authorities may therefore examine:

portability rights;

technical interoperability;

API access;

real-time transfer;

machine-readable formats;

authentication portability.

However, portability alone may not eliminate ecosystem concentration if the incumbent retains:

superior data volume;

superior algorithms;

network effects;

distribution advantages;

proprietary infrastructure.

18. Predictive Ecosystem Concentration and Interoperability

Interoperability can be particularly important where several markets are technologically connected.

Potential concerns include:

API restrictions;

compatibility limitations;

discriminatory technical access;

proprietary standards;

authentication barriers;

data-format restrictions.

A dominant ecosystem may theoretically use interoperability control to protect predictive advantages.

The legal analysis must distinguish legitimate security or technical reasons from exclusionary conduct.

19. Risk Indicators

A competition-law compliance programme can monitor the following indicators:

These are risk indicators, not automatic findings of infringement.

20. A Predictive Ecosystem Competition-Risk Framework

A useful analytical model can contain six stages.

Stage 1 — Identify the ecosystem

Map:

platforms;

suppliers;

customers;

infrastructure;

data;

algorithms;

complementary services.

Stage 2 — Identify concentration points

Determine where control is concentrated over:

data;

users;

infrastructure;

standards;

algorithms;

distribution.

Stage 3 — Identify predictive capabilities

Ask what the undertaking can predict concerning:

demand;

prices;

customer switching;

competitor entry;

supplier behaviour.

Stage 4 — Identify possible leveraging

Examine whether predictive information is used across markets.

Stage 5 — Test competitive effects

Consider:

foreclosure;

exclusion;

coordination;

reduced innovation;

higher barriers to entry;

reduced consumer choice.

Stage 6 — Examine procompetitive explanations

Consider whether the conduct creates:

efficiencies;

improved security;

better product quality;

reduced transaction costs;

innovation;

lower prices.

This final stage is essential because predictive integration can generate substantial legitimate efficiencies.

21. False Positives and the Need for Effects Analysis

Predictive capability should not itself be equated with anticompetitive conduct.

For example:

an AI pricing system may independently optimise prices;

a platform may combine data to improve cybersecurity;

interoperability restrictions may be justified by genuine security risks;

personalised recommendations may improve consumer welfare;

vertical integration may generate efficiencies.

Competition law must therefore distinguish:

concentration of predictive capability

from

abusive or coordinated use of predictive capability.

This distinction is central to avoiding over-enforcement.

22. Indian Competition-Law Application

For India, predictive ecosystem concentration can be examined through the Competition Act, 2002.

Section 3

Potential issues include:

algorithmic coordination;

information exchange;

platform-mediated collusion;

common pricing systems.

Section 4

Potential issues include:

denial of market access;

discriminatory access;

leveraging;

tying;

self-preferencing;

exclusionary pricing;

refusal to deal.

Sections 5 and 6

Merger analysis can consider whether a transaction significantly strengthens:

data concentration;

network effects;

ecosystem integration;

AI capabilities;

innovation barriers;

control over important digital infrastructure.

The Competition Commission of India has increasingly examined digital-market issues involving platforms, data, intermediation and market power, making predictive ecosystem analysis potentially relevant to future enforcement and merger assessment.

23. Relationship With Other Competition-Law Doctrines

Predictive ecosystem concentration overlaps with several established doctrines.

24. Key Legal Principles Emerging From the Case Law

The cases collectively indicate several important principles.

First

Information can itself become competitively significant when it reduces uncertainty about future conduct.

Second

Digital mechanisms do not remove traditional Article 101 principles.

Third

Dominant undertakings cannot necessarily rely upon technological sophistication as a defence to exclusionary conduct.

Fourth

Predictive pricing does not automatically constitute predatory pricing.

Fifth

Control over infrastructure can become relevant where competitors depend upon access.

Sixth

Vertical integration can create concerns where an ecosystem uses its infrastructure or informational advantage to disadvantage downstream competitors.

Seventh

Merger analysis increasingly needs to consider dynamic competition, innovation and potential competitors rather than relying exclusively on present market shares.

25. Overall Analytical Model

The competition implications can be expressed as:

Data Concentration
↓
Predictive Advantage
↓
Network Effects
↓
Ecosystem Expansion
↓
Cross-Market Leveraging
↓
Higher Entry and Switching Barriers
↓
Potential Entrenchment of Market Power

But the final step is not automatic.

The legal assessment must determine whether the resulting conduct satisfies the requirements of the applicable competition-law provision.

26. Conclusion

Predictive ecosystem concentration represents an important development in modern competition-law analysis because market power can increasingly arise from the combination of data, artificial intelligence, prediction, infrastructure, network effects and ecosystem control.

The principal competition concern is not simply that a firm possesses sophisticated predictive technology. Rather, the concern arises when predictive capabilities are combined with substantial market power and are used to:

exclude competitors;

coordinate market behaviour;

disadvantage dependent businesses;

leverage power between markets;

reinforce entry barriers;

restrict interoperability;

acquire emerging competitive threats;

or entrench ecosystem control.

The case law of T-Mobile Netherlands, Dole Food, Eturas, AC-Treuhand, Google Shopping, Microsoft, Intel, AKZO, France Télécom and Bronner demonstrates that existing competition-law doctrines already provide many of the tools needed to analyse these problems.

The principal challenge for competition authorities is therefore to adapt established doctrines to an economy in which competitive advantage may depend increasingly on the ability to predict markets rather than merely participate in them.

At the same time, predictive analytics can produce genuine efficiencies and innovation. Accordingly, a sound competition-law framework should distinguish legitimate predictive innovation from exclusionary or coordinative use of predictive ecosystem power.

Case-Law Summary