Competition Law and Antitrust Challenges in Computational Economies

1. Introduction

A computational economy is an economy in which important commercial decisions are increasingly made or influenced by algorithms, artificial intelligence, machine-learning systems, digital platforms, automated marketplaces, recommendation engines, and large-scale data analysis.

Traditional competition law was developed mainly for markets in which human managers determined prices, negotiated agreements, selected suppliers, and decided how products would be distributed. Computational markets complicate this model because algorithms can now determine prices in real time, rank businesses, recommend products, allocate advertising, predict consumer behaviour, and coordinate enormous volumes of transactions.

The central competition-law question is therefore not whether algorithms themselves can violate the law. Rather, the question is whether businesses use computational technologies in ways that restrict competition, facilitate coordination, exclude competitors, reinforce monopoly power, or harm the competitive process.

The main legal frameworks remain familiar: Article 101 TFEU addresses anticompetitive agreements and concerted practices in the European Union, while Article 102 TFEU addresses abuse of dominance. In the United States, Sections 1 and 2 of the Sherman Act address agreements restraining trade and monopolization respectively.

Computational economies do not replace these rules. They create new factual situations in which the existing principles must be applied.

2. Algorithmic Pricing and Automated Coordination

One of the most significant antitrust problems is algorithmic pricing.

Businesses increasingly use software that continuously examines demand, supply, competitors' behaviour and historical transactions before recommending or automatically setting prices.

Using an algorithm to determine a firm's own price is not automatically unlawful. Problems arise when competing businesses supply sensitive information to the same system or otherwise use technology to coordinate their commercial decisions.

A common pricing platform could potentially reduce genuine independent decision-making. For example, if several competing sellers provide confidential pricing, output and demand information to the same algorithm and then rely upon its recommendations, competition authorities may investigate whether the arrangement amounts to unlawful coordination or information exchange.

The problem becomes especially important because computational systems can process information and respond to competitors much faster than human managers.

3. Algorithmic Collusion

Traditional cartels normally involve some form of agreement between competitors.

Computational economies create more difficult possibilities. Algorithms might:

• implement an agreement already made by humans;
• act as communication mechanisms between competitors;
• use competitors' confidential data to generate common recommendations;
• rapidly detect and punish deviations from coordinated behaviour; or
• independently learn strategies that result in highly parallel market conduct.

The first situations fit relatively comfortably within traditional competition law.

The last situation is considerably more difficult. If independently developed algorithms simply learn that matching competitors' prices maximizes profits, establishing the legally required agreement or concerted practice may be difficult.

Consequently, competition law must distinguish unlawful coordination from lawful conscious parallel behaviour.

4. Data as a Source of Market Power

Computational businesses often depend heavily on data.

Large platforms may possess information concerning:

• consumer searches;
• purchasing behaviour;
• advertising performance;
• location patterns;
• transaction histories;
• seller performance;
• browsing behaviour; and
• user interactions.

Large datasets can improve algorithms, which may attract additional users and generate still more data.

This creates a possible feedback mechanism:

More users → more data → better algorithms → better services → more users.

Such advantages are not automatically anticompetitive. Competition law generally does not punish a business merely because it has developed superior technology.

However, concerns can arise where control over data is combined with exclusionary contractual practices, discriminatory access rules or other conduct capable of preventing effective competition.

5. Digital Platforms and Computational Gatekeepers

Many computational markets operate through multi-sided platforms connecting consumers, advertisers, merchants, developers or service providers.

Platforms can simultaneously act as:

1. marketplace operator;
2. rule-maker;
3. data collector;
4. ranking system;
5. advertising intermediary; and
6. competitor to businesses using the platform.

This creates potential conflicts of interest.

For example, a platform might control the algorithm determining which businesses appear prominently while simultaneously selling its own competing services.

Competition authorities therefore examine whether ranking, recommendation and access systems constitute competition on the merits or whether they unfairly leverage market power into neighbouring markets.

6. Self-Preferencing

Self-preferencing occurs when a platform gives preferential treatment to its own products or services compared with competing products operating through its ecosystem.

Computational systems make this especially important because visibility can be controlled through algorithms.

A platform might influence:

• search ranking;
• recommendation placement;
• default applications;
• advertising visibility;
• access to consumer information; or
• transaction conditions.

Not every advantage given to an integrated firm's own service necessarily constitutes an antitrust infringement. The legal analysis depends upon matters such as dominance, the nature of the conduct, market circumstances and its capability to restrict competition.

7. Network Effects

Computational markets frequently contain strong network effects.

A service becomes more valuable when additional users join it. Social networks are a straightforward example, but similar effects can exist in marketplaces, operating systems, payment networks and communication platforms.

Network effects can produce substantial efficiencies.

However, they can also make entry difficult because a new competitor must persuade consumers to leave an established network containing substantially more users.

When combined with data advantages, economies of scale and switching costs, network effects can contribute to durable market power.

8. Interoperability and Switching Costs

Competition may also depend upon whether consumers and businesses can move between computational ecosystems.

Potential barriers include:

• incompatible technical standards;
• inability to transfer data;
• closed application ecosystems;
• contractual restrictions;
• loss of digital purchases;
• proprietary interfaces; and
• dependence on platform-specific services.

High switching costs may make customers effectively locked into a particular ecosystem.

Competition authorities therefore increasingly examine whether interoperability restrictions represent legitimate product design decisions or exclusionary mechanisms.

9. Computational Market Definition

Traditional market definition asks which products consumers regard as substitutes.

Computational economies make this difficult because many digital services have a monetary price of zero.

Consumers may instead effectively exchange:

• attention;
• personal data;
• advertising exposure; or
• engagement.

Platforms can also serve several customer groups simultaneously.

For example, a search service may interact with users, advertisers, websites and merchants. Competitive conditions on one side can influence another.

Consequently, competition authorities increasingly consider factors beyond monetary price, including quality, privacy, innovation and access to data.

10. Algorithmic Transparency

Algorithms can be extremely complex and sometimes commercially confidential.

This creates an enforcement problem.

Authorities may need to determine:

• what information entered an algorithm;
• how recommendations were produced;
• whether competitors supplied sensitive information;
• whether ranking rules disadvantaged competitors;
• whether humans could override automated decisions; and
• how frequently businesses followed algorithmic recommendations.

The relevant evidence may therefore include source-code documentation, internal communications, datasets, model specifications and records showing how algorithmic outputs were implemented.

Importantly, complexity does not itself establish illegality.

Important Case Laws and Enforcement Proceedings

1. Eturas UAB and Others v Lithuanian Competition Council

Court of Justice of the European Union, Case C-74/14, 2016

Eturas concerned Lithuanian travel agencies participating in a common online booking system.

The system administrator communicated a restriction concerning discounts available through the platform, and a technical modification limited discount rates.

The Court of Justice considered when travel agencies using a shared computerized system could be regarded as participating in a concerted practice.

The Court held, in substance, that awareness of the communication concerning the restriction could be relevant to establishing participation, subject to applicable evidentiary safeguards and the presumption of innocence.

Importance

Eturas is one of the most significant early cases for computational coordination.

It demonstrates that competitors cannot necessarily avoid competition-law responsibility simply because coordination is implemented through software rather than through traditional meetings.

It is particularly relevant to modern algorithmic platforms because a common technological intermediary can potentially facilitate coordinated commercial conduct.

2. Google and Alphabet v European Commission (Google Shopping)

Court of Justice of the European Union, Case C-48/22 P, judgment 10 September 2024

The dispute concerned Google's treatment of comparison-shopping services in its general search results.

The European Commission had concluded that Google abused its dominant position by giving favourable treatment to its own comparison-shopping service while rival comparison-shopping services were disadvantaged in search results.

The General Court essentially upheld the Commission's decision.

In September 2024, the Court of Justice dismissed Google and Alphabet's appeal, leaving the approximately €2.4 billion fine in place. The judgment addressed the distinction between competition on the merits and conduct capable of excluding competitors.

Importance

Google Shopping demonstrates how algorithmic ranking architecture can become relevant under abuse-of-dominance law.

Control over a search or recommendation system may provide a dominant platform with substantial influence over competitors' visibility.

The case therefore has implications extending beyond shopping search to recommendation engines, app stores, digital marketplaces and AI-driven discovery systems.

3. United States v Google — Search Monopolization Litigation

U.S. District Court for the District of Columbia, 2024 judgment

The U.S. Department of Justice challenged Google's practices concerning general search services and search advertising.

In August 2024, the District Court concluded that Google had maintained monopoly power unlawfully through exclusionary distribution arrangements, particularly agreements securing default search-engine positions.

The subsequent remedies proceedings addressed measures intended to restore competitive opportunities, including restrictions involving distribution arrangements and access to certain search-related data and services.

Importance

This litigation demonstrates the relationship between defaults, data, scale and computational feedback loops.

Search engines improve through enormous volumes of queries and user interactions. Distribution arrangements that increase usage can therefore reinforce data and scale advantages.

The case illustrates that antitrust analysis of computational economies must examine not merely algorithms themselves but also the contractual and distribution infrastructure supplying those algorithms with users and data.

4. United States and Plaintiff States v RealPage, Inc.

U.S. antitrust proceedings beginning in 2024

The U.S. Department of Justice and several states brought proceedings against RealPage concerning revenue-management software used in rental housing.

The government's complaint alleged that competing landlords supplied non-public, competitively sensitive information that was incorporated into RealPage's pricing systems. According to the allegations, the system generated pricing recommendations and facilitated alignment among participating landlords.

The complaint invoked both Section 1 and Section 2 of the Sherman Act.

Later proceedings produced proposed settlements and remedies concerning the use of competitors' sensitive information in rental-pricing software.

Importance

RealPage represents one of the clearest modern examples of the antitrust issues surrounding common pricing algorithms.

Its significance lies in the principle that inserting an algorithm between competitors does not necessarily eliminate traditional concerns about information exchange or coordination.

The central question remains whether independent competitive decision-making has been replaced or impaired by an arrangement involving competitors and a common computational intermediary.

5. United States v Agri Stats, Inc.

U.S. Department of Justice proceedings, commenced 2023

Agri Stats provided information and benchmarking services to participants in meat-processing industries.

The U.S. government alleged that Agri Stats collected and distributed competitively sensitive information concerning matters including prices, costs and output.

The government argued that these information exchanges reduced normal competitive uncertainty among processors.

A federal court denied the defendant's motion to dismiss in 2024, allowing the government's case to proceed. In 2026, the DOJ announced proposed settlement documentation in the proceeding.

Importance

Agri Stats is highly relevant to computational economies even though it is not simply an AI-pricing case.

Modern algorithms depend on information.

Therefore, competition law increasingly focuses not only on the final algorithm but also on the data architecture feeding the system.

If competitors systematically contribute detailed confidential information to a centralized analytics service, authorities may examine whether that system facilitates coordination.

6. European Commission v Google — Android

The European Commission's Android proceedings concerned contractual restrictions associated with Google's Android mobile ecosystem.

The Commission concluded that Google had imposed restrictions on Android device manufacturers and mobile network operators that unlawfully strengthened its position in general internet search.

The case involved issues including pre-installation, distribution arrangements and restrictions associated with Google's Android ecosystem.

The General Court largely upheld the Commission's findings in 2022, while reducing the fine.

Importance

Android illustrates the competitive significance of computational ecosystems.

An operating system is not merely software. It can control access to applications, search services, data and distribution channels.

The case demonstrates how tying, default settings and ecosystem rules can reinforce an existing position in computational markets.

7. European Commission's Google AdSense Proceedings

The European Commission also investigated Google's practices involving online search advertising intermediation through AdSense for Search.

The Commission concluded that contractual provisions imposed on third-party websites restricted competing search-advertising intermediaries.

Although subsequent judicial proceedings affected aspects of the Commission's decision, the dispute remains important for understanding digital-market competition.

Importance

Computational advertising markets depend upon algorithms that determine advertising placement, relevance and pricing.

Control over the infrastructure connecting advertisers, publishers and users can therefore create strategic bottlenecks.

The case demonstrates why competition law examines contractual restrictions surrounding computational infrastructure in addition to the operation of the underlying algorithms.

11. Artificial Intelligence and Future Antitrust Problems

Generative AI creates another layer of competition concerns.

Developing advanced AI systems may require enormous quantities of:

• computing power;
• semiconductor capacity;
• training data;
• specialist engineers;
• cloud infrastructure; and
• capital.

If control over these inputs becomes highly concentrated, competition concerns may arise even before downstream AI markets fully develop.

Competition authorities may therefore examine partnerships, exclusive cloud agreements, access to computing infrastructure, acquisitions of emerging competitors and control over valuable datasets.

The objective is not to penalize technological leadership. The relevant issue is whether firms acquire or maintain market power through conduct that unlawfully restricts competitive opportunities.

12. Autonomous Algorithmic Collusion

One of the hardest theoretical questions concerns algorithms that independently learn coordinated behaviour.

Suppose competing companies independently install AI pricing systems without communicating with one another.

The algorithms repeatedly observe market conditions and eventually learn that aggressive price competition reduces profits. Each system therefore begins matching higher prices generated by the others.

Economically, the outcome might resemble coordination.

Legally, however, Article 101 TFEU and Section 1 of the Sherman Act ordinarily require some form of agreement or concerted action.

Mere parallel conduct generally does not automatically establish an unlawful agreement.

This creates a potential enforcement gap between economic coordination and legally provable collusion.

13. Computational Evidence and Antitrust Enforcement

Future investigations will increasingly involve digital evidence.

Authorities may examine:

Input data: What information was provided to the system?

Model design: What objectives was the algorithm instructed to maximize?

Communications: Did competitors discuss the algorithm or its parameters?

Recommendations: What commercial decisions did the system recommend?

Acceptance rates: How frequently were those recommendations followed?

Overrides: Could individual businesses independently reject recommendations?

Feedback mechanisms: Did competitors' decisions become inputs for later recommendations?

This allows authorities to determine whether software merely assisted independent decision-making or operated as an infrastructure for coordinated behaviour.

14. Innovation Versus Intervention

Competition law must also avoid treating technological success itself as suspicious.

Computational technologies can produce major competitive benefits:

• lower transaction costs;
• better demand forecasting;
• improved product matching;
• faster price adjustments;
• reduced waste;
• personalized services;
• easier market entry; and
• improved consumer information.

Overly broad intervention could discourage innovation.

The challenge is therefore to distinguish efficiency-producing computational behaviour from conduct that suppresses independent rivalry.

15. Major Antitrust Risks in Computational Economies

The major competition concerns can be summarized as follows:

1. Algorithmic price coordination — common algorithms may facilitate alignment between competitors.
2. Sensitive data exchanges — computational intermediaries can aggregate information that competitors ordinarily should determine independently.
3. Self-preferencing — dominant platforms may use ranking systems to favour their own services.
4. Data concentration — exclusive access to large datasets can reinforce barriers to entry.
5. Network effects — established computational platforms may become increasingly difficult to challenge as their user bases expand.
6. Default positions and distribution restrictions — controlling default access points can substantially influence consumer behaviour.
7. Interoperability restrictions — closed ecosystems can increase switching costs and restrict entry.
8. Algorithmic opacity — complex systems can make anticompetitive conduct harder to identify and prove.
9. AI infrastructure concentration — control of computing capacity, cloud services, chips or training data can influence competition in emerging AI markets.
10. Autonomous coordination — independently operating algorithms may generate coordinated outcomes without an easily identifiable traditional agreement.

Conclusion

Competition law in computational economies is fundamentally concerned with preserving independent competitive decision-making in markets increasingly governed by software, data and artificial intelligence.

Cases such as Eturas, Google Shopping, United States v Google, RealPage, Agri Stats, Google Android, and the Google AdSense proceedings illustrate different dimensions of the problem. They cover computerized coordination, algorithmic ranking, default distribution, sensitive-data exchanges, platform ecosystems and control of digital infrastructure.

The fundamental antitrust principles have therefore not disappeared. Agreements restricting competition can still violate Article 101 TFEU or Section 1 of the Sherman Act, while exclusionary conduct by firms possessing substantial market power can fall under Article 102 TFEU or Section 2 of the Sherman Act.

What has changed is the mechanism through which market power and coordination operate.

In a computational economy, the relevant competitive instrument may no longer be a meeting, written agreement or conventional distribution contract. It may instead be a pricing algorithm, ranking system, dataset, API, default setting, recommendation engine or shared analytical platform.

For that reason, modern antitrust analysis increasingly requires competition authorities and courts to understand not only markets and contracts but also data flows, algorithms, platform architecture and computational decision-making.

Add six clearly numbered case lawsFix the missing section numbers

Add six clearly numbered case laws

Fix the missing section numbers

Add jurisdiction-specific legal analysis