Competition Law and Future Institutional Frameworks for Intelligent Markets

Introduction

Intelligent markets are markets in which artificial intelligence, machine learning, automated decision-making, algorithmic pricing, predictive analytics, autonomous agents, cloud computing, large-scale data systems, and interconnected digital platforms substantially influence commercial decisions.

Competition law was traditionally designed around human firms making identifiable decisions. Intelligent markets complicate this model because market behaviour may increasingly be generated by algorithms that:

• set or adjust prices autonomously;
• rank products and services;
• determine access to platforms;
• allocate customers and resources;
• recommend products;
• predict competitor behaviour;
• control interoperability;
• learn from competitors' conduct;
• coordinate through automated systems; and
• operate across several interconnected markets simultaneously.

The OECD has specifically identified algorithmic pricing, algorithmic collusion, self-preferencing, exclusion and AI-related concentration as emerging competition concerns. It has also noted that competition authorities are developing technological expertise and algorithmic-audit capabilities.

The future institutional framework therefore cannot consist merely of traditional antitrust rules applied by conventional administrative structures. It is likely to require a combination of competition authorities, digital regulators, technical experts, data specialists, algorithmic-audit units, sector regulators, consumer authorities and international cooperation mechanisms.

I. Meaning of Intelligent Markets

An intelligent market can be understood as a market in which significant competitive variables are determined or influenced by computational intelligence.

Major characteristics

1. Algorithmic decision-making

Businesses increasingly use algorithms to determine:

• prices;
• discounts;
• product placement;
• advertising;
• inventory;
• credit;
• search rankings;
• customer segmentation; and
• market entry strategies.

2. Autonomous commercial conduct

An AI system may modify its behaviour without a new human instruction.

This creates an important legal question:

Who is legally responsible when an autonomous system produces an anticompetitive outcome?

Possible answers include:

• the deploying firm;
• the algorithm's developer;
• the platform controlling the infrastructure;
• multiple firms using the same system;
• or, depending on the legal framework, none of them under traditional doctrines unless human involvement or attributable conduct can be established.

3. Continuous market adaptation

Traditional competition investigations often examine conduct during a defined period.

AI systems, however, can continuously change:

input → analysis → decision → market reaction → new input → new decision.

Consequently, competition authorities increasingly need continuous or near-real-time monitoring capabilities.

4. Data as a competitive asset

Data may constitute an important input for:

• training AI models;
• improving recommendations;
• predicting consumer behaviour;
• detecting demand;
• developing advertising systems;
• improving search;
• and creating network effects.

Thus, control over data can become a source of market power.

5. Interconnected markets

A single enterprise may simultaneously operate:

cloud infrastructure → operating system → app store → search engine → advertising → AI model → consumer service.

Competition problems may therefore arise from leveraging power from one market into another.

II. Why Traditional Competition Institutions May Become Insufficient

Traditional competition authorities generally rely upon:

1. complaints;
2. investigations;
3. document production;
4. economic analysis;
5. witness evidence;
6. inspections;
7. expert evidence; and
8. judicial review.

These tools remain important, but intelligent markets create additional problems.

A. Speed

An investigation may take years, whereas an algorithm can alter market conditions within minutes.

B. Opacity

Modern machine-learning systems may be difficult to understand even for their operators.

C. Attribution

There may be no conventional human instruction saying:

"Fix the price at X."

Instead, an algorithm may independently learn a pricing strategy.

D. Scale

A single platform can simultaneously affect millions of transactions.

E. Multi-market leverage

AI, cloud, data, search and platform services can reinforce one another.

F. Evidence preservation

Important evidence may exist in:

• source code;
• model weights;
• logs;
• training datasets;
• API calls;
• prompts;
• system instructions;
• version histories;
• model outputs; and
• automated decision records.

The institutional framework therefore has to become technically capable as well as legally capable.

III. Existing Competition-Law Foundation

Future institutions do not necessarily require abandonment of conventional competition law.

Existing doctrines remain relevant.

1. Anti-competitive agreements

Article 101 TFEU, for example, prohibits agreements and concerted practices restricting competition, including price-fixing and market-sharing arrangements.

In intelligent markets, this can cover:

• algorithmic price coordination;
• hub-and-spoke arrangements;
• shared pricing systems;
• automated information exchange;
• restrictive platform agreements.

2. Abuse of dominance

Article 102 TFEU addresses abusive conduct by dominant firms. The European Commission's 2026 exclusionary-abuse guidelines reinforce the continuing importance of Article 102 in modern markets.

Potential intelligent-market abuses include:

• AI self-preferencing;
• discriminatory ranking;
• tying AI services to cloud infrastructure;
• refusal of interoperability;
• exclusionary access conditions;
• discriminatory data access;
• predatory strategies; and
• leveraging dominance from one technological layer into another.

3. Merger control

Traditional turnover thresholds may fail to capture acquisitions of:

• AI startups;
• data-rich businesses;
• emerging foundation-model developers;
• technology developers with little current revenue but substantial competitive potential.

Future merger institutions may therefore require greater attention to:

• innovation;
• data;
• computing capacity;
• ecosystems;
• nascent competition;
• interoperability; and
• future competitive constraints.

The European Commission's 2026 review of its merger guidelines expressly reflects the transformation of competitive dynamics caused by digitalisation and other structural changes.

IV. Important Case Laws

1. Google Search (Google Shopping)

European Commission / General Court – Google Shopping

The Google Shopping litigation concerned Google's preferential treatment of its own comparison-shopping service in search results.

Significance

The case demonstrated that:

• ranking systems can affect competition;
• algorithmic placement can constitute an important competitive variable;
• a dominant platform can influence downstream competitors through search architecture;
• technological neutrality cannot simply be assumed.

Institutional lesson

Competition authorities require technical specialists capable of examining ranking algorithms and platform design, rather than relying exclusively on conventional economic evidence.

The issue has subsequently been reinforced by the EU's Digital Markets Act framework, which imposes specific obligations concerning self-preferencing. In July 2026, the European Commission announced a €460 million DMA fine concerning Google's self-preferencing in Search.

2. Google Android

European Commission v Google – Android

The Android case concerned Google's conduct involving mobile operating systems, applications and related services.

Competition significance

The case illustrated the importance of:

• ecosystem power;
• tying;
• defaults;
• network effects;
• interoperability;
• platform access; and
• leveraging dominance between related technological markets.

Institutional lesson

Intelligent-market competition cannot always be analysed market-by-market.

A regulator must understand the architecture of the entire ecosystem.

For example:

Operating system → browser → search → advertising → data → AI assistant

may constitute an interconnected competitive structure.

3. United States v Microsoft

United States v Microsoft Corp.

Although predating modern generative AI, Microsoft remains an important institutional precedent.

The case concerned Microsoft's use of operating-system power and its relationship with web browsers and competing technologies.

Relevance to intelligent markets

The case illustrates the importance of:

• technological bottlenecks;
• platform leverage;
• default settings;
• network effects;
• exclusionary conduct; and
• innovation competition.

Future institutional lesson

Competition authorities must investigate not merely the immediate product but the technological layer controlling access to future markets.

This is particularly relevant to:

• AI operating systems;
• AI assistants;
• cloud platforms;
• foundation models;
• app ecosystems; and
• autonomous-agent marketplaces.

4. FTC v Qualcomm

FTC v Qualcomm Inc.

The litigation involving Qualcomm concerned licensing practices and competition in mobile communications technology.

Significance

It illustrates the difficulties surrounding:

• technology markets;
• intellectual property;
• licensing;
• standards;
• vertical relationships;
• network effects; and
• access to technologically important inputs.

Institutional lesson

Future intelligent markets may involve similar bottlenecks involving:

• AI chips;
• cloud computing;
• foundation models;
• proprietary datasets;
• model APIs;
• AI inference infrastructure.

A competition authority may therefore need specialists capable of analysing technical dependency chains.

5. Ohio v American Express

Ohio v American Express Co., 585 U.S. 529 (2018)

The U.S. Supreme Court considered competition in a two-sided transaction platform.

Significance

The case is particularly important for platform economics.

Digital intelligent markets frequently involve multiple user groups:

consumer ↔ platform ↔ merchant

or:

developer ↔ operating system ↔ consumer

or:

advertiser ↔ platform ↔ user.

Institutional lesson

Competition authorities must increasingly understand:

• two-sided markets;
• indirect network effects;
• platform pricing;
• cross-group effects;
• platform rules; and
• ecosystem-level competition.

Intelligent-market regulation therefore requires multi-sided market analysis, rather than looking exclusively at the price paid by one group.

6. Epic Games v Apple

Epic Games, Inc. v Apple Inc.

The dispute concerning Apple's App Store practices provides an important example of competition issues involving:

• digital distribution;
• platform rules;
• payment systems;
• developer access;
• steering;
• commissions; and
• ecosystem control.

Institutional lesson

Digital platforms can simultaneously function as:

1. infrastructure;
2. marketplace;
3. gatekeeper;
4. competitor; and
5. rule-maker.

This creates an institutional problem that conventional competition law may address only after substantial harm has occurred.

The EU's DMA represents one response to this problem by imposing specified obligations on designated gatekeepers. The Commission has also been conducting investigations and specification proceedings concerning interoperability, steering and other platform obligations.

7. Amazon Marketplace Investigations and Enforcement

Amazon's treatment of marketplace data, sellers and its own retail activities has generated significant competition-law scrutiny in multiple jurisdictions.

Institutional significance

The Amazon experience illustrates the platform-as-referee-and-player problem:

The same entity may operate the marketplace while competing with businesses dependent upon that marketplace.

An intelligent platform can potentially use:

• seller data;
• transaction data;
• search information;
• ranking systems;
• pricing information;
• inventory data; and
• algorithmic recommendations

to influence downstream competition.

Future institutional lesson

Authorities may need dedicated platform-monitoring divisions capable of examining algorithmic ranking, data access and self-preferencing continuously.

V. Core Elements of a Future Institutional Framework

1. Specialist Competition Authority

The future competition authority should contain multidisciplinary divisions.

Possible structure

Competition Authority

→ Legal Division
→ Economics Division
→ Digital Markets Division
→ AI & Algorithms Division
→ Data & Privacy Division
→ Merger Intelligence Division
→ Cyber/Forensic Technology Division
→ Consumer Protection Division
→ International Cooperation Division

The European Commission already provides an example of increasing technological institutional capacity: its Chief Technology Officer function supplies technological advice on algorithmic, data and forensic issues within competition enforcement.

VI. Algorithmic Audit Unit

A dedicated Algorithmic Competition Audit Unit could investigate:

• pricing algorithms;
• recommendation systems;
• ranking algorithms;
• advertising auctions;
• matching systems;
• allocation algorithms;
• AI agents;
• automated bidding;
• discriminatory algorithms;
• self-preferencing systems.

It could examine:

Inputs

What information does the system receive?

Objective function

What is the system designed to optimise?

Constraints

What restrictions are programmed into it?

Learning mechanism

How does it change its behaviour?

Outputs

What decisions does it generate?

Feedback loop

How does market reaction influence future decisions?

VII. AI and Algorithmic Evidence Infrastructure

Traditional evidence rules must evolve.

Authorities may need legally protected access to:

• model logs;
• audit trails;
• API records;
• source-code segments;
• model documentation;
• training-data categories;
• decision histories;
• system prompts;
• deployment configurations;
• model versions; and
• automated communications.

This is particularly important because algorithmic conduct can be difficult to reconstruct after the event.

VIII. Real-Time Competition Monitoring

A future competition authority may move from a predominantly reactive model to a combination of:

Ex post enforcement

Investigate conduct after it occurs.

Ex ante regulation

Set rules before harmful conduct develops.

Continuous monitoring

Monitor high-risk markets during their operation.

This is increasingly relevant because AI systems can change commercial behaviour rapidly. The OECD has observed that algorithmic pricing creates both efficiency benefits and risks of coordination and exclusion.

IX. Competition Sandboxes

Competition authorities could establish regulatory competition sandboxes for emerging technologies.

Participants could test:

• autonomous pricing;
• AI marketplaces;
• algorithmic bidding;
• autonomous agents;
• data-sharing arrangements;
• interoperability systems.

The authority could identify competition risks before large-scale deployment.

X. Institutional Coordination

Intelligent markets frequently cross regulatory boundaries.

A single AI platform could implicate:

• competition law;
• consumer law;
• data protection;
• cybersecurity;
• financial regulation;
• telecommunications regulation;
• intellectual property;
• AI regulation.

Consequently, a future framework should provide a formal inter-regulatory coordination mechanism.

The EU is already moving toward greater coordination between competition and data-protection authorities; in 2026 the European Commission services and EDPB announced joint work concerning the interaction between competition and data protection law.

XI. International Competition Governance

Intelligent markets are inherently international.

An AI model can be:

• developed in one country;
• trained using global data;
• hosted on foreign cloud infrastructure;
• deployed through another country's platform; and
• used by consumers worldwide.

Therefore, competition authorities need:

• information-sharing agreements;
• coordinated investigations;
• compatible merger review;
• common terminology;
• cross-border algorithmic evidence procedures;
• coordinated remedies; and
• international technical standards.

The OECD's work specifically emphasises international responses to algorithmic pricing and digital competition.

XII. New Institutional Concept: Competition Technology Office

A particularly important future institution could be a Competition Technology Office (CTO).

Its functions could include:

1. Algorithmic intelligence

Detect suspicious patterns.

2. Market simulation

Model possible competitive outcomes.

3. AI auditing

Test algorithmic systems for exclusionary or collusive effects.

4. Digital forensics

Recover evidence from complex technological environments.

5. Merger simulation

Evaluate the effect of AI acquisitions on innovation and future competition.

6. Early-warning systems

Identify emerging concentration before traditional market-share statistics reveal it.

XIII. Future Merger-Control Framework

Traditional merger control relies heavily upon:

• turnover;
• market shares;
• concentration ratios;
• entry barriers;
• efficiencies.

Intelligent markets require additional indicators.

Possible future indicators

1. control over critical datasets;
2. access to computing power;
3. ownership of foundation models;
4. control over AI distribution channels;
5. interoperability;
6. access to APIs;
7. switching costs;
8. network effects;
9. developer dependence;
10. control over essential technical standards;
11. innovation pipelines; and
12. potential competition.

This is particularly significant where a small AI company has low revenue but possesses technology capable of becoming a major competitive constraint.

XIV. Intelligent Markets and Essential Facilities

Future competition disputes may increasingly involve access to:

• cloud computing;
• AI chips;
• model APIs;
• datasets;
• operating systems;
• app stores;
• digital identity systems;
• payment infrastructure;
• interoperability interfaces.

The traditional essential-facilities doctrine may therefore acquire renewed importance, although whether a particular facility satisfies the applicable legal test will depend upon the jurisdiction and facts.

XV. Algorithmic Collusion

One of the most difficult future problems is algorithmic coordination.

Suppose:

Firm A → AI pricing algorithm

and

Firm B → AI pricing algorithm

Both algorithms continuously observe market prices and learn that aggressive price reductions reduce profits.

They may independently converge toward higher prices.

There may be:

• no email;
• no meeting;
• no express agreement;
• no telephone call.

This raises difficult questions concerning:

• agreement;
• concerted practice;
• intent;
• foreseeability;
• attribution;
• causation;
• liability.

The OECD has identified algorithmic collusion and the attribution problem as important emerging issues.

XVI. Self-Learning AI and Liability

Future legislation may need to distinguish between:

A. Human-directed conduct

The company deliberately instructs an AI system to produce an anticompetitive outcome.

B. Predictable autonomous conduct

The company did not expressly order the conduct but could reasonably foresee it.

C. Unforeseeable autonomous conduct

The system produces an outcome that was genuinely unexpected.

This could lead to future doctrines based upon:

• control;
• foreseeability;
• deployment responsibility;
• monitoring obligations;
• compliance design; and
• failure to intervene.

XVII. Competition Compliance in Intelligent Firms

Future competition compliance programmes may need to include AI compliance by design.

Companies could be required or encouraged to maintain:

• algorithmic risk assessments;
• competition-risk documentation;
• model governance;
• audit logs;
• human oversight;
• automated alerts;
• algorithmic testing;
• documentation of model changes.

Competition compliance would therefore evolve from:

"Do not agree with competitors."

toward:

"Design, deploy and supervise automated systems so that they do not facilitate prohibited coordination or exclusion."

XVIII. Remedies in Intelligent Markets

Traditional remedies include:

• fines;
• injunctions;
• behavioural commitments;
• structural separation;
• divestiture.

Intelligent markets may require technological remedies.

Possible remedies

1. Interoperability

Require systems to communicate with competitors.

2. Data portability

Enable users or competitors to transfer relevant data where legally appropriate.

3. Algorithmic neutrality

Require ranking systems to apply non-discriminatory criteria.

4. API access

Provide fair access to important interfaces.

5. Separation of data

Prevent competitive information obtained from marketplace participants from being improperly used by the platform's competing business.

6. Model-access remedies

Where legally justified, provide access to particular technological inputs.

7. Algorithmic monitoring

Require independent audits.

XIX. Ex Ante and Ex Post Institutional Models

The future framework is likely to combine two approaches.

The EU's Digital Markets Act represents an important example of the ex-ante approach, while traditional Articles 101 and 102 TFEU remain central to ex-post enforcement.

XX. Future Institutional Model

A comprehensive future model could therefore be represented as:

Intelligent Market

↓

Competition Monitoring System

↓

Algorithmic Detection

↓

Economic Analysis + Technical Audit

↓

Competition Authority

↙ ↓ ↘

Antitrust Enforcement — Ex Ante Regulation — Merger Control

↓

Inter-Regulatory Coordination

↓

National Courts / Specialized Tribunal

↓

International Competition Cooperation

This represents a transition from a firm-centred competition authority toward an ecosystem-centred competition institution.

XXI. Major Institutional Challenges

1. Technological expertise

Authorities must recruit:

• AI engineers;
• data scientists;
• economists;
• cybersecurity specialists;
• software engineers;
• forensic experts.

2. Independence

Technical experts must operate independently from both regulated firms and political institutions.

3. Confidentiality

Algorithmic investigations may expose:

• trade secrets;
• source code;
• proprietary datasets;
• security vulnerabilities.

4. Due process

Automated enforcement cannot replace legal standards of:

• notice;
• evidence;
• proportionality;
• reasoned decisions;
• appeal.

5. Regulatory overlap

Competition, AI, privacy and sector regulators may investigate the same technology.

6. False positives

An algorithmic anomaly does not necessarily establish an infringement.

Authorities must distinguish:

correlation → economic effect → competitive harm → legally actionable conduct.

XXII. Future Evolution of Competition Authorities

The competition authority of the future is likely to evolve through several stages.

Stage 1 — Traditional Authority

Primarily lawyers and economists.

Stage 2 — Digital Competition Authority

Addition of economists specialising in platforms and digital markets.

Stage 3 — Algorithmic Authority

Dedicated AI, data and algorithmic investigation capabilities.

Stage 4 — Predictive Competition Authority

Continuous monitoring and early-warning systems.

Stage 5 — Intelligent Competition Authority

Use of AI itself to:

• detect suspicious conduct;
• simulate markets;
• identify emerging concentration;
• analyse enormous datasets;
• monitor algorithmic changes.

Stage 6 — Networked Competition Governance

Domestic and international authorities exchange technical intelligence and coordinate remedies.

XXIII. Six Major Principles for Future Institutional Frameworks

Principle 1 — Technological neutrality

Competition law should regulate competitive effects rather than favour or prohibit particular technologies merely because they are technologically novel.

Principle 2 — Algorithmic accountability

Firms should remain responsible for competition risks created by systems they deploy, subject to applicable legal standards.

Principle 3 — Ecosystem analysis

Competition authorities should examine interconnected technological ecosystems rather than isolated products.

Principle 4 — Continuous supervision

High-risk intelligent markets may require continuing monitoring rather than one-time investigations.

Principle 5 — Multidisciplinary enforcement

Competition enforcement should integrate:

law + economics + computer science + data science + cybersecurity.

Principle 6 — Procedural safeguards

Greater technological powers must be accompanied by:

• judicial oversight;
• confidentiality protection;
• transparency;
• proportionality;
• rights of defence;
• independent review.

XXIV. Overall Legal Significance

The central transformation is from:

competition between firms

toward:

competition within technologically mediated ecosystems.

The cases involving Google, Microsoft, Qualcomm, American Express, Apple and Amazon demonstrate different aspects of this transition:

These precedents do not themselves establish a single future legal framework. Rather, they provide different doctrinal foundations from which future institutions can develop.

Conclusion

Competition law in intelligent markets will require institutional transformation rather than merely new substantive rules.

The future competition authority will need to be:

• technologically competent;
• economically sophisticated;
• capable of algorithmic auditing;
• equipped for digital forensics;
• able to analyse ecosystems;
• capable of continuous monitoring;
• coordinated with privacy, AI and sector regulators;
• prepared for cross-border investigations; and
• subject to strong procedural safeguards.

The fundamental challenge is that market power may increasingly reside not simply in ownership of a product, but in control over algorithms, data, computing infrastructure, interfaces, standards, distribution systems and intelligent decision-making architectures.

The emerging institutional response is already visible in developments such as specialised technological capacity within competition authorities, ex-ante digital-market regulation, algorithmic-pricing research and greater coordination between competition and other regulators. The OECD has specifically identified the need for competition authorities to adapt their analytical tools and institutional capabilities to rapidly evolving digital markets.

Accordingly, the future framework can be conceptualised as:

Traditional Antitrust + Digital Regulation + Algorithmic Auditing + Data Governance + Technical Expertise + Continuous Market Intelligence + International Cooperation = Institutional Framework for Intelligent Markets.

The objective is not to replace traditional competition law, but to make its principles effective in markets where increasingly important competitive decisions are made, assisted or continuously modified by intelligent systems.