Competition Law and Data Infrastructure Monopolisation Concerns

1. Introduction

Data infrastructure refers to the physical and technological infrastructure through which data is generated, stored, processed, transmitted and accessed. It includes:

data centres;

cloud-computing infrastructure;

server capacity and storage;

internet backbone and connectivity;

content-delivery networks;

data-processing facilities;

application programming interfaces (APIs);

identity and authentication infrastructure;

database and storage systems;

AI-computing infrastructure and GPU clusters;

hyperscale cloud platforms;

network infrastructure and interconnection facilities.

The competition-law concern arises when control over such infrastructure becomes sufficiently concentrated that a firm can exclude competitors, increase switching costs, restrict access, discriminate against downstream businesses, leverage infrastructure dominance into adjacent markets, or make rivals dependent upon its infrastructure.

This issue is becoming particularly important in cloud computing and AI. In June 2026, the European Commission stated that Amazon Web Services (AWS) and Microsoft Azure appeared to have entrenched positions in EU cloud services, with lock-in effects, high switching costs and extensive ecosystems being relevant concerns. (Digital Markets Act (DMA)) The UK's Competition and Markets Authority has similarly concluded that competition concerns exist in public cloud infrastructure and recommended attention to the two largest providers, Microsoft and AWS. (GOV.UK)

The central legal question is therefore not simply:

“Is a company very large?”

but rather:

“Is the company using control over essential or strategically important data infrastructure to obtain or maintain market power through conduct that harms the competitive process?”

2. Why Data Infrastructure Can Produce Monopoly Power

Data infrastructure possesses several characteristics that make monopolisation particularly significant.

A. Extremely high fixed costs

Building hyperscale data centres requires enormous expenditure on:

land;

electricity;

cooling;

servers;

networking equipment;

fibre connectivity;

security;

backup systems;

specialised hardware.

These costs can create substantial economies of scale.

A large infrastructure operator can therefore have a lower average cost than a smaller entrant.

B. Economies of scope

A hyperscaler can use the same infrastructure to provide:

cloud storage;

computing;

databases;

AI services;

cybersecurity;

analytics;

machine learning;

content delivery;

enterprise software.

Consequently, infrastructure dominance in one market may reinforce power in another.

C. Switching costs

Businesses may invest millions in:

cloud migration;

proprietary APIs;

data architecture;

security systems;

software integration;

employee training.

Once these investments are made, moving to another infrastructure provider can become expensive.

D. Network and ecosystem effects

A dominant infrastructure provider may have:

infrastructure → developers → applications → customers → more data → greater infrastructure utilisation → lower costs → more customers.

This creates a self-reinforcing ecosystem.

E. Data accumulation

Infrastructure providers may obtain extensive information concerning:

customer behaviour;

workloads;

applications;

traffic;

performance;

enterprise demand;

technological development.

This information can potentially provide a competitive advantage in downstream markets.

The FTC's examination of major cloud/AI partnerships has specifically identified concerns about access to computing resources, switching costs and access to commercially sensitive information. (Federal Trade Commission)

3. Relevant Competition-Law Framework

A. Abuse of dominance

Under Article 102 TFEU, a dominant undertaking may not abuse its position.

The analysis generally involves:

defining the relevant market;

establishing dominance;

identifying exclusionary or exploitative conduct;

assessing competitive effects;

considering objective justification and efficiencies.

The European Commission expressly identifies market definition and assessment of dominance as the initial steps in an Article 102 investigation. (Competition Policy)

B. US monopolisation law

Section 2 of the Sherman Act prohibits monopolisation, attempted monopolisation and conspiracy to monopolise.

The traditional analysis asks:

whether the undertaking possesses monopoly power; and

whether that power has been acquired or maintained through exclusionary conduct rather than superior competitive performance.

The distinction is critical because:

A monopoly obtained through innovation is not automatically unlawful.

The problem arises where the monopolist uses exclusionary mechanisms to prevent rivals from competing.

C. Indian competition law

In India, the principal provision is Section 4 of the Competition Act 2002, concerning abuse of dominant position.

Potentially relevant conduct includes:

unfair or discriminatory conditions;

denial of market access;

limiting technical development;

leveraging dominance from one market into another;

tying and bundling;

discriminatory access;

exclusionary pricing.

For data infrastructure, Section 4 becomes particularly relevant where a dominant cloud, data-processing, platform or network infrastructure provider uses its position to disadvantage downstream competitors.

4. What Constitutes a “Data Infrastructure Monopoly”?

A data infrastructure monopoly can exist at several levels.

Level 1 — Physical infrastructure

For example:

data centres;

fibre networks;

submarine cables;

electricity-linked infrastructure;

specialised computing clusters.

Level 2 — Cloud infrastructure

For example:

Infrastructure-as-a-Service;

Platform-as-a-Service;

cloud databases;

storage;

virtual machines.

Level 3 — Technical infrastructure

For example:

APIs;

operating systems;

authentication;

developer tools;

interoperability layers.

Level 4 — Data infrastructure

For example:

large-scale datasets;

data repositories;

data aggregation systems;

real-time data feeds.

Level 5 — AI infrastructure

For example:

GPU clusters;

AI training infrastructure;

inference capacity;

specialised accelerators;

AI-cloud platforms.

The competition concern increases when one firm controls multiple layers simultaneously.

5. Six Major Case Laws

Case 1 — United States v. Terminal Railroad Association, 224 U.S. 383 (1912)

Principle

This is one of the foundational cases concerning control over infrastructure.

The Terminal Railroad Association controlled essential railroad terminal facilities in St. Louis. Competitors effectively needed access to the terminal system to compete.

The Supreme Court treated the infrastructure arrangement as an unlawful obstruction of competition.

Relevance to data infrastructure

The case establishes an important conceptual principle:

Control over infrastructure that competitors cannot practically bypass can create competition-law obligations concerning access.

A similar question can arise where a company controls:

indispensable cloud infrastructure;

a critical data-processing facility;

a strategically essential network;

an important interconnection point.

However, modern competition law does not automatically impose a duty to deal merely because infrastructure is important.

6. Case 2 — Otter Tail Power Co. v. United States, 410 U.S. 366 (1973)

Otter Tail controlled electricity transmission facilities and was accused of restricting municipalities' access to transmission services.

The Supreme Court found that the company's conduct could constitute unlawful monopolisation.

Competition principle

The case illustrates how a vertically integrated infrastructure operator can use control over an upstream infrastructure layer to disadvantage downstream competitors.

Data-infrastructure analogy

Imagine:

Cloud infrastructure provider → computing infrastructure → downstream SaaS competitors.

If the infrastructure provider refuses access to competitors while simultaneously competing against those competitors downstream, the competitive concern becomes significantly stronger.

The issue is particularly serious where the infrastructure provider controls a bottleneck.

7. Case 3 — MCI Communications Corp. v. AT&T, 708 F.2d 1081 (7th Cir. 1983)

This is one of the most important US cases concerning the essential-facilities doctrine.

AT&T controlled significant telecommunications infrastructure. MCI sought access to the network.

The Seventh Circuit identified circumstances relevant to an obligation to provide access, including:

control of the essential facility by a monopolist;

inability of competitors reasonably to duplicate it;

denial of access;

feasibility of providing access.

Importance for data infrastructure

The case provides a useful framework for analysing:

network access;

cloud interconnection;

critical APIs;

infrastructure bottlenecks;

data transmission facilities.

Suppose a dominant infrastructure company controls an interface that competitors technically need to access customers or data.

The competition question becomes:

Is the infrastructure genuinely indispensable, or can competitors reasonably replicate or bypass it?

That distinction is crucial.

8. Case 4 — Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585 (1985)

Facts

Aspen Skiing involved four major ski facilities. Aspen Skiing Company eventually controlled three of them and discontinued a cooperative ticket arrangement with the smaller rival.

The Supreme Court found the conduct capable of constituting exclusionary monopolisation.

Importance

The case is particularly relevant to refusal-to-deal theories.

It suggests that a monopolist's decision to abandon a profitable course of cooperation with a rival can, in exceptional circumstances, constitute exclusionary conduct.

Application to data infrastructure

Consider a dominant cloud provider that:

historically allowed interoperability;

cooperated with competing platforms;

supplied infrastructure to a rival;

then terminates the relationship specifically to disadvantage that rival.

The historical course of dealing can become highly significant.

9. Case 5 — Verizon Communications Inc. v. Trinko, 540 U.S. 398 (2004)

Trinko is equally important because it establishes a limit on compulsory-access theories.

The Supreme Court held that antitrust law generally does not impose a broad obligation upon monopolists to assist competitors.

The Court was sceptical about turning antitrust law into a regulatory system requiring courts to supervise commercial relationships.

Why Trinko matters enormously for data infrastructure

Suppose a cloud provider refuses to provide access to infrastructure.

That fact alone does not necessarily establish an antitrust violation.

A claimant generally needs something more, such as:

exclusionary intent;

discriminatory treatment;

unlawful tying;

predatory conduct;

anticompetitive interoperability restrictions;

a relevant statutory duty;

conduct falling within a recognised exception.

Thus:

Infrastructure importance ≠ automatic essential-facilities obligation.

This is one of the most important qualifications in data-infrastructure monopolisation analysis.

10. Case 6 — United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)

Microsoft is perhaps the most important technology-monopolisation precedent for understanding data infrastructure.

Microsoft possessed monopoly power in the market for Intel-compatible PC operating systems.

The court examined Microsoft's conduct toward competing browser technologies and its use of control over the operating-system platform to disadvantage rivals.

Key principle

A firm possessing control over an important technological platform cannot necessarily use that control to exclude competing technologies.

Relevance to cloud infrastructure

The analogy is strong:

Operating-system platform → application ecosystem

can be compared with:

Cloud infrastructure → software/application ecosystem.

A dominant cloud provider might potentially leverage infrastructure dominance by:

tying cloud services;

restricting interoperability;

disadvantaging competing software;

using privileged access to customer information;

making rival applications technically inferior;

imposing discriminatory contractual terms.

The Microsoft case therefore illustrates the concept of leveraging platform power into adjacent markets.

11. Case 7 — United States v. Google LLC

The modern Google litigation provides an important digital-platform example.

The US Department of Justice's search case concerns monopolisation and exclusionary conduct in digital markets. The proceedings continue to involve remedies and questions concerning how a dominant technology platform's conduct should be addressed. (Department of Justice)

Relevance to infrastructure

Google demonstrates how infrastructure-like technological assets can support market power.

Search infrastructure, distribution channels, default arrangements, data and advertising technology can interact to create reinforcing competitive advantages.

The broader lesson is:

Competition authorities increasingly analyse technological ecosystems rather than isolated products.

That is highly relevant to cloud and data infrastructure.

12. Additional Case — United Brands v Commission, Case 27/76

The European Court of Justice established important principles concerning dominance.

Dominance involves a position of economic strength that enables an undertaking to behave to an appreciable extent independently of competitors, customers and consumers.

Data infrastructure application

A cloud or data infrastructure provider may possess dominance where competitors face:

significant switching costs;

infrastructure replication costs;

scarcity of specialised capacity;

ecosystem dependence;

limited alternatives;

long-term contractual lock-in.

Market share remains important, but it is not the only consideration.

13. Case — Bronner v Mediaprint, Case C-7/97

This case is particularly relevant to essential facilities.

The ECJ adopted a restrictive approach to compulsory access.

For refusal to supply to amount to abuse, the facility must generally be indispensable, meaning that there is no realistic alternative and duplication is not economically or technically feasible.

Importance for data infrastructure

A cloud provider cannot be compelled to provide infrastructure merely because competitors would prefer access.

The claimant should demonstrate something approaching:

indispensability + lack of realistic alternative + exclusionary consequences.

This prevents competition law from becoming a general price/access regulator.

14. The Central Problem: Cloud Infrastructure Lock-In

One of the most important competition concerns is cloud lock-in.

A customer may initially choose a cloud provider because of:

price;

reliability;

performance;

geographic coverage.

But over time the customer may become dependent upon:

proprietary databases;

APIs;

storage formats;

machine-learning tools;

identity systems;

security architecture;

monitoring systems.

Switching then becomes increasingly difficult.

This can produce:

initial competition → migration costs → dependency → reduced contestability.

The EU's current cloud investigation specifically focuses on issues including interoperability, financial conditions and contractual conditions surrounding cloud procurement. (Digital Markets Act (DMA))

15. Data Egress Fees

One particularly important issue is the charging of fees when customers move data away from a cloud provider.

Suppose:

Provider A charges low storage prices;

Customer stores enormous quantities of data;

Customer later wants to migrate to Provider B;

Provider A imposes substantial data-transfer/egress costs.

The customer may effectively become locked in.

Competition law could examine whether such fees are:

cost justified;

commercially reasonable;

discriminatory;

designed to discourage switching;

part of an exclusionary strategy.

However, high switching costs alone do not automatically establish abuse.

The complete competitive context matters.

16. Interoperability Restrictions

Interoperability is perhaps even more important than price.

A dominant infrastructure provider might design its system so that:

third-party software functions less effectively;

competing clouds cannot communicate efficiently;

data cannot easily be exported;

APIs are proprietary;

customers cannot use multiple providers efficiently.

This can create technical foreclosure.

A rival may technically exist but remain unable to compete effectively because access to the dominant infrastructure ecosystem is restricted.

17. Self-Preferencing

A particularly important concern arises where the infrastructure provider competes with companies that depend upon its infrastructure.

For example:

Cloud provider → supplies infrastructure to independent AI company
↓
Cloud provider → also offers competing AI product.

The infrastructure provider potentially possesses information concerning:

workload requirements;

computing demand;

customer behaviour;

technical performance;

pricing;

application architecture.

If the provider uses that information to favour its own downstream service, competition concerns arise.

This is one reason the FTC has examined major cloud/AI partnerships and their potential effects on access to computing resources and sensitive information. (Federal Trade Commission)

18. Tying and Bundling

Suppose a dominant cloud provider says:

“You may purchase our cloud computing capacity only if you also purchase our cybersecurity service.”

Or:

“Our AI infrastructure is available on favourable terms only if you use our database.”

This can raise tying or bundling concerns.

The competition authority would consider:

whether the firm is dominant in the tying product;

whether the products are distinct;

whether customers are effectively forced to purchase the tied product;

whether rivals are foreclosed;

whether there are legitimate efficiencies.

19. Predatory Pricing and Infrastructure Subsidisation

Large infrastructure providers may have enormous financial resources.

A dominant provider could potentially:

price cloud infrastructure below cost;

subsidise one service with revenues from another;

temporarily underprice competitors;

acquire customers at unsustainable prices;

subsequently increase prices after competitors exit.

But competition law must distinguish:

aggressive competition

from

predatory exclusion.

Low prices benefit consumers unless there is evidence that pricing forms part of a credible exclusionary strategy.

20. Vertical Foreclosure

Vertical foreclosure is particularly important in data infrastructure.

Consider:

Layer 1: Data centre
↓
Layer 2: Cloud computing
↓
Layer 3: Database
↓
Layer 4: AI platform
↓
Layer 5: Consumer application

If one company controls all five layers, it may have the ability and incentive to disadvantage independent competitors.

This creates the possibility of:

Input foreclosure

Competitors cannot obtain infrastructure on reasonable terms.

Customer foreclosure

The dominant infrastructure company channels customers toward its own downstream services.

Information foreclosure

The dominant firm obtains information from customers and uses it competitively.

21. Data as a Barrier to Entry

Data infrastructure creates a particularly unusual barrier because data and infrastructure reinforce one another.

For example:

More customers
↓
More workloads
↓
More data
↓
Better optimisation
↓
Lower costs
↓
Better service
↓
More customers.

This can produce a data-infrastructure feedback loop.

However, competition authorities should avoid assuming that possession of large quantities of data automatically establishes dominance.

The relevant questions include:

Is the data unique?

Is it difficult to reproduce?

Is it commercially valuable?

Is it updated continuously?

Can rivals obtain equivalent data elsewhere?

Does access to the data materially improve the competing product?

Does the infrastructure provider combine data across markets?

22. AI Makes the Infrastructure Problem More Serious

AI substantially increases the importance of data infrastructure.

Training advanced AI systems requires:

enormous computing power;

specialised GPUs;

high-speed networking;

storage;

electricity;

cooling;

data pipelines.

Consequently, control over AI infrastructure can become a strategic competitive bottleneck.

The FTC's investigation of large cloud/AI partnerships has identified potential concerns about access to computing resources, switching costs and commercially sensitive information. (Federal Trade Commission)

This produces a potentially important competitive relationship:

Cloud dominance → AI infrastructure control → AI developer dependency → downstream AI market power.

23. Minority Investments and Infrastructure Control

A company does not necessarily need 100% ownership to create competition concerns.

A dominant cloud company may invest in an AI developer while obtaining:

exclusivity;

preferred access;

board rights;

technical information;

revenue-sharing;

computing commitments;

distribution rights.

The competitive issue becomes:

Does the investment merely finance innovation, or does it strategically restrict access to an important competitor or input?

The FTC's cloud/AI investigation specifically examined these types of investment and partnership arrangements. (Federal Trade Commission)

24. Data Centre Concentration

Competition concerns may also arise at the physical data-centre level.

A region might have:

limited power availability;

limited fibre connectivity;

scarce suitable land;

limited cooling capacity;

specialised network connections.

If one company controls a disproportionate amount of available capacity, it could potentially obtain bottleneck power.

The issue is especially important where AI demand causes enormous increases in computing requirements.

Current market developments demonstrate the strategic importance of physical data-centre capacity, with major technology companies competing for rapidly expanding infrastructure and electricity capacity. (Financial Times)

25. Essential Facilities Doctrine and Data Centres

The essential-facilities doctrine should be applied cautiously.

A facility is more likely to raise serious concerns where:

a dominant undertaking controls it;

competitors cannot reasonably duplicate it;

access is indispensable;

refusal eliminates effective competition;

access is technically feasible;

the requested access can be provided without disproportionate disruption.

But Bronner, Trinko and MCI demonstrate that courts do not ordinarily transform every commercially important facility into a mandatory-access facility.

26. Merger Control

Monopolisation concerns can arise before a monopoly actually develops.

Competition authorities may scrutinise acquisitions involving:

cloud providers;

data-centre operators;

AI companies;

GPU infrastructure;

cybersecurity companies;

database providers;

connectivity networks.

A merger may raise concerns if it:

eliminates an emerging competitor;

gives the acquirer access to strategically important data;

forecloses rival infrastructure providers;

combines complementary bottlenecks;

increases switching costs.

The European Commission continues to scrutinise transactions involving computing infrastructure and data-centre assets under merger-control rules. (Competition Case Search)

27. Network Effects and Contestability

Traditional competition analysis often asks:

How many competitors exist?

For data infrastructure, another question is crucial:

Can customers realistically switch between competitors?

A market may contain five providers but still be weakly contestable if:

migration is prohibitively expensive;

data transfer is difficult;

APIs are incompatible;

contracts restrict switching;

technical expertise is provider-specific.

Thus, contestability may be more important than nominal competitor numbers.

28. Competition Between Hyperscalers

The existence of multiple hyperscalers does not automatically eliminate competition concerns.

For example:

AWS;

Microsoft Azure;

Google Cloud;

regional providers;

specialised AI clouds.

may compete at the headline level while customers remain locked into one provider because of technical architecture.

Therefore, competition authorities should distinguish:

Inter-provider competition

Can AWS, Azure and Google effectively compete for customers?

from:

Intra-customer contestability

Can an existing customer realistically move from one provider to another?

The second question is increasingly important.

29. Remedies

Where monopolisation or abuse is established, remedies may include:

A. Interoperability obligations

Require infrastructure providers to permit effective technical interoperability.

B. Data portability

Customers should be able to transfer data efficiently.

C. Restrictions on discriminatory access

The infrastructure provider should not favour its own downstream services without legitimate justification.

D. Prohibition of exclusionary contracts

Restrictions on exclusivity may be necessary.

E. Transparency

Greater transparency regarding:

pricing;

egress fees;

contractual terms;

technical limitations.

F. Access remedies

In exceptional circumstances, infrastructure access may be mandated.

G. Structural separation

The most intrusive remedy would involve separating:

infrastructure operation

from

downstream competitive services.

This should generally be reserved for situations where behavioural remedies cannot effectively restore competition.

30. Comparative Case-Law Principles

31. Key Legal Tests for a Data Infrastructure Monopoly

A competition authority should ideally proceed through the following sequence.

Step 1 — Define the market

Possible markets include:

public cloud infrastructure;

cloud storage;

cloud computing;

data-centre services;

AI computing;

specialised GPU infrastructure;

database services;

network connectivity.

Step 2 — Establish market power

Consider:

market share;

entry barriers;

switching costs;

economies of scale;

network effects;

infrastructure scarcity;

customer dependency;

financial strength.

Step 3 — Identify the infrastructure bottleneck

Ask:

What exactly does the dominant company control?

Step 4 — Identify exclusionary conduct

Examples:

refusal to deal;

discriminatory access;

tying;

bundling;

predatory pricing;

exclusivity;

interoperability restrictions;

self-preferencing;

excessive switching costs;

discriminatory use of customer data.

Step 5 — Establish competitive harm

Possible effects include:

foreclosure;

reduced innovation;

increased costs;

reduced choice;

weaker entry;

degraded interoperability;

reduced quality.

Step 6 — Examine efficiencies

The firm may argue that the conduct creates:

security benefits;

efficiency;

reliability;

innovation;

economies of scale;

technical integration.

Competition authorities must distinguish legitimate integration from exclusionary integration.

32. The Most Important Distinction: Monopoly vs Monopolisation

This distinction should always be maintained.

Monopoly

A company may become dominant because it:

innovated successfully;

built better infrastructure;

achieved economies of scale;

offered superior reliability.

That is not necessarily unlawful.

Monopolisation

The legal concern arises when the company:

uses exclusionary conduct to acquire, maintain or extend monopoly power.

Therefore:

Large market share ≠ automatic violation.

Infrastructure ownership ≠ automatic violation.

Data accumulation ≠ automatic violation.

High switching costs ≠ automatic violation.

The unlawful element normally requires anticompetitive conduct connected to the exercise or maintenance of market power.

33. Emerging Regulatory Direction

The regulatory approach is increasingly moving beyond traditional price-based analysis.

The EU's current cloud investigations specifically examine interoperability, financial conditions and contractual conditions. (Digital Markets Act (DMA)) The Commission has also indicated that AWS and Azure appear to possess entrenched positions accompanied by lock-in and switching-cost effects. (Digital Markets Act (DMA))

Similarly, the FTC has investigated cloud computing in relation to competition, security, single points of failure and AI. (Federal Trade Commission)

This indicates an emerging conception of competition in which:

technical architecture itself can become a competitive instrument.

34. Overall Legal Assessment

Data infrastructure monopolisation represents a particularly difficult competition-law problem because infrastructure, data, technology and market power reinforce each other.

The strongest concerns arise where a dominant infrastructure provider:

controls a difficult-to-replicate bottleneck;

has substantial and durable market power;

faces limited competitive constraints;

imposes significant switching costs;

restricts interoperability;

discriminates against infrastructure-dependent rivals;

uses customer data to compete downstream;

ties infrastructure to complementary services;

enters exclusive arrangements with important downstream firms;

leverages infrastructure dominance into adjacent markets.

The case law provides an important balance.

Terminal Railroad, Otter Tail and MCI demonstrate why control of bottleneck infrastructure can be competitively significant. Aspen Skiing demonstrates that refusal to cooperate can, in exceptional circumstances, be exclusionary. Trinko and Bronner establish that competition law does not generally require monopolists to share their assets with competitors. Microsoft demonstrates how technological platform control can be leveraged to foreclose adjacent competition.

The modern cloud and AI environment adds a new dimension: the infrastructure itself may become the competitive bottleneck. Current regulatory investigations into AWS, Azure, cloud interoperability and major cloud-AI partnerships show that competition authorities are increasingly examining these structural dependencies. (Federal Trade Commission)

Conclusion

The central principle can therefore be stated as follows:

Competition law should not punish firms merely for building superior data infrastructure or becoming large. It should intervene where control over indispensable or strategically important infrastructure is converted into exclusionary market power that prevents rivals from competing effectively.

In the emerging AI economy, this principle is especially important because data centres, cloud computing, GPUs, storage, networking, data and AI models are becoming vertically integrated components of a single competitive ecosystem. The future competition-law challenge will therefore be not simply identifying a “dominant platform”, but determining whether control over the underlying infrastructure gives that platform the ability and incentive to determine who can enter, scale, innovate and compete in downstream digital markets.