1. Introduction

The development of artificial intelligence (AI) economies creates a major institutional challenge for competition authorities. Traditional competition authorities were largely designed around identifiable markets, relatively stable products, observable prices, and conventional forms of market power.

AI markets can operate differently.

Competitive power may arise from control over:

• foundation models;
• computing infrastructure;
• advanced GPUs and AI accelerators;
• cloud platforms;
• training data;
• proprietary datasets;
• AI researchers and engineers;
• application programming interfaces (APIs);
• model distribution;
• app stores;
• agentic AI ecosystems;
• cloud–model partnerships;
• semiconductor supply chains;
• venture financing;
• technical standards.

Consequently, institutional redesign for competition authorities in AI economies means adapting competition institutions so that they can detect, investigate and remedy competition problems arising from rapidly changing AI ecosystems.

The objective is not to create special protection for particular competitors. It is to ensure that competition authorities have sufficient expertise, investigative powers, technological capabilities, merger-review mechanisms and remedial tools to preserve competitive markets.

2. Why AI Economies Require Institutional Redesign

Traditional competition enforcement often relies on:

1. defining the relevant market;
2. measuring market shares;
3. identifying dominance;
4. examining conduct;
5. assessing effects;
6. imposing remedies.

AI can complicate each stage.

For example, a company may have relatively little revenue from an AI model while possessing enormous strategic importance because it controls:

• critical computing resources;
• a highly valuable model;
• unique training data;
• developer access;
• an ecosystem of applications.

Similarly, an AI company may depend upon another undertaking for cloud computing and chips.

The competitive structure can therefore resemble:

chips → cloud → compute → foundation model → API → applications → distribution → users

rather than a conventional single-product market.

3. Meaning of Institutional Redesign

Institutional redesign involves changing the capacity, organization, procedures and analytical tools of competition authorities.

It can include:

A. Specialized AI divisions

Authorities may require teams specializing in:

• machine learning;
• algorithms;
• cloud computing;
• semiconductors;
• data economics;
• AI safety;
• technical standards.

B. Enhanced merger review

Authorities may need to examine acquisitions of:

• start-ups;
• model developers;
• AI research teams;
• data companies;
• compute providers.

C. Algorithmic investigation capabilities

Authorities need the ability to investigate:

• pricing algorithms;
• recommendation systems;
• ranking algorithms;
• automated contracting;
• AI agents.

D. Technical auditing

Authorities may need to inspect technological systems rather than rely exclusively on documentary evidence.

E. Continuous market monitoring

AI markets can change faster than conventional regulatory cycles.

4. AI Competition Is Multi-Layered

An AI economy can be divided into several layers:

A competition authority designed only around conventional product markets may overlook interactions between these layers.

5. AI Ecosystems and Vertical Integration

AI markets frequently involve vertical relationships.

For example:

chip manufacturer → cloud provider → AI model developer → AI application → enterprise customer.

Vertical integration can generate legitimate efficiencies.

It can also create foreclosure risks.

A vertically integrated undertaking might potentially:

• prioritize its own models;
• discriminate against rival models;
• restrict access to computing resources;
• bundle cloud and AI services;
• impose exclusive arrangements;
• restrict interoperability.

Institutional redesign therefore requires authorities to develop vertical-foreclosure expertise specific to AI ecosystems.

6. Case Law 1 — United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)

The Microsoft litigation is a foundational authority for understanding competition in technology ecosystems.

Microsoft possessed substantial power in PC operating systems and was accused of using exclusionary strategies against technologies that could threaten its position.

The court examined Microsoft's treatment of emerging technological threats, including browser competition.

Institutional relevance

The case demonstrates why competition authorities need the ability to examine technological evolution, not merely current market shares.

In AI economies, analogous questions may concern whether an incumbent uses control over:

• operating systems;
• cloud services;
• app stores;
• APIs;
• data;

to suppress emerging AI technologies.

Institutional lesson

Competition authorities require expertise capable of identifying future technological constraints before they become conventional market competitors.

7. Case Law 2 — Microsoft Corp. v Commission, Case T-201/04

The European Microsoft case involved interoperability information and the integration of Windows Media Player into Windows.

The General Court substantially upheld the Commission's findings.

AI relevance

AI ecosystems similarly depend upon interoperability between:

• models;
• applications;
• cloud platforms;
• APIs;
• operating systems;
• data systems.

A competition authority investigating AI therefore needs technical capacity to determine whether an interoperability restriction is:

• legitimate;
• technically necessary;
• discriminatory;
• exclusionary.

Institutional lesson

Legal investigators alone may not be sufficient. Competition authorities need technical engineers and software specialists who can understand system architecture.

8. Case Law 3 — Intel Corp. v Commission, Case C-413/14 P

Intel concerned exclusionary rebates offered by Intel to computer manufacturers and a retailer.

The Court of Justice emphasized the need for careful assessment of whether the conduct was capable of restricting competition.

AI relevance

Similar economic analysis may be necessary where AI infrastructure providers use:

• rebates;
• cloud credits;
• exclusivity agreements;
• capacity discounts;
• preferential pricing.

For example, a dominant AI infrastructure provider could potentially use commercial incentives to encourage customers to remain exclusively within its ecosystem.

Institutional lesson

Authorities require sophisticated effects-based economic analysis rather than assuming that every restrictive commercial practice has identical effects.

9. Case Law 4 — Qualcomm Inc. v FTC

Qualcomm's litigation concerning modem chipsets and licensing illustrates the difficulty of analysing competition in highly technical markets.

The Ninth Circuit ultimately rejected the FTC's principal antitrust theory.

Institutional significance

The case demonstrates that competition authorities must understand:

• technological architecture;
• intellectual property;
• licensing models;
• supply chains;
• innovation incentives.

These requirements become even more significant in AI because technical systems are frequently built through multiple vertically connected layers.

Institutional lesson

AI enforcement requires authorities to combine legal, economic and engineering expertise.