Competition Law and Data Infrastructure Monopolisation Risks

1. Introduction

Data infrastructure refers to the physical, technical and organisational infrastructure through which data is generated, stored, processed, transmitted and made available to other businesses. It includes:

data centres;

cloud-computing infrastructure;

storage and computing capacity;

application programming interfaces (APIs);

data-hosting and database systems;

network and interconnection infrastructure;

identity and authentication infrastructure;

data marketplaces and exchanges;

interoperability systems;

AI-training infrastructure;

data-management platforms; and

critical datasets that are difficult for competitors to reproduce.

The competition-law problem arises when a firm controls infrastructure that competitors cannot realistically reproduce, bypass or replace, and uses that control to exclude rivals from downstream markets.

This is particularly significant in cloud computing. The European Commission has recently identified interoperability barriers, restricted or conditional access to business-user data, tying/bundling and problematic contractual terms as potential competition concerns in cloud markets. In June 2026, it also preliminarily concluded that AWS and Microsoft Azure may qualify as important gateways despite not satisfying the ordinary quantitative DMA thresholds. (Digital Markets Act (DMA))

The central competition-law question is therefore:

When does ownership or control of data infrastructure cease to be legitimate competitive advantage and become a means of monopolisation or exclusion?

2. Meaning of Data Infrastructure Monopolisation

Data infrastructure monopolisation occurs where an undertaking acquires or maintains substantial market power through control over infrastructure that competitors need in order to compete.

It can occur through several mechanisms:

A. Physical infrastructure control

A firm may control:

data centres;

fibre networks;

submarine cables;

server capacity;

specialised computing facilities;

interconnection points.

Competitors may face enormous capital costs in replicating such infrastructure.

B. Cloud infrastructure control

Large cloud providers may control:

compute capacity;

storage;

databases;

networking;

AI accelerators;

cloud operating environments;

proprietary APIs.

This can create substantial switching and interoperability barriers.

C. Data-access control

A dominant undertaking may possess datasets unavailable to competitors.

Examples include:

transaction histories;

search data;

consumer behaviour;

location data;

business intelligence;

financial data;

AI-training datasets.

D. Technical infrastructure control

Monopolisation can also occur through control of:

APIs;

authentication systems;

operating-system interfaces;

interoperability protocols;

proprietary standards.

E. Ecosystem infrastructure

A company may combine infrastructure with adjacent services so that customers become increasingly dependent upon its ecosystem.

This is especially important where cloud, AI, operating systems, databases and enterprise software are vertically integrated.

3. Why Data Infrastructure Creates Competition Problems

Data infrastructure possesses several characteristics that can facilitate market concentration.

3.1 High sunk costs

Building large-scale data centres and cloud infrastructure requires enormous capital expenditure.

A new entrant may need to invest in:

servers;

land;

electricity;

cooling;

network connectivity;

cybersecurity;

backup systems;

specialised processors.

Therefore, an incumbent can obtain a significant structural advantage.

3.2 Economies of scale

Large infrastructure providers can spread fixed costs over millions of customers.

For example:

More customers → greater infrastructure utilisation → lower average cost → greater ability to compete → more customers.

This can produce a self-reinforcing competitive advantage.

3.3 Network effects

Data infrastructure frequently exhibits direct and indirect network effects.

A cloud platform with a large ecosystem of:

developers;

software vendors;

enterprise customers;

AI providers;

cybersecurity companies;

becomes increasingly attractive.

Competitors therefore face a chicken-and-egg problem: they need customers to attract developers, but need developers to attract customers.

3.4 Switching costs

Switching from one infrastructure provider to another may require:

migrating databases;

rewriting applications;

retraining employees;

changing APIs;

transferring large quantities of data;

renegotiating contracts;

changing cybersecurity systems.

Consequently, even if another supplier offers lower prices, customers may remain with the incumbent.

3.5 Data portability problems

Data may technically belong to the customer but still be difficult to extract in a commercially usable form.

This distinction is important:

Legal ownership of data does not necessarily mean effective competitive access to data.

A provider can therefore create competitive lock-in through technical rather than formal legal ownership.

4. Relevant Competition-Law Theories

Several established doctrines can apply.

4.1 Abuse of dominance

Under Article 102 TFEU, Section 2 of the Sherman Act, and Section 4 of India's Competition Act, 2002, dominance or monopoly power is not itself unlawful.

The problem arises from exclusionary conduct.

Examples include:

refusal to supply;

discriminatory access;

tying;

bundling;

self-preferencing;

interoperability restrictions;

exclusionary rebates;

discriminatory pricing;

contractual restrictions;

foreclosure of competing infrastructure.

5. Essential Facilities Doctrine

The most obvious competition-law theory is the essential facilities doctrine.

The basic proposition is that a dominant undertaking controlling an indispensable facility may, in exceptional circumstances, have a duty to provide access to competitors.

Traditional requirements generally involve:

control of the facility by a dominant undertaking;

practical or legal inability of competitors to reasonably duplicate it;

denial of access;

elimination or substantial restriction of competition;

absence of an objective justification.

However, courts have traditionally applied the doctrine cautiously.

The doctrine is particularly controversial when applied to data, because data is different from a physical bridge or railway:

data can often be reproduced;

data may be subject to privacy rights;

datasets can become obsolete;

data collection can sometimes be duplicated;

mandatory sharing can reduce incentives to invest.

Consequently, not every commercially valuable dataset qualifies as an essential facility.

6. Case Law

Case 1 — United States v Terminal Railroad Association of St. Louis (1912)

This is one of the foundational American essential-facilities cases.

The defendants controlled terminal facilities necessary for competing railroads to enter St. Louis.

The Supreme Court objected to the collective control of essential infrastructure that prevented competitors from effectively accessing the market.

Relevance to data infrastructure

The case establishes an important principle:

Control over a strategically indispensable infrastructure bottleneck can become an antitrust problem when the bottleneck is used to exclude competitors.

Modern equivalents could include:

critical data centres;

network interconnection facilities;

cloud infrastructure;

specialised computing infrastructure.

The analogy is strongest where competitors genuinely cannot economically reproduce the infrastructure.

7. Case 2 — MCI Communications Corp. v AT&T (7th Cir., 1983)

This is one of the leading American essential-facilities decisions.

AT&T controlled telecommunications infrastructure that MCI needed to compete.

The court formulated a frequently cited framework involving:

control of the essential facility by a monopolist;

competitor's inability reasonably to duplicate it;

denial of access; and

feasibility of providing access.

Importance for data infrastructure

The case provides a useful conceptual model for infrastructure monopolisation.

For example, suppose a cloud provider controls a technical interface necessary for interoperability with a dominant ecosystem.

The analysis could ask:

First: Does the cloud provider possess monopoly power?

Second: Is the interface practically indispensable?

Third: Can competitors reproduce or bypass it?

Fourth: Has access been denied?

Fifth: Is providing access technically and economically feasible?

The case therefore remains highly relevant to cloud and network infrastructure.

8. Case 3 — Magill — Joined Cases C-241/91 P and C-242/91 P

The European Court of Justice considered whether refusal to license copyright-protected information could constitute abuse of dominance.

The Court established exceptional circumstances under which refusal to provide access to information could constitute an abuse.

The important factors included:

indispensability;

elimination of competition;

prevention of a new product for which consumer demand existed;

lack of justification.

Relevance to data monopolisation

Magill is particularly important because it demonstrates that information itself can become relevant to an exclusionary-abuse analysis.

A dominant infrastructure operator could therefore face competition concerns if it controls uniquely valuable information and prevents downstream competitors from developing competing products.

But the doctrine remains exceptional.

9. Case 4 — Oscar Bronner GmbH & Co. KG v Mediaprint

Case C-7/97

This is one of the most important European cases for essential facilities.

Bronner sought access to Mediaprint's newspaper distribution system.

The Court applied a strict test.

The facility had to be indispensable, meaning that there was no actual or potential substitute and that duplication was not economically viable.

Key principle

The mere fact that an alternative facility is less advantageous does not make an existing facility indispensable.

Application to data infrastructure

This is crucial.

A company cannot simply argue:

"AWS is the best cloud infrastructure, therefore AWS must provide access to its infrastructure."

Instead, the competition authority must examine whether:

alternative cloud providers exist;

private infrastructure is technically feasible;

multi-cloud arrangements are possible;

migration is realistically achievable;

interoperability can be achieved;

duplication is economically feasible.

Thus, commercial importance ≠ essential facility.

10. Case 5 — IMS Health GmbH & Co. OHG v NDC Health GmbH

Case C-418/01

The Court again addressed refusal to license an intellectual-property-based infrastructure.

IMS Health controlled a pharmaceutical data structure that competitors needed.

The Court reaffirmed the exceptional nature of compulsory access.

Three particularly important conditions emerged:

access must be indispensable;

refusal must prevent the emergence of a new product or service for which consumer demand exists;

refusal must lack objective justification.

Data infrastructure significance

This is highly relevant to:

proprietary databases;

data architectures;

industry-standard datasets;

APIs;

data formats;

interoperable technical systems.

The case demonstrates that a dominant firm does not automatically have to share its intellectual property or infrastructure merely because competitors would benefit from access.

11. Case 6 — Microsoft Corp. v Commission

Case T-201/04

This is arguably one of the most significant European competition cases for technology infrastructure.

The European Commission found that Microsoft had abused its dominant position through refusal to supply interoperability information and through tying conduct.

The General Court largely upheld the Commission's reasoning.

Why it matters

Microsoft's control over operating-system information created an important interoperability advantage.

Competitors required information to ensure their products could work effectively with Microsoft's dominant operating system.

Data infrastructure lesson

Modern data infrastructure can produce a similar problem.

Suppose a dominant cloud provider:

controls the dominant cloud environment;

possesses critical technical information;

restricts interoperability;

prevents competing services from functioning effectively.

Competition authorities may regard this as an exclusionary strategy rather than legitimate product differentiation.

12. Case 7 — Slovak Telekom v Commission

Joined Cases C-152/19 P and C-165/19 P

The European Court of Justice considered access to telecommunications infrastructure controlled by a dominant undertaking.

The case is important because it concerns infrastructure-based competition and exclusionary access conditions.

The Court confirmed that refusal or restriction of access to infrastructure can constitute abuse where the circumstances satisfy the strict conditions applicable to exclusionary conduct.

Data-centre relevance

The analogy is particularly strong for:

fibre infrastructure;

data-centre interconnection;

cloud connectivity;

network access;

physical server infrastructure.

It demonstrates that competition authorities may look beyond prices and ask whether infrastructure control prevents effective downstream competition.

13. Case 8 — Google Shopping

Case T-612/17

Although this case did not concern physical data centres, it is important for data-driven infrastructure and self-preferencing.

The European Commission found that Google systematically favoured its own comparison-shopping service in search results.

The General Court upheld the core finding of abuse.

Relevance

A vertically integrated data-infrastructure provider might similarly use control over infrastructure to favour its downstream services.

For example:

Cloud infrastructure → AI computing → AI model → AI application

If the infrastructure owner systematically provides its downstream business with:

superior access;

better technical interfaces;

preferential data;

lower infrastructure costs;

better latency;

privileged APIs;

while disadvantaging rivals, competition concerns may arise.

The key issue is therefore not merely ownership of infrastructure, but how infrastructure control is exercised.

14. Case 9 — Napp Pharmaceutical Holdings Ltd v Director General of Fair Trading

[2002] CAT 1

The UK Competition Appeal Tribunal considered abuse of dominance involving pricing and exclusionary conduct.

Although not a data-infrastructure case, Napp is useful for understanding how dominant firms can employ commercial conditions to disadvantage competitors.

Data infrastructure application

A dominant cloud provider could potentially create foreclosure through:

discriminatory discounts;

minimum-spend requirements;

preferential pricing;

loyalty arrangements;

bundled infrastructure services.

The competition analysis would focus on whether the commercial arrangement forecloses equally efficient competitors.

15. Case 10 — FTC v Qualcomm

969 F.3d 974 (9th Cir. 2020)

This case concerned technology licensing and the relationship between intellectual property, technological standards and competition.

The Ninth Circuit ultimately rejected the FTC's theory on the particular facts.

Importance

The case illustrates an important limitation:

Not every technologically important asset controlled by a powerful company creates an antitrust duty to deal.

Competition law must distinguish between:

legitimate monetisation of technology; and

conduct specifically designed to exclude competition.

This distinction is critical for cloud infrastructure and proprietary data systems.

16. Indian Competition-Law Perspective

The Indian framework is particularly interesting because Section 4 of the Competition Act, 2002 prohibits abuse of dominant position.

Relevant provisions include:

Section 4(2)(a)

Prohibits unfair or discriminatory:

conditions; or

prices.

Section 4(2)(c)

Prohibits denial of market access.

This provision is especially relevant to data infrastructure.

If a dominant infrastructure provider prevents competitors from obtaining access necessary to compete, the conduct may potentially fall within the concept of denial of market access.

Section 4(2)(d)

Concerns supplementary obligations unrelated to the subject matter of the contract.

This can become relevant where cloud infrastructure is bundled with unrelated services.

Section 4(2)(e)

Addresses leveraging dominance in one relevant market to enter or protect another market.

This is particularly significant for vertically integrated cloud companies.

17. The MCX Stock Exchange Ltd. v NSE Principle

The Indian competition-law jurisprudence concerning network infrastructure is particularly instructive.

The Competition Commission of India found concerns regarding the use of dominance and pricing strategy by NSE in relation to the currency derivatives market.

The case demonstrates that infrastructure and network advantages can materially affect competition in adjacent markets.

The broader principle is applicable to digital infrastructure:

An undertaking that possesses a substantial infrastructural or network advantage cannot necessarily use that advantage to eliminate competition in a neighbouring market.

This is relevant to cloud ecosystems where infrastructure can be leveraged into:

AI;

cybersecurity;

enterprise software;

database services;

financial technology;

analytics.

18. Data Centres as Potential Bottleneck Infrastructure

A modern data centre can become competitively significant because it combines:

physical space;

electricity;

cooling;

networking;

connectivity;

computing capacity;

storage;

cybersecurity.

If a firm controls a critical cluster of data-centre capacity, competitors may encounter substantial barriers to entry.

However, competition authorities should distinguish between:

Legitimate scale

A company simply has more efficient infrastructure.

and

Exclusionary infrastructure control

The company deliberately:

reserves scarce capacity;

prevents competitors from obtaining connectivity;

imposes discriminatory access conditions;

ties infrastructure to unrelated services;

acquires strategically important facilities solely to exclude rivals.

The second category presents much greater competition-law risk.

19. Cloud Computing and Infrastructure Monopolisation

Cloud markets are particularly susceptible to concentration.

A simplified ecosystem looks like:

Data centre → Cloud infrastructure → Platform services → AI/database tools → Applications

A firm controlling several layers may gain considerable vertical power.

The EU's current cloud investigation illustrates this concern. The Commission's investigation specifically considers interoperability, data access, tying/bundling and contractual conditions. (Digital Markets Act (DMA))

As of June 2026, the Commission had preliminarily taken the view that AWS and Azure could constitute important gateways under the DMA despite not meeting the ordinary quantitative designation thresholds. (Digital Markets Act (DMA))

20. Cloud Lock-In as a Competition Concern

One of the most important risks is cloud lock-in.

Consider:

Customer → AWS → proprietary API → proprietary database → proprietary AI tools

After several years, moving to another provider may become extremely expensive.

The customer may therefore become economically dependent on the infrastructure provider.

Lock-in may result from:

data-transfer charges;

proprietary APIs;

incompatible architectures;

proprietary databases;

contractual restrictions;

technical migration barriers;

preferential treatment of internal services.

Competition-law significance

High switching costs can create durable market power even where nominally competing providers exist.

The relevant question is therefore not merely:

"Are there three cloud providers?"

but:

"Can customers realistically switch between them?"

21. Interoperability as a Competition Remedy

Interoperability can be more effective than compulsory ownership transfer.

Possible remedies include:

API access

Competitors receive access to necessary interfaces.

Data portability

Customers can export data in usable formats.

Interoperability standards

Infrastructure providers must permit systems to communicate.

Technical documentation

Competitors receive sufficient information to develop interoperable services.

Non-discriminatory access

Infrastructure must be supplied on equivalent conditions.

This approach resembles the logic underlying the Microsoft interoperability case.

22. Self-Preferencing Risks

A vertically integrated infrastructure provider may compete against its own customers.

For example:

Cloud provider

→ supplies computing infrastructure to AI companies

→ develops its own AI model

→ develops its own AI application

→ gives its own AI products preferential access to infrastructure.

Potential problems arise if the provider gives itself:

priority computing capacity;

cheaper access;

superior APIs;

privileged datasets;

faster processing;

better technical support.

This can transform infrastructure dominance into downstream foreclosure.

23. Tying and Bundling

A dominant provider could require:

"To obtain cloud storage, you must also purchase our cybersecurity service."

or:

"To receive access to our AI infrastructure, you must use our database."

Such arrangements can raise concerns under abuse-of-dominance rules where:

the undertaking is dominant in the tying market;

the products are distinct;

customers are effectively forced to obtain the tied product;

the practice forecloses competition.

The concern becomes stronger where infrastructure itself is indispensable.

24. Data Accumulation and Infrastructure Power

Data infrastructure produces another competitive advantage: data feedback loops.

For example:

More customers

↓

More data

↓

Better algorithms

↓

Better services

↓

More customers

↓

More data

This can create a self-reinforcing competitive cycle.

The incumbent may therefore possess both:

infrastructure power; and

informational power.

This combination can be considerably stronger than either advantage individually.

25. Data as an Essential Facility — A Caution

Competition authorities should not automatically classify data as an essential facility.

Several questions should be asked.

Question 1 — Is the data genuinely indispensable?

Could competitors obtain equivalent data elsewhere?

Question 2 — Is duplication feasible?

Can competitors collect similar data themselves?

Question 3 — Is the dataset sufficiently unique?

Ordinary consumer information may be replicable.

Question 4 — Does refusal eliminate competition?

The effect must be more than merely making competition harder.

Question 5 — Is sharing technically feasible?

The provider may face legitimate security or architecture constraints.

Question 6 — Are privacy obligations implicated?

Personal data cannot simply be disclosed because competition law favours access.

Question 7 — Would compulsory access undermine investment?

Mandatory sharing could reduce incentives to develop innovative infrastructure.

These considerations explain why Bronner, Magill and IMS Health remain important.

26. Privacy and Competition Law

Data-infrastructure monopolisation cannot be analysed purely through antitrust law.

Three regimes may intersect:

Competition law

→ access and foreclosure

Data-protection law

→ lawful processing and privacy

Cybersecurity law

→ security and resilience

A competition remedy requiring data sharing may therefore need:

anonymisation;

aggregation;

purpose limitation;

access controls;

security safeguards;

audit mechanisms.

Consequently, the remedy must not simply say:

"Share all data with competitors."

It should identify what data, why, how, with whom, and under what safeguards.

27. Potential Monopolisation Strategies

A dominant data-infrastructure provider could potentially engage in:

28. Merger-Control Risks

Infrastructure monopolisation can also arise through acquisitions.

A dominant cloud provider acquiring:

a major data-centre operator;

an AI infrastructure provider;

a database company;

a cybersecurity provider;

a data marketplace;

could increase vertical or conglomerate market power.

The competition authority should examine:

Horizontal effects

Does the transaction eliminate a competitor?

Vertical effects

Does the merged entity control an essential input?

Conglomerate effects

Can infrastructure dominance be leveraged into another market?

Data effects

Does the transaction combine datasets that competitors cannot reproduce?

Innovation effects

Does the acquisition eliminate an emerging technological threat?

29. Appropriate Competition-Law Remedies

Remedies should be proportionate.

Structural remedies

In extreme cases:

divestiture;

separation of infrastructure and downstream businesses;

compulsory licensing;

separation of business units.

Behavioural remedies

More commonly:

non-discriminatory access;

interoperability;

data portability;

API access;

prohibition of self-preferencing;

prohibition of tying;

transparent pricing;

restrictions on exclusivity.

Technical remedies

Increasingly important:

standardised APIs;

interoperable data formats;

cloud migration tools;

portability mechanisms;

independent technical audits.

30. The Most Important Legal Test

A useful analytical framework for examining data-infrastructure monopolisation is:

Step 1 — Define the relevant market

Possible markets include:

cloud infrastructure services;

data-centre capacity;

cloud storage;

cloud computing;

AI computing;

database services;

data-access services.

Step 2 — Determine dominance

Consider:

market share;

infrastructure capacity;

financial strength;

switching costs;

entry barriers;

network effects;

ecosystem power.

Step 3 — Identify the infrastructure bottleneck

Ask:

What infrastructure does the dominant firm control that competitors cannot reasonably reproduce?

Step 4 — Examine conduct

Identify:

refusal;

discrimination;

tying;

bundling;

self-preferencing;

exclusionary pricing;

interoperability restrictions.

Step 5 — Assess foreclosure

Does the conduct:

exclude rivals?

raise their costs?

prevent entry?

restrict innovation?

reduce customer choice?

Step 6 — Examine objective justification

The undertaking may invoke:

security;

privacy;

technical limitations;

intellectual-property rights;

efficiency;

investment incentives.

Step 7 — Assess proportionality

The authority should determine whether a less restrictive alternative could achieve the same legitimate objective.

31. Six Core Case Laws to Remember

For examination purposes, the following cases provide a strong doctrinal foundation:

32. Overall Legal Position

The most important principle is:

Competition law does not prohibit a company from becoming dominant through superior data infrastructure; it prohibits the strategic use of that dominance to exclude competition where the conduct lacks sufficient competitive justification.

The strongest monopolisation risks arise where three conditions converge:

Infrastructure indispensability

Durable market power

Exclusionary conduct

=

Significant competition-law risk

This is particularly important for cloud computing because the infrastructure layer can become the foundation for downstream AI, software, analytics and data markets. The EU's ongoing cloud investigations demonstrate that regulators are increasingly examining interoperability, switching barriers, data access and contractual restrictions rather than looking solely at traditional market-share measures. (Digital Markets Act (DMA))

At the same time, the essential-facilities cases—especially Bronner, Magill, IMS Health and MCI v AT&T—show that mandatory access remains an exceptional remedy, not a general rule requiring successful infrastructure companies to share every commercially valuable asset.

Conclusion

The future competition-law debate over data infrastructure will therefore move beyond the traditional question of "Who owns the data?" toward a broader question:

Who controls the infrastructure through which competitors must obtain, process, transport, store or exploit data—and can that control be used to prevent effective competition?