Competition Law and Future Institutional Frameworks for Cognitive Economies

Introduction

A cognitive economy is an economy in which artificial intelligence (AI), machine learning, algorithms, large-scale data, automated decision-making, foundation models, autonomous agents, and digital platforms become major determinants of production, distribution, innovation, and market access.

In such an economy, competition may no longer depend only upon traditional factors such as price, output, and physical assets. Control over data, computing infrastructure, AI models, algorithms, interfaces, standards, talent, cloud capacity, and ecosystem access may determine market power.

Traditional competition institutions—competition commissions, courts, sector regulators, data-protection authorities, and consumer authorities—therefore face a need for institutional adaptation. The future framework is likely to require continuous market monitoring, algorithmic auditing, interoperability supervision, data-access governance, merger scrutiny, and coordination between competition and technology regulators.

1. Meaning of Cognitive Economies

A cognitive economy is characterised by the increasing use of systems capable of:

• learning from large datasets;
• predicting consumer behaviour;
• making or assisting commercial decisions;
• generating content and products;
• optimising prices and supply chains;
• coordinating autonomous agents;
• allocating resources automatically;
• detecting patterns unavailable to human decision-makers; and
• continuously adapting to market conditions.

Examples include:

1. AI-powered search;
2. generative-AI platforms;
3. autonomous vehicles;
4. algorithmic financial markets;
5. AI-powered healthcare;
6. automated logistics;
7. smart energy systems;
8. AI advertising markets;
9. cloud-computing ecosystems; and
10. autonomous commercial agents.

The competition-law difficulty is that market power can become embedded in technological architecture rather than merely in prices.

2. Why Traditional Competition Law Requires Institutional Adaptation

Traditional antitrust analysis generally examines:

• relevant market;
• market share;
• barriers to entry;
• market power;
• exclusionary conduct;
• collusion;
• mergers;
• consumer harm; and
• efficiency.

These concepts remain relevant, but cognitive economies introduce additional questions.

A. Data as a competitive resource

A dominant enterprise may possess:

• proprietary datasets;
• real-time behavioural data;
• training data;
• transaction histories;
• biometric information;
• feedback data; and
• model-performance data.

A competitor may therefore face a data-entry barrier even where the underlying technology is publicly available.

B. Computational capacity

AI competition may depend upon access to:

• GPUs;
• specialised AI chips;
• cloud computing;
• high-performance data centres;
• energy infrastructure; and
• model-training capacity.

Control over these resources may create new forms of bottleneck power.

C. Algorithmic interdependence

Algorithms can observe market conditions and modify prices rapidly. This creates difficult questions concerning:

• algorithmic collusion;
• tacit coordination;
• autonomous pricing;
• personalised pricing; and
• responsibility for AI-generated competitive harm.

D. Ecosystem lock-in

A firm controlling an operating system, cloud platform, app store, payment system, AI assistant, or search engine may extend its power into adjacent markets.

E. Speed of technological change

A traditional investigation may take years, whereas an AI market can change substantially within months.

Future competition institutions therefore need continuous rather than purely retrospective supervision.

3. Major Competition Concerns in Cognitive Economies

3.1 AI Model Concentration

A small number of firms may control:

• foundation models;
• training infrastructure;
• cloud platforms;
• AI chips;
• distribution interfaces.

This can create vertical and horizontal concentration.

The competition question is whether control at one technological layer can be leveraged into another.

3.2 Data Advantage and Data Accumulation

A company with millions of users may continuously generate additional data.

This can create a feedback loop:

More users → more data → better AI → better service → more users → more data

Such a loop can make entry increasingly difficult.

Competition authorities may therefore need to consider dynamic data advantages, rather than merely static market shares.

3.3 Self-Preferencing by AI Platforms

A platform providing AI recommendations could favour:

• its own products;
• its own search results;
• affiliated services;
• its own advertising products;
• its own AI applications.

The institutional question is whether regulators should impose:

• transparency obligations;
• ranking-neutrality requirements;
• interoperability;
• separation of platform and downstream activities; or
• behavioural remedies.

4. Algorithmic Collusion

Traditional cartel law generally requires evidence of an agreement or concerted practice.

AI systems complicate this because competing algorithms can independently:

• monitor competitors;
• adjust prices;
• predict reactions;
• avoid aggressive competition; and
• converge on similar prices.

The institutional challenge is distinguishing:

independent intelligent adaptation

from

algorithmically facilitated coordination.

Future institutions may require firms deploying high-risk pricing algorithms to maintain:

• audit logs;
• model documentation;
• decision records;
• training-data records;
• governance protocols; and
• compliance controls.

5. AI and Merger Control

Cognitive economies require broader merger analysis.

A transaction involving a relatively small AI company may nevertheless be competitively significant because the target may possess:

• valuable datasets;
• specialised engineers;
• unique algorithms;
• intellectual property;
• strategic computing relationships;
• a promising AI model; or
• an important user community.

Thus, turnover-based merger thresholds may under-detect acquisitions of future competitors.

Future institutional frameworks may use:

Transaction-value thresholds

A very high acquisition price may indicate that a target possesses significant competitive potential even where current revenues are low.

Innovation-based analysis

Authorities may examine whether the target could become a future technological competitor.

Data concentration analysis

Authorities may assess whether the transaction combines complementary datasets capable of reinforcing market power.

6. Interoperability as a Competition Remedy

Interoperability may become a central institutional tool.

For example, regulators could require dominant AI ecosystems to permit competitors reasonable access to:

• APIs;
• technical interfaces;
• data portability mechanisms;
• communication protocols;
• payment systems;
• cloud interfaces; and
• interoperability standards.

The objective is not necessarily to eliminate successful firms but to prevent technological ecosystems from becoming permanently closed.

7. Competition Law and the Essential-Facility Problem

Some cognitive-economy infrastructure may become sufficiently important that denial of access raises competition concerns.

Potential bottlenecks include:

• cloud infrastructure;
• AI computing;
• digital identity systems;
• payment infrastructure;
• interoperability interfaces;
• critical datasets;
• app stores;
• operating systems; and
• technical standards.

The difficult institutional question is when access should remain commercially negotiated and when competition authorities should intervene.

8. Six Important Case Laws

8.1 United States v. Microsoft Corp. (2001)

Facts

Microsoft was found liable for unlawfully maintaining its monopoly in the market for Intel-compatible PC operating systems and engaging in exclusionary conduct concerning web browsers.

Competition-law significance

The case demonstrated how control over one technological layer can be used to protect dominance against emerging technologies.

Relevance to cognitive economies

The Microsoft framework remains relevant to AI ecosystems because:

• dominant platforms may control distribution;
• complementary technologies may become competitive threats;
• technical integration may have exclusionary consequences; and
• interoperability restrictions can affect market entry.

Principle: Competition authorities must examine not only current competitors but also technological innovations capable of becoming future competitive constraints.

8.2 European Commission v. Microsoft Corp. (Microsoft) — Commission Decision of 2004 and subsequent litigation

The European Commission addressed Microsoft's refusal to provide interoperability information and its tying of Windows Media Player.

The European competition framework treated interoperability as important to maintaining effective competition.

Relevance

In cognitive economies, interoperability may involve:

• AI agents;
• cloud services;
• operating systems;
• APIs;
• data formats;
• model interfaces.

The case supports the broader proposition that technological architecture can itself have competition-law consequences.

8.3 Google Search (Shopping) — European Commission, 2017

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

Competition significance

The case is particularly important for cognitive economies because algorithmic ranking can determine which businesses receive access to consumers.

Relevance to AI

Future AI assistants may decide:

Which restaurant should the consumer see?
Which product should be recommended?
Which financial service should be presented?
Which news source should be summarised?

If the AI system systematically favours affiliated products, algorithmic self-preferencing may become a major competition issue.

8.4 Google Android — European Commission, 2018

The European Commission found that Google had imposed contractual restrictions involving Android that were considered abusive of its dominant position.

The case concerned, among other things:

• tying;
• pre-installation;
• distribution agreements; and
• restrictions affecting competing services.

Relevance to cognitive economies

The case illustrates how control over a core platform can facilitate expansion into adjacent markets.

A comparable AI ecosystem could involve:

Operating system → AI assistant → search → advertising → payments → commerce

Competition institutions may therefore need to analyse the entire ecosystem rather than isolated products.

8.5 Google AdSense — European Commission, 2019

The European Commission found that Google had imposed restrictive contractual provisions concerning online search advertising intermediation.

Relevance

AI-driven advertising may increasingly involve:

• automated targeting;
• predictive advertising;
• real-time auctions;
• personalised recommendations; and
• algorithmic optimisation.

A dominant AI advertising intermediary could potentially use contractual restrictions or technical architecture to limit competing intermediaries.

8.6 Qualcomm — European Commission, 2018

The European Commission examined Qualcomm's payments to Apple concerning the supply of LTE baseband chipsets and concluded that the conduct constituted an abuse of dominance.

Relevance to cognitive economies

The case demonstrates the importance of input-level control.

AI markets may similarly depend on scarce technological inputs such as:

• advanced processors;
• AI accelerators;
• specialised chips;
• cloud computing;
• data-centre capacity.

Competition institutions may therefore need to analyse both downstream AI services and upstream technological infrastructure.

9. Additional Relevant Case Laws

9.7 Intel v. Commission

The Intel litigation concerned rebates provided by a dominant undertaking and the appropriate assessment of exclusionary effects.

Relevance

In AI markets, dominant infrastructure providers might provide:

• preferential cloud pricing;
• rebates;
• exclusive AI-computing arrangements;
• discounts for model developers; or
• bundled infrastructure services.

The Intel jurisprudence is therefore relevant to evaluating whether rebates actually exclude equally efficient competitors.

9.8 Bronner v. Mediaprint

The Court of Justice addressed the stringent conditions governing refusal-to-deal claims and access to infrastructure.

Relevance

Cognitive economies may generate claims for access to:

• proprietary datasets;
• AI interfaces;
• cloud infrastructure;
• technical standards; or
• essential digital infrastructure.

Bronner demonstrates that compulsory access should not automatically follow merely because an input is commercially valuable.

9.9 United Brands v. Commission

The case remains foundational concerning:

• dominance;
• relevant market;
• abuse;
• refusal to supply; and
• discriminatory conduct.

Relevance

Its principles can help structure analysis of dominant AI ecosystems where control over a particular input or distribution channel affects downstream competitors.

9.10 Ohio v. American Express Co. (2018)

The U.S. Supreme Court considered competition in a two-sided transaction platform involving merchants and cardholders.

Relevance

AI ecosystems frequently operate as multi-sided markets involving:

• consumers;
• advertisers;
• developers;
• data suppliers;
• merchants;
• content providers; and
• AI-model providers.

The case illustrates why competition authorities must consider interactions among multiple sides of a platform rather than examining only one group of users.

10. Future Institutional Framework

A future competition regime for cognitive economies could contain several institutional layers.

Layer 1 — Competition Authority

Responsible for:

• abuse of dominance;
• cartels;
• mergers;
• exclusionary conduct;
• market studies;
• structural remedies.

Layer 2 — Digital/AI Regulatory Authority

Responsible for:

• algorithmic transparency;
• AI-system governance;
• interoperability;
• model accountability;
• technical standards.

Layer 3 — Data Authority

Responsible for:

• data portability;
• data access;
• privacy;
• data-sharing frameworks;
• data interoperability.

Layer 4 — Sector Regulators

Specialised authorities could supervise:

• finance;
• telecommunications;
• healthcare;
• energy;
• transport; and
• digital infrastructure.

Layer 5 — Judicial Review

Courts remain necessary to ensure:

• legality;
• procedural fairness;
• proportionality;
• evidentiary reliability;
• protection of commercial rights.

11. Proposed Institutional Architecture

A possible future model can be represented as:

AI/Data Economy

↓

Continuous Market Monitoring

↓

Competition Authority + AI Regulator + Data Regulator

↓

Algorithmic Auditing

↓

Market-Power Assessment

↓

Conduct / Merger / Access Investigation

↓

Behavioural or Structural Remedy

↓

Continuous Post-Remedy Monitoring

This represents a shift from a purely ex-post antitrust model toward a combination of:

Ex-ante regulation + ex-post enforcement + continuous technological monitoring

12. Algorithmic Auditing

Competition authorities may increasingly need technical teams capable of examining:

• source-code behaviour where legally obtainable;
• model outputs;
• pricing algorithms;
• recommendation systems;
• ranking systems;
• API restrictions;
• data-access rules;
• model-training practices;
• automated contracting; and
• communications between autonomous systems.

This requires competition agencies to recruit:

• economists;
• data scientists;
• AI engineers;
• computer scientists;
• cybersecurity specialists;
• lawyers; and
• behavioural experts.

13. Institutional Problem of Explainability

A competition authority may be unable to determine why an AI system produced a particular competitive outcome.

For example:

Algorithm A repeatedly disadvantages sellers using competing payment systems.

The authority must determine whether the result arises from:

1. legitimate optimisation;
2. biased training data;
3. technical design;
4. intentional exclusion;
5. autonomous adaptation; or
6. discriminatory programming.

Future competition institutions therefore require technical evidentiary capabilities, not merely traditional documentary evidence.

14. Market Definition in Cognitive Economies

Traditional market definition can become difficult because AI products are often:

• free to consumers;
• bundled;
• rapidly evolving;
• multi-sided;
• differentiated by quality rather than price.

Authorities may therefore need to examine:

Price

Where users actually pay.

Quality

Including accuracy, reliability, speed and privacy.

Data

Whether users provide data instead of money.

Innovation

Whether firms compete through technological improvement.

Switching costs

Whether users can move their data and AI preferences.

Ecosystem effects

Whether competition occurs between individual products or entire ecosystems.

15. Competition and Open AI Ecosystems

Open standards and open-source AI may lower barriers to entry, but they also create competition questions.

For example:

• Can a dominant platform control access to open-source models?
• Can cloud providers discriminate against competing models?
• Can a dominant app ecosystem restrict competing AI assistants?
• Can open models be commercially neutral in practice?

The institutional framework must distinguish between genuine openness and nominal openness controlled by a dominant intermediary.

16. Structural Remedies

Traditional behavioural remedies may sometimes be insufficient.

Possible structural remedies include:

• separation of platform and downstream businesses;
• divestiture;
• prohibition of exclusive agreements;
• mandatory interoperability;
• data portability;
• restrictions on self-preferencing;
• separation of infrastructure and applications;
• access obligations; and
• limitations on cross-use of competitively sensitive data.

Structural intervention, however, requires careful economic and legal analysis because excessive intervention can reduce investment incentives.

17. International Cooperation

Cognitive economies are inherently cross-border.

An AI company may have:

• developers in one country;
• cloud infrastructure in another;
• users worldwide;
• data stored in several jurisdictions;
• intellectual property registered elsewhere.

Consequently, competition institutions will increasingly need cooperation concerning:

• merger investigations;
• algorithmic conduct;
• digital evidence;
• data access;
• cross-border cartels;
• remedies; and
• coordinated enforcement.

International competition networks and cooperation between major competition authorities will therefore become increasingly important.

18. Key Institutional Principles for the Future

A future cognitive-economy competition framework should ideally follow these principles:

1. Technological neutrality

Rules should regulate competitive harm rather than particular technologies alone.

2. Dynamic analysis

Authorities should examine future competitive constraints and innovation.

3. Continuous monitoring

Competition assessment should not end when an investigation closes.

4. Technical competence

Authorities require AI and computational expertise.

5. Interoperability

Interoperability can reduce ecosystem lock-in.

6. Data portability

Consumers and businesses should not become permanently dependent upon one ecosystem because of data barriers.

7. Innovation protection

Competition policy should protect the competitive process without unnecessarily suppressing technological development.

8. Procedural fairness

AI-related enforcement must preserve due process, confidentiality and judicial review.

9. Institutional coordination

Competition, data, consumer-protection and AI authorities should cooperate.

10. Proportionality

Remedies should correspond to the demonstrated competitive harm.

19. Core Challenges for Competition Authorities

20. Conclusion

The future of competition law in cognitive economies will depend upon the ability of institutions to understand where technological control becomes economic power.

The traditional concepts of dominance, exclusion, tying, refusal to deal, discriminatory conduct, vertical restraints and merger control remain relevant. However, they must be applied to markets in which data, algorithms, AI models, computational capacity, interoperability and autonomous decision-making increasingly determine competitive conditions.

The cases involving Microsoft, Google Shopping, Google Android, Google AdSense, Qualcomm, Intel, Bronner, United Brands and American Express provide important foundations. They demonstrate recurring competition-law concerns involving platform control, interoperability, self-preferencing, technological inputs, rebates, access to infrastructure and multi-sided markets.

The future institutional model is therefore likely to move toward:

Traditional Antitrust
↓
Digital Competition Regulation
↓
AI + Data + Infrastructure Oversight
↓
Algorithmic Monitoring and Auditing
↓
Continuous Competition Governance