Competition Law and Public Algorithm Infrastructure Concentration

1. Introduction

Public algorithm infrastructure concentration refers to a situation in which a small number of undertakings, platforms, technology providers, or infrastructure operators control algorithmic systems that are widely used to deliver, allocate, rank, authenticate, monitor, or optimise public or commercially significant services.

Such infrastructure can include algorithms used for:

public procurement;

transport allocation;

digital identity;

public-benefit administration;

healthcare systems;

energy management;

financial infrastructure;

public-sector cloud computing;

digital payments;

search and ranking;

automated eligibility decisions;

traffic management;

public data platforms; and

government technology services.

The competition-law concern arises when concentration of such algorithmic infrastructure creates market power, entry barriers, exclusionary effects, discriminatory access, or opportunities to leverage dominance into neighbouring markets.

The important distinction is that an algorithm itself is not normally a competition problem. The issue is the economic and competitive power created by control over algorithmic infrastructure.

2. Meaning of Public Algorithm Infrastructure

Public algorithm infrastructure can be understood as the technological layer on which government agencies, public institutions, businesses, or citizens depend for important functions.

For example:

Data → Algorithm → Decision/Allocation → Market outcome

If one undertaking controls the algorithmic layer, it may potentially influence downstream market conditions.

Consider a public procurement platform:

Suppliers → algorithmic tender platform → ranking/evaluation → contract allocation.

If the platform operator also competes in the downstream market, its access to information and control over the algorithm may create a potential competitive conflict.

3. What Is Infrastructure Concentration?

Infrastructure concentration occurs when critical technological capacity is controlled by a limited number of undertakings.

It can arise through:

acquisitions;

exclusive government contracts;

proprietary technology;

network effects;

economies of scale;

control over datasets;

interoperability restrictions;

intellectual-property rights;

switching costs;

technical standards.

Concentration can produce genuine efficiencies.

For example, one algorithmic platform may reduce:

administrative costs;

duplication;

transaction costs;

fraud;

processing time.

The competition concern arises when concentration becomes sufficiently strong that competitors cannot effectively enter or expand.

4. Why Algorithms Can Create Market Power

Algorithms can create market power through several mechanisms.

A. Data advantage

An algorithm becomes more effective when trained on large quantities of data.

This can create:

more users → more data → better algorithm → more users.

This feedback loop can make entry difficult.

B. Network effects

A public algorithmic infrastructure may become more valuable as more:

agencies;

businesses;

suppliers;

consumers

use it.

A competing platform may therefore face substantial difficulties in achieving sufficient scale.

C. Switching costs

Once institutions integrate their systems with a particular algorithmic infrastructure, switching may require:

new software;

retraining;

migration;

cybersecurity testing;

contractual changes;

interoperability work.

These costs can reinforce incumbent market power.

5. Public Procurement as an Example

Consider a government procurement algorithm that:

receives supplier bids;

evaluates bids;

ranks suppliers;

recommends winners.

If the company operating the infrastructure also sells goods or services to the government, several competition concerns may arise.

It might theoretically:

favour affiliated suppliers;

manipulate ranking parameters;

use competitor information;

impose discriminatory access conditions.

Such conduct would require evidence and legal analysis; the mere existence of a dual role does not establish an infringement.

6. Algorithmic Infrastructure and Abuse of Dominance

Where an undertaking has substantial market power, competition law may become relevant to conduct such as:

discriminatory access;

refusal to interoperate;

self-preferencing;

tying;

exclusionary contractual terms;

discriminatory algorithms;

degradation of competitors' access.

Under an abuse-of-dominance framework, the central question is generally whether the conduct unfairly excludes or disadvantages competitors or otherwise constitutes prohibited abuse.

7. Algorithmic Infrastructure and Essential Facilities

Certain algorithmic infrastructures could potentially be argued to constitute an essential facility.

Examples might include:

a dominant identity-verification system;

an infrastructure-access platform;

a government-mandated transaction network;

a critical digital authentication service.

The analysis would require determining whether:

access is indispensable;

alternatives exist;

duplication is realistically possible;

refusal eliminates effective competition;

access can technically be provided.

The threshold for imposing compulsory access is generally demanding.

8. Public Algorithm Infrastructure and Information Advantages

A platform operator may have access to large quantities of information concerning competitors.

For example, a procurement platform might know:

bid prices;

supplier costs;

capacity;

geographic coverage;

future bids.

If the platform operator also participates in the market, the information asymmetry could become competitively important.

Potential safeguards include:

data firewalls;

anonymisation;

restricted employee access;

independent governance;

audit mechanisms.

9. Important Case Laws

1. United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)

Microsoft is a foundational case for understanding technological infrastructure and monopoly power.

Microsoft possessed a dominant position in PC operating systems and engaged in conduct designed to protect that position against competing technologies.

The court examined restrictions affecting browser competition and Microsoft's use of its operating-system position.

Relevance

A dominant digital infrastructure provider can potentially use control over a foundational technological layer to disadvantage competitors operating at another layer.

For public algorithm infrastructure, the analogous concern arises where the infrastructure provider uses control over the core system to disadvantage downstream competitors.

10. European Commission v. Microsoft, Case T-201/04 (2007)

The European Microsoft litigation involved Microsoft's refusal to provide interoperability information to competing work-group server products.

The EU courts upheld the Commission's findings concerning Microsoft's conduct.

Relevance

This case is particularly important for algorithmic infrastructure because interoperability may determine whether competing systems can function effectively.

If a dominant infrastructure operator prevents competitors from obtaining necessary interoperability information, competition may potentially be restricted.

However, the case involved exceptional circumstances and should not be interpreted as creating a general duty to disclose proprietary technology.

11. Bronner v. Mediaprint, Case C-7/97 (1998)

Bronner concerned access to a newspaper home-delivery network.

The Court established a demanding standard for compulsory access.

Relevance

The case provides an important framework for analysing whether algorithmic infrastructure can be considered indispensable.

The key questions include:

Is the infrastructure genuinely indispensable?

Can competitors develop alternatives?

Is duplication economically feasible?

Does refusal eliminate effective competition?

This prevents competition law from automatically transforming every important infrastructure into a mandatory shared facility.

12. IMS Health v. NDC Health, Case C-418/01 (2004)

IMS Health concerned access to a pharmaceutical data structure protected by intellectual-property rights.

The Court considered when refusal to license could constitute abuse of dominance.

Relevance

Public algorithmic infrastructure may similarly incorporate:

proprietary software;

databases;

technical structures;

intellectual-property rights.

The case demonstrates the need to balance innovation incentives with the possibility of exclusionary use of intellectual property.

13. Google Shopping, Case AT.39740 / T-612/17

The Google Shopping proceedings involved Google's treatment of its own comparison-shopping service within its search results.

The European Commission found that Google had favoured its own service, and the General Court largely upheld the Commission's decision.

Relevance

The case demonstrates the importance of algorithmic ranking and gateway control.

A company controlling a critical algorithmic infrastructure can potentially influence which competitors receive visibility or access.

For public algorithm infrastructure, the analogous concern could arise where the infrastructure determines:

supplier rankings;

access priority;

eligibility;

search results;

service allocation.

14. FTC v. Qualcomm, 969 F.3d 974 (9th Cir. 2020)

The Qualcomm litigation concerned cellular technology, licensing and standard-essential patents.

The Ninth Circuit ultimately reversed the district court's judgment against Qualcomm.

Relevance

The case is useful because technical standards can become foundational infrastructure for an entire ecosystem.

It demonstrates that:

technological importance alone does not establish an antitrust violation;

intellectual-property rights can create substantial commercial power;

competition analysis must identify actual exclusionary conduct.

This principle is highly relevant when assessing public algorithmic standards and infrastructure.

15. Rambus Inc. v. FTC, 522 F.3d 456 (D.C. Cir. 2008)

Rambus concerned conduct in a standard-setting environment and allegations involving disclosure of intellectual-property interests.

Relevance

Standard-setting is highly relevant to algorithmic infrastructure.

A technical standard adopted by a public authority or industry may become extremely important to downstream competitors.

Manipulation of standard-setting can potentially create:

artificial entry barriers;

technological foreclosure;

intellectual-property advantages.

The case demonstrates why competition authorities may need to examine how an infrastructure standard became dominant, not merely its current market position.

16. North Carolina State Board of Dental Examiners v. FTC, 574 U.S. 494 (2015)

The U.S. Supreme Court examined antitrust immunity for a state professional board whose members were active participants in the regulated market.

The Court held that active state supervision was necessary for the board to receive state-action immunity.

Relevance to public algorithm infrastructure

The case provides a broader governance principle.

Where market participants themselves exercise control over infrastructure or regulatory mechanisms, there can be a risk that infrastructure rules are used to protect incumbents.

This principle can be relevant where private technology providers participate in designing or operating public algorithmic systems.

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

The case involved telecommunications infrastructure and AT&T's refusal to provide access to its network under the circumstances considered by the court.

It is an important historical authority concerning essential facilities and telecommunications.

Relevance

Telecommunications networks are examples of infrastructure that can become critical inputs for competitors.

The analogy extends to digital algorithmic infrastructure where a dominant system controls access to a critical technical layer.

The case also demonstrates that infrastructure-related antitrust analysis is highly dependent upon market structure and factual circumstances.

18. Algorithmic Self-Preferencing

Self-preferencing occurs where a platform gives preferential treatment to its own products or affiliated services.

An algorithm may:

rank affiliated products higher;

provide better access;

allocate more favourable opportunities;

reduce competitors' visibility.

This can be particularly significant when the infrastructure is a gateway.

For example:

Public platform → algorithmic ranking → supplier visibility → contract opportunities.

If the platform operator participates downstream, the combination of infrastructure control and competitive activity can create potential foreclosure concerns.

19. Algorithmic Discrimination

Algorithmic discrimination can take several forms:

Technical discrimination

Different API access or system functionality.

Ranking discrimination

Different placement or visibility.

Access discrimination

Different eligibility or authentication requirements.

Pricing discrimination

Different access fees or transaction charges.

Data discrimination

Different access to relevant information.

Not every differentiation violates competition law.

The relevant issue is whether discrimination is connected to market power and produces legally relevant competitive harm.

20. Public Infrastructure and Competitive Neutrality

Where government-owned and private enterprises use the same algorithmic infrastructure, competitive neutrality can become important.

A public platform should ideally avoid unjustified advantages for:

state-owned enterprises;

affiliated businesses;

incumbent providers.

At the same time, governments may legitimately give preference to particular suppliers to pursue:

national security;

environmental objectives;

social policy;

public-service obligations.

The competition analysis must therefore distinguish legitimate public policy from unjustified competitive exclusion.

21. Public Algorithm Infrastructure and Merger Control

Algorithmic infrastructure concentration may also arise through acquisitions.

For example:

Large cloud company + government technology provider

or:

procurement platform + supplier analytics company

may combine:

datasets;

algorithms;

users;

infrastructure;

distribution channels.

Merger analysis may therefore examine whether the transaction creates:

increased entry barriers;

data advantages;

interoperability restrictions;

vertical foreclosure;

reduced innovation.

22. Data Feedback Loops

Algorithmic infrastructure can produce a particularly powerful feedback mechanism:

More users

↓

More data

↓

Better algorithm

↓

Better service

↓

More users

This can generate legitimate efficiencies.

However, it can also make market entry difficult because a new competitor may initially lack the data necessary to match the incumbent's performance.

Competition analysis must therefore determine whether the advantage is:

contestable;

replicable;

durable;

exclusionary.

23. Public Algorithm Infrastructure and Innovation

Concentration may have both positive and negative innovation effects.

Potential benefits

A large infrastructure provider may possess:

greater R&D resources;

economies of scale;

better cybersecurity;

advanced algorithms;

greater investment capacity.

Potential risks

Excessive concentration may:

discourage new entrants;

prevent alternative technologies;

reduce experimentation;

create technological dependency.

Competition law must therefore avoid assuming that concentration is either inherently beneficial or inherently harmful.

24. Indian Competition-Law Perspective

The Competition Act, 2002 provides several possible routes for addressing competition concerns.

Section 3 — Anti-competitive agreements

Potential issues include agreements between technology providers concerning:

interoperability;

standards;

market allocation;

technical exclusion.

Section 4 — Abuse of dominant position

Potential concerns include:

denial of market access;

discriminatory conditions;

tying;

leveraging;

exclusionary conduct.

Sections 5 and 6 — Combinations

Acquisitions involving algorithmic infrastructure may require consideration of:

data concentration;

network effects;

vertical integration;

innovation;

foreclosure.

25. Public Algorithm Infrastructure and Government Procurement

Government procurement can be particularly sensitive.

Suppose a private company operates a government procurement algorithm while also selling products to the government.

The system could potentially provide access to:

competitor bids;

market prices;

supplier capacities;

procurement forecasts.

Appropriate institutional safeguards could include:

independent administration;

access controls;

audit logs;

confidential-data separation;

transparent algorithmic rules;

conflict-of-interest safeguards.

These measures can protect both procurement integrity and competitive neutrality.

26. Interoperability as a Competition Remedy

Interoperability can reduce dependence on a single algorithmic infrastructure.

Potential approaches include:

open APIs;

standardised data formats;

common technical standards;

data portability;

multi-provider access.

But mandatory interoperability should account for:

cybersecurity;

privacy;

intellectual property;

technical reliability.

27. Algorithmic Transparency and Competition

Transparency can assist competition by allowing market participants to understand:

access requirements;

ranking criteria;

technical standards;

pricing;

interoperability rules.

However, complete disclosure of algorithms may itself create risks.

It could expose:

trade secrets;

cybersecurity vulnerabilities;

gaming opportunities.

Consequently, competition regulation may sometimes favour procedural transparency and independent auditing rather than full public disclosure of source code.

28. Essential Questions for Competition Analysis

When examining public algorithm infrastructure concentration, authorities should ask:

Market structure

Who controls the infrastructure?

How many alternative providers exist?

Entry

Can new providers enter?

Are switching costs high?

Data

Does the incumbent possess unique data?

Can competitors obtain equivalent data?

Interoperability

Can alternative systems communicate with the incumbent?

Conduct

Is access discriminatory?

Is there self-preferencing?

Is there tying or bundling?

Effects

Are competitors foreclosed?

Is innovation reduced?

Are consumers or public bodies harmed?

Justification

Is the conduct necessary for security?

Is it required for technical integrity?

Are there less restrictive alternatives?

29. Competition Remedies

Where unlawful conduct is established, possible remedies may include:

1. Non-discrimination obligations

Equivalent competitors must receive equivalent access.

2. Interoperability

The dominant infrastructure may be required to permit technical compatibility.

3. Data-access safeguards

Access can be provided under controlled conditions.

4. Structural separation

In exceptional circumstances, infrastructure and downstream commercial activities may be separated.

5. Independent auditing

An independent body may examine algorithmic operation.

6. Compliance monitoring

Ongoing monitoring may ensure that remedies remain effective.

30. Difference Between Concentration and Anticompetitive Conduct

This distinction is fundamental.

Concentration ≠ infringement.

A company may dominate algorithmic infrastructure because it:

developed superior technology;

invested heavily;

achieved economies of scale;

attracted users;

provided better security.

Competition law generally does not punish successful innovation merely because it produces a large market share.

The concern arises where market power is accompanied by prohibited exclusionary or exploitative conduct.

31. Overall Analytical Framework

A useful framework is:

Algorithmic infrastructure

↓

Market adoption

↓

Network effects / data advantages

↓

Market power

↓

Potential conduct

refusal to interoperate;

discrimination;

self-preferencing;

tying;

exclusion;

strategic data use.

↓

Competitive effects

entry barriers;

foreclosure;

reduced innovation;

reduced consumer choice.

↓

Objective justification / efficiencies

↓

Proportionate remedy

This approach prevents both under-enforcement and over-enforcement.

32. Conclusion

Public algorithm infrastructure can become an important source of economic and competitive power because algorithms increasingly perform functions that were previously carried out through physical infrastructure or conventional administrative systems.

The cases of Microsoft, European Microsoft, Bronner, IMS Health, Google Shopping, Qualcomm, Rambus, MCI Communications and North Carolina Dental Board illustrate different aspects of the legal problem:

technological platform dominance;

interoperability;

refusal to deal;

control over critical information;

standard-setting;

intellectual-property rights;

infrastructure access; and

conflicts between regulation and competitive interests.

The central competition-law principle is that algorithmic infrastructure should not be treated as anticompetitive merely because it is concentrated. Concentration can generate substantial efficiencies and innovation. The critical issue is whether control over the infrastructure is used to exclude rivals, restrict interoperability, discriminate against competitors, leverage market power, or create durable barriers to entry.

For public-sector and public-facing algorithmic systems, competition policy is particularly important because the infrastructure may simultaneously affect government procurement, private markets, innovation, access to services, and the competitive opportunities of businesses. A balanced framework therefore combines competition enforcement with interoperability, competitive neutrality, appropriate data governance, transparent access rules, and proportionate regulatory safeguards.