Global Convergence of AI Competition Regulatory Regimes

Introduction

The global convergence of AI competition regulatory regimes refers to the increasing similarity among competition-law approaches adopted by major jurisdictions toward artificial intelligence markets. Although the United States, European Union, United Kingdom, China, Japan, India, Australia and other jurisdictions retain different statutory frameworks, regulators are increasingly confronting the same structural AI risks:

• concentration of computing capacity;
• control over foundation models;
• access to high-quality data;
• vertically integrated cloud–AI ecosystems;
• exclusive or preferential AI partnerships;
• AI-related mergers and acquisitions;
• interoperability and switching barriers;
• algorithmic coordination;
• discriminatory access to APIs and compute;
• self-preferencing by integrated platforms; and
• dependence of downstream businesses on a small number of AI infrastructure providers.

The convergence is therefore not complete harmonisation of legislation. Rather, it is a convergence of regulatory concepts, enforcement theories, investigative techniques and remedies.

1. Meaning of Regulatory Convergence in AI Competition Law

Traditional competition law generally asks whether firms have acquired or exercised market power through conduct such as:

1. cartels;
2. exclusionary agreements;
3. abuse of dominance;
4. monopolisation;
5. anticompetitive mergers; or
6. vertical restraints.

AI introduces an additional layer because market power may arise from control over technological infrastructure rather than merely conventional products.

An AI ecosystem can be represented as:

Chips → Cloud/Compute → Data → Foundation Models → APIs → Applications → Users

Control at one layer can reinforce power at another.

For example:

GPU scarcity → cloud concentration → foundation-model concentration → API dependency → application-level foreclosure.

Consequently, competition authorities increasingly analyse AI through an ecosystem and vertical-stack perspective.

2. Major Drivers of Global Convergence

A. Similar AI Market Structures

The largest AI markets frequently exhibit similar characteristics:

• high fixed costs;
• enormous computational requirements;
• network effects;
• economies of scale;
• data advantages;
• technical switching costs;
• ecosystem integration;
• proprietary infrastructure; and
• substantial barriers to entry.

These characteristics make traditional competition concepts such as essential facilities, foreclosure, tying, leveraging, refusal to supply and vertical integration increasingly relevant.

B. Convergence Around Foundation Models

Foundation models create competition concerns because a small number of firms may control:

• training infrastructure;
• training datasets;
• model weights;
• safety infrastructure;
• inference capacity;
• APIs;
• developer ecosystems; and
• distribution channels.

Regulators are therefore examining whether partnerships between major technology companies and AI developers create de facto concentration without a conventional merger.

This is particularly important because a minority investment or exclusive cloud arrangement may achieve some of the competitive effects of an acquisition.

3. United States: Antitrust Applied to AI

The United States generally approaches AI competition through existing antitrust statutes, particularly:

• Sherman Act §1;
• Sherman Act §2;
• Clayton Act §7; and
• FTC Act §5.

The principal concern is whether AI arrangements create:

• monopolisation;
• attempted monopolisation;
• exclusionary conduct;
• unlawful agreements;
• anticompetitive acquisitions; or
• vertical foreclosure.

The U.S. approach therefore demonstrates an important form of convergence: new technology does not necessarily require an entirely new competition statute.

4. European Union: Digital Markets and Competition Convergence

The European Union combines traditional competition law with the Digital Markets Act (DMA) and other digital regulation.

AI competition concerns can intersect with:

• Article 101 TFEU;
• Article 102 TFEU;
• merger control;
• DMA obligations;
• data-access principles;
• interoperability;
• self-preferencing;
• tying and bundling.

The EU is particularly concerned with preventing dominant digital ecosystems from extending their market power into AI.

The European approach therefore tends toward ex ante plus ex post regulation.

5. United Kingdom

The UK increasingly combines traditional competition law with the digital-markets framework administered by the Competition and Markets Authority.

AI-related concerns include:

• strategic market status;
• access to computing resources;
• control of foundation models;
• cloud concentration;
• interoperability;
• partnerships between incumbent platforms and AI developers;
• data advantages; and
• ecosystem leverage.

The UK approach is significant because it attempts to intervene before entrenched AI market power becomes irreversible.

6. China

China approaches AI competition through its broader framework involving:

• Anti-Monopoly Law;
• platform regulation;
• algorithm governance;
• data regulation;
• digital-economy supervision.

Chinese enforcement is particularly relevant where large platforms use:

• algorithms;
• data;
• platform ecosystems;
• preferential treatment;
• technological integration

to reinforce market power.

China's approach consequently illustrates convergence between competition regulation and technology governance.

7. India

India's Competition Act framework increasingly encounters AI issues through:

• abuse of dominance;
• vertical restraints;
• combinations;
• access to essential technological inputs;
• data advantages;
• platform ecosystems;
• algorithmic pricing.

The Competition Commission of India can potentially examine AI competition through established doctrines rather than waiting for AI-specific legislation.

This is important for global convergence because the same theories used in U.S., EU and UK enforcement can potentially be adapted to Indian digital markets.

8. Australia and Other Jurisdictions

Australia, Japan, Canada and other developed competition regimes similarly face issues involving:

• cloud concentration;
• digital platforms;
• AI partnerships;
• data access;
• interoperability;
• algorithmic coordination; and
• technology-sector mergers.

The result is a gradually developing international competition-policy vocabulary for AI.

9. At Least Six Important Case Laws

Because AI-specific reported competition decisions remain relatively limited, many of the most important precedents are pre-AI digital, technology, infrastructure and platform cases whose principles are being applied or considered in AI markets.

Case 1: United States v. Microsoft Corp. (2001)

The Microsoft litigation is one of the foundational cases for modern digital competition law.

Microsoft was found to have engaged in exclusionary conduct designed to protect its operating-system monopoly, particularly through conduct concerning Internet Explorer and competing technologies.

Relevance to AI

The case illustrates how an incumbent controlling an important technological platform can use that position to suppress emerging competitors.

The same theory can arise where:

dominant cloud platform + proprietary AI services + preferential distribution

creates barriers for competing AI developers.

Principle

A firm may violate competition law when it uses dominance in one technological layer to exclude competition in another.

10. Case 2: United States v. Google — Search and Advertising Litigation

The modern Google antitrust litigation demonstrates the application of traditional monopolisation principles to digital ecosystems.

The central concern has been whether Google used exclusionary arrangements and control over distribution to preserve dominance.

AI significance

The case provides a framework for examining whether dominant digital platforms can use:

• default placement;
• distribution agreements;
• proprietary interfaces;
• ecosystem integration

to favour their own AI products.

For example, competition authorities could ask whether a dominant search or mobile ecosystem systematically privileges its own generative-AI assistant over rival systems.

Broader principle

Distribution control can be a source of AI market power even when the underlying AI model is technically contestable.

11. Case 3: European Commission v Google Shopping

The Google Shopping decision is particularly relevant to AI because it concerns self-preferencing.

The European Commission found that Google had abused its dominant position by favouring its own comparison-shopping service in search results.

The case demonstrates that competition concerns can arise when a platform controls an important infrastructure layer while simultaneously competing with businesses dependent upon that infrastructure.

AI application

A dominant platform providing:

AI infrastructure + AI marketplace + competing AI applications

could potentially create analogous concerns if it systematically gives its own AI services preferential access or visibility.

12. Case 4: Google Android — European Commission

The EU Android decision concerned Google's use of contractual arrangements involving Android that the Commission considered capable of reinforcing Google's dominance in general search.

AI relevance

The case illustrates the competition risks of leveraging ecosystem control.

An AI provider controlling:

• operating systems;
• app stores;
• cloud;
• search;
• advertising;
• AI assistants

could potentially extend dominance from one market into another.

The important lesson is that AI competition cannot always be analysed as a standalone "AI model market."

13. Case 5: Qualcomm — European Commission

The Qualcomm litigation concerning exclusivity and chipset markets provides important principles concerning conditional rebates and exclusionary conduct.

AI relevance

Comparable arrangements could arise in AI infrastructure markets.

For example, a powerful cloud provider might offer:

• discounted compute;
• preferential GPU access;
• credits;
• model-hosting benefits

on condition that customers use its AI ecosystem exclusively.

Such arrangements could make rival AI infrastructure economically unviable.

Principle

Competition authorities increasingly examine economic dependency and foreclosure effects, rather than merely the formal wording of contractual arrangements.

14. Case 6: Intel v European Commission

The Intel litigation concerned rebates and exclusionary effects in the x86 CPU market.

It is important for AI because computing hardware remains a fundamental input into AI development.

AI application

The same analytical question may arise concerning:

• GPUs;
• AI accelerators;
• TPUs;
• specialised inference chips;
• compute clusters.

If a dominant supplier uses pricing arrangements to exclude competing infrastructure providers, competition authorities may apply established theories developed in Intel.

Broader principle

Competition at the hardware layer can determine competition at the AI-model layer.

15. Case 7: Bronner v Mediaprint

The European Court of Justice's decision in Oscar Bronner GmbH & Co. KG v Mediaprint is important for the law of essential facilities.

The Court adopted a restrictive approach to when refusal to provide access to an infrastructure facility constitutes abuse of dominance.

AI significance

The essential-facilities question is increasingly relevant to:

• compute;
• cloud infrastructure;
• data repositories;
• AI model interfaces;
• specialised AI chips.

Suppose a dominant provider controls infrastructure that competitors cannot reasonably replicate.

The question becomes:

At what point does private control over AI infrastructure create a competition-law obligation to provide access?

Bronner provides a critical doctrinal starting point.

16. Case 8: IMS Health

The IMS Health litigation concerned access to a commercially important database.

It is particularly relevant to AI because data is one of the fundamental productive inputs for machine learning.

AI relevance

A dominant company possessing an indispensable dataset might theoretically obtain competitive advantage by denying access to rivals.

The IMS Health principles help frame the question of when refusal to license or share an information resource may become abusive.

AI implication

The case contributes to convergence around:

data access + dominance + indispensability + foreclosure.

17. Case 9: Slovak Telekom

The Slovak Telekom litigation concerned exclusionary conduct involving access to telecommunications infrastructure.

Its importance extends beyond telecommunications because it demonstrates how competition law can address vertical foreclosure involving infrastructure.

AI relevance

Modern AI markets can have similar vertical structures:

infrastructure → cloud → model → application

A dominant infrastructure provider may therefore be scrutinised if it makes access to downstream competitors commercially or technically difficult.

18. Case 10: AT.40437 — Apple App Store / App Store Competition Proceedings

European competition enforcement concerning Apple's App Store practices illustrates the increasing attention paid to digital gatekeepers controlling access between developers and consumers.

AI relevance

The same gatekeeper problem may emerge in AI ecosystems.

For example:

AI operating system → AI assistant → AI application marketplace → consumers.

If the platform imposes discriminatory terms on competing AI applications while favouring its own services, regulators may examine:

• self-preferencing;
• tying;
• discriminatory access;
• interoperability;
• commission structures.

This is one of the strongest areas of convergence between digital-platform regulation and AI competition policy.

19. Convergent Regulatory Principles

Across jurisdictions, several principles are becoming increasingly similar.

19.1 Compute as a Strategic Competition Input

Traditional competition law focused on:

• raw materials;
• transportation;
• electricity;
• telecommunications infrastructure.

AI adds compute capacity.

Competition authorities may therefore analyse:

• GPU availability;
• cloud capacity;
• inference capacity;
• specialised chips;
• data-centre access.

Compute concentration can become an upstream source of downstream AI dominance.

20. Data as a Competitive Asset

Data can generate:

• economies of scale;
• model improvement;
• prediction advantages;
• personalisation;
• switching costs.

Consequently, competition authorities increasingly investigate whether dominant firms can use data advantages to prevent entry.

The convergence is particularly visible between:

competition law + data protection + digital regulation.

21. AI Partnerships as a Substitute for Acquisitions

A major emerging issue is the use of:

• minority investments;
• exclusive cloud agreements;
• preferential hosting;
• licensing agreements;
• distribution agreements;
• board arrangements;
• long-term compute contracts.

These arrangements may produce some competitive consequences similar to mergers.

Therefore, global regulators are increasingly asking:

Does formal ownership accurately describe the actual degree of economic control?

22. AI Merger Control

AI markets create novel merger concerns because a transaction involving a relatively small AI company may nevertheless eliminate a future competitive threat.

Traditional turnover-based thresholds can therefore miss strategically important acquisitions.

This has encouraged greater attention to:

• innovation competition;
• nascent competition;
• potential competition;
• access to data;
• access to compute;
• technological capabilities.

23. Algorithmic Collusion

AI systems can independently learn pricing strategies.

This creates a difficult question:

Can autonomous algorithms facilitate collusion even when humans never explicitly agree?

Global regulators increasingly distinguish between:

Traditional cartel

Human agreement → algorithm implements agreement.

Algorithmic coordination

Algorithms interact → pricing becomes coordinated.

Autonomous collusion

AI systems independently discover strategies that produce coordinated outcomes.

The third category presents one of the greatest challenges to traditional doctrines of agreement and intent.

24. Self-Learning Systems and Attribution

Competition law traditionally attributes conduct to firms and individuals.

AI complicates this because:

• models evolve;
• agents adapt;
• pricing systems learn;
• autonomous systems interact.

Regulators therefore face the question:

Who is legally responsible for an AI-generated anticompetitive outcome?

Possible approaches include:

1. responsibility of the deploying company;
2. responsibility of the system designer;
3. responsibility of the platform operator;
4. shared responsibility; or
5. liability based on foreseeable risks rather than direct intent.

25. Interoperability as a Convergent Remedy

Interoperability is increasingly important because AI ecosystems can generate switching costs.

Potential remedies include:

• API access;
• portability;
• interoperability;
• technical standards;
• data portability;
• model portability;
• cloud switching;
• reduced data-egress barriers.

The European Union's digital regulation has strongly influenced this broader international regulatory discussion.

26. Structural Remedies

Where behavioural remedies fail, authorities may consider:

• divestiture;
• separation of business units;
• restrictions on exclusive agreements;
• access obligations;
• licensing;
• interoperability requirements.

The central debate is whether AI concentration should be addressed through behavioural regulation or structural separation.

27. The Emerging Global Regulatory Model

The emerging model can be expressed as:

Traditional Antitrust

↓

Digital Competition Regulation

↓

AI-Specific Competition Oversight

↓

Infrastructure + Data + Models + Platforms

↓

Ex Ante + Ex Post Regulation

This represents convergence without complete legal uniformity.

28. Differences That Remain

Global convergence should not be confused with identical laws.

The jurisdictions therefore converge in substance, while differing in institutional design and legal terminology.

29. Key Challenges to Convergence

A. Different definitions of dominance

A firm may be dominant under one jurisdiction's methodology but not another's.

B. Different merger thresholds

Strategic AI acquisitions can escape review where traditional turnover thresholds are insufficient.

C. National-security considerations

AI infrastructure overlaps with:

• semiconductors;
• defence;
• cloud infrastructure;
• critical infrastructure.

Competition regulators may therefore interact with national-security authorities.

D. Regulatory fragmentation

Different rules concerning:

• AI safety;
• data;
• privacy;
• cybersecurity;
• competition

can produce contradictory obligations.

30. Future Direction

The most likely direction is functional convergence rather than formal global harmonisation.

Competition authorities are likely to converge around five questions:

1. Who controls compute?

2. Who controls data?

3. Who controls foundation models?

4. Who controls distribution?

5. Can competitors realistically switch or enter?

These questions could eventually become the global analytical framework for AI competition law.

Conclusion

The global convergence of AI competition regulatory regimes represents the transformation of conventional antitrust into a framework capable of addressing technologically complex ecosystems.

The emerging consensus is that AI competition cannot be examined solely at the level of the final AI product. Competition authorities must analyse the entire stack:

chips → compute → cloud → data → models → APIs → applications → distribution.

The major precedents—Microsoft, Google Shopping, Google Android, Qualcomm, Intel, Bronner, IMS Health and Slovak Telekom—provide the doctrinal foundations for this development even though most pre-date generative AI.

The central regulatory movement is therefore from company-centric antitrust toward ecosystem-centric competition regulation. The future global competition regime is likely to combine traditional antitrust, merger control, interoperability obligations, data-access principles, digital-platform regulation and AI-specific oversight.

Ultimately, convergence is likely to occur not because every country adopts identical AI competition legislation, but because regulators increasingly confront the same economic reality: control over compute, data, models and digital distribution can produce mutually reinforcing forms of market power.