Competition Law and Synthetic Knowledge Platform Concentration

1. Introduction

A synthetic knowledge platform may be understood as a digital platform that uses artificial intelligence, large language models, retrieval systems, synthetic data, automated summarisation, knowledge graphs, and other computational techniques to create, aggregate, organise, transform, and distribute knowledge.

Examples may include AI-powered research engines, generative-AI knowledge assistants, scientific-information platforms, legal-research systems, educational AI platforms, and enterprise knowledge systems.

Competition-law concerns arise when a small number of firms control critical layers of this ecosystem—such as foundation models, computing infrastructure, training data, proprietary datasets, search/discovery channels, distribution platforms, or downstream knowledge applications. Concentration can therefore occur through mergers, acquisitions, exclusive agreements, vertical integration, control over data, or network effects.

The central competition-law question is not simply whether a platform is large, but whether its concentration enables it to foreclose rivals, raise barriers to entry, exploit users, restrict innovation, or extend market power from one layer of the knowledge ecosystem into another.

2. Structure of the Synthetic Knowledge Platform Ecosystem

A synthetic knowledge ecosystem can be divided into several layers:

1. Computational infrastructure
   • cloud computing;
   • GPUs and specialised AI chips;
   • data centres;
   • model-training infrastructure.
2. Data and knowledge inputs
   • books;
   • scientific publications;
   • databases;
   • websites;
   • proprietary datasets;
   • user-generated information.
3. Foundation-model layer
   • large language models;
   • multimodal models;
   • specialised scientific models;
   • knowledge-retrieval models.
4. Knowledge-generation layer
   • summarisation;
   • synthesis;
   • reasoning;
   • question answering;
   • research assistance.
5. Distribution layer
   • search engines;
   • browsers;
   • operating systems;
   • app stores;
   • educational platforms;
   • enterprise software.
6. Downstream applications
   • legal research;
   • medical research;
   • education;
   • scientific discovery;
   • financial intelligence;
   • enterprise knowledge management.

Concentration at one level can consequently reinforce concentration at another.

3. Relevant Competition-Law Framework

A. Abuse of Dominance

A dominant synthetic knowledge platform may engage in:

• discriminatory access to data;
• self-preferencing;
• tying;
• exclusive dealing;
• refusal to supply;
• discriminatory API access;
• interoperability restrictions;
• exploitative data practices;
• predatory pricing;
• loyalty-inducing arrangements.

In India, these issues primarily arise under Sections 3 and 4 of the Competition Act, 2002.

In the EU, Articles 101 and 102 TFEU are central, supplemented by the Digital Markets Act for designated gatekeepers.

In the United States, analysis may involve Sections 1 and 2 of the Sherman Act, Section 7 of the Clayton Act, and Section 5 of the FTC Act.

4. Merger and Acquisition Concentration

Concentration may arise when:

• a foundation-model company acquires a knowledge database;
• a search company acquires an AI research platform;
• a cloud provider acquires an AI model developer;
• a dominant platform acquires a downstream AI application;
• a company acquires a potential competitor before it becomes significant.

Traditional market-share analysis can be inadequate because many AI products are supplied at a zero monetary price.

Competition authorities therefore increasingly need to consider:

• data concentration;
• computing capacity;
• user attention;
• technical capabilities;
• access to distribution;
• switching costs;
• ecosystem effects;
• innovation competition;
• potential competition.

5. Data as a Source of Market Power

Data can constitute an important competitive advantage.

A synthetic knowledge platform may possess:

• proprietary research datasets;
• user queries;
• feedback data;
• clickstream information;
• academic databases;
• specialised professional information;
• interaction histories.

A larger user base may generate more data, which improves the AI system, which attracts more users.

This creates a possible data-network-effect feedback loop:

More Users → More Data → Better Model → Better Knowledge Services → More Users

If competitors cannot obtain comparable data, concentration can become self-reinforcing.

However, possession of large quantities of data does not automatically establish dominance. Competition authorities must examine quality, uniqueness, substitutability, access, portability, and the competitive significance of the particular dataset.

6. Vertical Integration

A particularly important issue is vertical integration.

Suppose one enterprise controls:

Cloud → AI Chips → Foundation Model → Knowledge Platform → Search → Distribution

The integrated firm could potentially disadvantage competitors through:

• preferential cloud pricing;
• discriminatory computing access;
• tying cloud services to AI models;
• restricting interoperability;
• preferential placement in search;
• exclusive distribution;
• withholding technical information.

This does not mean vertical integration is inherently anticompetitive. It can also produce efficiencies such as:

• lower costs;
• better security;
• faster innovation;
• improved integration;
• reduced transaction costs.

The competition analysis therefore focuses on whether integration forecloses effective competition.

7. Self-Preferencing

A synthetic knowledge platform integrated into a search or digital ecosystem may prefer its own AI-generated knowledge services.

For example:

Search Platform → Own AI Answer → Own Knowledge Product → Reduced Visibility of Rival Research Services

Possible concerns include:

• preferential ranking;
• reduced traffic to competing knowledge providers;
• preferential access to APIs;
• use of competitor data to improve the platform's own service;
• manipulation of recommendation systems.

The concern becomes stronger where the platform controls an important gateway to users.

8. Exclusive Access to Knowledge and Data

Exclusive agreements may prevent competitors from obtaining essential or strategically important information.

Examples include:

• exclusive licensing of scientific databases;
• exclusive access to publisher content;
• exclusive agreements with universities;
• exclusive access to specialised professional datasets;
• agreements preventing data portability.

Such arrangements may be problematic where they substantially reduce the ability of rivals to compete.

The analysis should distinguish between:

legitimate exclusive licensing
and
exclusivity designed or capable of substantially foreclosing competitors.

9. Interoperability and API Restrictions

Competition can also be affected by restrictions on APIs.

A dominant knowledge platform could:

• limit API functionality;
• impose discriminatory API pricing;
• restrict third-party integration;
• prevent interoperability with competing AI models;
• prohibit data export;
• impose technical restrictions on switching.

These practices may increase switching costs and reinforce ecosystem dependence.

10. Network Effects

Synthetic knowledge platforms may exhibit strong direct and indirect network effects.

For example:

Users → Queries → Training/Feedback → Improved Platform → More Users

At the same time:

Users → Developers → Applications → More Users

Once a platform reaches substantial scale, competitors may struggle to achieve comparable network effects.

This can create a tipping risk, particularly when combined with:

• high switching costs;
• data advantages;
• strong brand recognition;
• proprietary infrastructure;
• default distribution;
• interoperability restrictions.

11. Six Important Case Laws

The following cases are particularly useful for understanding concentration and digital-platform competition. They do not all concern synthetic knowledge platforms specifically; rather, they establish principles that can be applied to AI-driven knowledge ecosystems.

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

Facts

Microsoft was found to have used its dominance in PC operating systems to restrict competing technologies and browsers.

Competition Principle

The case demonstrates how dominance in one technological layer can be leveraged to protect or extend market power into another layer.

Relevance to Synthetic Knowledge Platforms

The same principle can become relevant where a company controls:

• an operating system;
• browser;
• search engine;
• cloud platform;

and then uses that position to favour its own AI knowledge service.

For example, preferential integration of a proprietary AI assistant into a dominant distribution platform could potentially disadvantage independent AI knowledge providers.

Key lesson

Control of an important technological gateway can facilitate extension of market power into adjacent markets.

Case 2: Google Search (Shopping) — European Commission, 2017

Facts

The European Commission found that Google had systematically favoured its comparison-shopping service in search results while placing competing comparison-shopping services at a disadvantage.

Competition Principle

The case is important for self-preferencing.

Relevance

A synthetic knowledge platform incorporated into a dominant search ecosystem could theoretically favour:

• its own AI-generated answers;
• its own research database;
• its own knowledge service;
• its own AI applications.

If rival knowledge providers are systematically demoted or deprived of equivalent access, competition concerns may arise.

Key lesson

A vertically integrated platform may not necessarily be permitted to use control over a critical gateway to favour its own downstream service in a manner that harms competition.

Case 3: Google Android — European Commission, 2018

Facts

The European Commission examined Google's contractual practices concerning Android devices, including restrictions involving search and browser distribution.

Competition Principle

The case illustrates how contractual arrangements across complementary digital products can reinforce an ecosystem's market position.

Relevance

Synthetic knowledge ecosystems may similarly involve:

Operating System + Browser + Search + AI Assistant + Knowledge Platform

Bundling or contractual arrangements could make it difficult for competing AI knowledge providers to obtain distribution.

Key lesson

Competition analysis must consider the ecosystem, rather than examining each digital product entirely in isolation.

Case 4: Google Search (AdSense) — European Commission, 2019

Facts

The Commission examined contractual restrictions that allegedly prevented third-party search providers from accessing important advertising opportunities on publisher websites.

Competition Principle

The case demonstrates the potential competitive significance of contractual restrictions imposed through a dominant digital intermediary.

Relevance

A dominant synthetic knowledge intermediary could potentially use contractual arrangements to restrict:

• rival AI assistants;
• alternative knowledge APIs;
• competing search providers;
• independent knowledge aggregators.

Key lesson

Contractual restrictions can contribute to foreclosure where a dominant intermediary controls an important route to customers.

Case 5: United States v. Google LLC — Search and Search Advertising

Facts

The U.S. Department of Justice challenged Google's practices concerning distribution and default arrangements in search.

Competition Principle

The litigation illustrates the importance of distribution advantages and default status in digital markets.

Relevance

Synthetic knowledge services may depend heavily on distribution.

A company controlling:

• browsers;
• mobile operating systems;
• search;
• app stores;
• enterprise software;

could potentially give its own AI knowledge service preferential access to users.

Key lesson

Market power may be reinforced not only through product superiority but also through control over distribution channels and default positions.

Case 6: FTC v. Meta Platforms, Inc.

Facts

The U.S. Federal Trade Commission challenged Meta's historical acquisitions of Instagram and WhatsApp, arguing that the acquisitions contributed to the preservation of monopoly power in personal social networking.

Competition Principle

The case illustrates the importance of potential competition and acquisitions of emerging competitors.

Relevance to Synthetic Knowledge Platforms

A dominant AI knowledge platform might acquire:

• a promising research AI;
• an emerging knowledge-search engine;
• a specialised scientific model;
• a new AI-agent technology;
• a rapidly growing data platform.

Even where the target's present market share is small, the transaction may warrant examination if the target represents a meaningful source of future competition.

Key lesson

Merger control must sometimes consider future competitive significance, not merely current market shares.

12. Additional Relevant Case: Amazon Marketplace

Competition authorities have increasingly examined the relationship between a platform's intermediary role and its own commercial activities.

The underlying competition concern is particularly relevant to synthetic knowledge platforms because an intermediary may simultaneously:

1. provide infrastructure;
2. collect information about users and competitors;
3. operate a marketplace or distribution channel;
4. offer its own competing service.

The resulting conflict can create incentives for self-preferencing and exploitation of commercially valuable platform information.

13. Synthetic Knowledge Platforms and Essential Facilities

A difficult question is whether certain knowledge resources could qualify as an essential facility.

Potential examples might include:

• uniquely valuable datasets;
• specialised scientific databases;
• indispensable technical interfaces;
• infrastructure necessary for model development.

The essential-facility doctrine traditionally requires a particularly high threshold.

Generally, the claimant must establish factors such as:

1. control by a dominant undertaking;
2. indispensability;
3. inability to reasonably duplicate the facility;
4. absence of a viable alternative;
5. potential elimination of effective competition;
6. feasibility of supplying access.

Therefore, not every large dataset or AI model constitutes an essential facility.

14. Competition Between AI Models

Concentration may occur at the foundation-model level.

Potential competition problems include:

A. Model access

A dominant provider could restrict access to its model through discriminatory API conditions.

B. Compute dependency

Competitors may depend upon the dominant firm's cloud infrastructure.

C. Data dependency

Rivals may lack comparable training datasets.

D. Distribution dependency

AI applications may depend upon dominant search engines, operating systems or app stores.

E. Talent concentration

Acquisition of specialist AI teams can reduce potential competition.

15. Merger Theories of Harm

Competition authorities could examine several theories.

1. Horizontal concentration

Two competing AI knowledge platforms merge.

2. Vertical foreclosure

A cloud provider acquires a foundation-model company and disadvantages rival models.

3. Conglomerate effects

A dominant search platform acquires a knowledge-generation platform and links the products.

4. Data concentration

The transaction combines datasets that competitors cannot replicate.

5. Killer acquisition

An established platform acquires a small but potentially disruptive AI company.

6. Innovation harm

The transaction eliminates an independent technological trajectory.

16. Dynamic Competition and Innovation

Traditional competition analysis often focuses on:

• price;
• output;
• market shares.

Synthetic knowledge markets require greater attention to:

• model quality;
• accuracy;
• latency;
• privacy;
• innovation;
• research capabilities;
• interoperability;
• reliability;
• transparency;
• user choice.

Many AI knowledge services are supplied without direct monetary payment.

Therefore:

Price Competition ≠ Entire Competition Analysis

A platform could theoretically maintain a zero monetary price while reducing competition through lower quality, reduced privacy, diminished innovation, or restricted interoperability.

17. Competition Concerns from User Lock-In

Switching from one synthetic knowledge platform to another may be costly because users have accumulated:

• prompts;
• workflows;
• personalised knowledge bases;
• embeddings;
• documents;
• proprietary configurations;
• enterprise integrations;
• API dependencies.

High switching costs can reduce competitive pressure.

Competition authorities may therefore examine:

• data portability;
• export rights;
• interoperability;
• open APIs;
• migration tools;
• contractual termination conditions.

18. Remedies

Possible remedies include:

Structural remedies

• divestiture;
• separation of business units;
• prohibition of certain acquisitions.

Behavioural remedies

• non-discrimination;
• interoperability obligations;
• API access;
• data portability;
• prohibition of tying;
• restrictions on self-preferencing.

Merger remedies

• divestiture of assets;
• licensing commitments;
• access commitments;
• firewall arrangements;
• restrictions on exclusive agreements.

Regulatory remedies

Digital-market regulation may supplement traditional competition law where ordinary enforcement is too slow or difficult.

19. Indian Competition-Law Perspective

For India, synthetic knowledge-platform concentration can be analysed principally under the Competition Act, 2002.

Relevant issues include:

Section 3

Agreements that cause or are likely to cause an appreciable adverse effect on competition.

Potential examples:

• AI-platform exclusivity;
• restrictive licensing;
• coordinated pricing;
• information exchange;
• restrictive distribution agreements.

Section 4

Abuse of dominant position.

Potential concerns include:

• unfair or discriminatory conditions;
• denial of market access;
• tying;
• leveraging;
• discriminatory API access;
• exploitative contractual terms.

Sections 5 and 6

Combinations may require examination where acquisitions, mergers or amalgamations produce substantial competitive effects.

For digital markets, the assessment should not depend exclusively on traditional turnover or current revenue because rapidly growing AI businesses may possess substantial competitive significance before achieving conventional financial scale.

20. Synthetic Knowledge Platform Concentration: Competition Matrix

21. Key Legal Principles Emerging from the Cases

The six principal cases collectively demonstrate several important competition-law principles:

1. Technological dominance can be leveraged into adjacent markets.
2. Distribution and default arrangements can reinforce market power.
3. Self-preferencing may become significant where a platform controls an important gateway.
4. Vertical contractual restrictions can produce foreclosure effects.
5. Potential competition matters in merger analysis.
6. Digital ecosystems require analysis beyond conventional market-share measurements.

22. Challenges for Competition Authorities

Synthetic knowledge markets present several distinctive challenges:

A. Defining the relevant market

Is the market:

• search?
• AI search?
• knowledge retrieval?
• generative AI?
• professional research?
• scientific information?
• enterprise knowledge management?

The answer may differ according to the particular competitive issue.

B. Measuring market power

Market share alone may not capture:

• data advantages;
• model quality;
• compute access;
• distribution;
• ecosystem control.

C. Rapid technological change

A market that appears concentrated today may evolve quickly.

D. Multi-sided platforms

The same platform may simultaneously serve:

• users;
• publishers;
• developers;
• advertisers;
• enterprises.

E. Zero-price services

Traditional price-based tests become less informative.

23. Conclusion

Synthetic knowledge platform concentration represents a convergence of competition law, AI economics, data governance, and digital-platform regulation.

The central concern is not concentration by itself. Competition analysis must determine whether concentration creates or strengthens the ability to:

• exclude rivals;
• control essential inputs;
• restrict interoperability;
• exploit data advantages;
• favour affiliated services;
• raise switching costs;
• eliminate potential competitors;
• reduce innovation.

The principles developed in United States v. Microsoft, Google Shopping, Google Android, Google AdSense, the U.S. Google Search litigation, and FTC v. Meta provide useful analytical foundations, even though these cases arose in earlier generations of digital markets.