Intelligence Inequality as an Economic Structure Driver

1. Introduction

Intelligence inequality as an economic structure driver refers to a situation in which unequal access to intelligence-producing capabilities—such as advanced AI, computational resources, data, expert knowledge, algorithms, predictive systems, and decision-making infrastructure—becomes a structural determinant of economic power.

Traditionally, economic inequality has been explained through differences in:

capital;

land;

labour;

education;

technology;

financial resources.

In an AI-intensive economy, another productive factor becomes increasingly important:

access to superior intelligence and decision-making capacity.

The critical competition-law question is therefore whether unequal access to intelligence merely reflects legitimate differences in innovation and investment, or whether it develops into structural market power that suppresses competition and entrenches economic inequality.

2. Meaning of Intelligence Inequality

Intelligence inequality should be distinguished from differences in human intelligence.

In competition and economic-law analysis, it is better understood as unequal access to economically valuable intelligence infrastructure.

This can include unequal access to:

foundation AI models;

high-performance computing;

GPUs;

proprietary datasets;

predictive analytics;

algorithms;

machine-learning talent;

autonomous agents;

specialised AI chips;

cloud infrastructure;

proprietary knowledge graphs;

real-time market intelligence.

Thus:

Intelligence inequality = unequal capacity to observe, predict, optimise and act in economic markets.

3. From Knowledge Inequality to Intelligence Inequality

Traditional knowledge economies already created disparities between firms with different levels of expertise.

AI changes the scale and speed of that disparity.

A conventional firm may have:

100 analysts producing market forecasts.

An AI-intensive firm may have:

automated systems processing millions of variables continuously.

The difference is not merely quantitative.

AI can create a feedback advantage:

More data

↓

Better intelligence

↓

Better decisions

↓

More customers

↓

More data

↓

Even better intelligence

This can produce cumulative economic advantage.

4. Intelligence as a Productive Asset

Intelligence can affect almost every stage of production.

Input markets

AI can identify:

cheaper suppliers;

optimal logistics;

commodity-price movements.

Production

AI can optimise:

manufacturing;

energy consumption;

inventory;

labour allocation.

Distribution

AI can predict:

consumer demand;

purchasing behaviour;

geographic demand.

Pricing

AI can calculate:

willingness to pay;

elasticity;

competitor responses.

Investment

AI can predict:

financial risk;

market movements;

acquisition targets.

The undertaking possessing superior intelligence may therefore compete more effectively across multiple markets simultaneously.

5. Intelligence Inequality and Market Power

Competition law traditionally measures market power through indicators such as:

market share;

barriers to entry;

financial strength;

vertical integration;

network effects.

AI requires an additional question:

Who possesses the superior ability to generate commercially relevant predictions?

A firm with a comparatively modest current market share may nevertheless possess extraordinary future competitive power because it controls:

unique data;

superior models;

specialised compute;

autonomous decision systems.

6. Data as the Foundation of Intelligence Inequality

Data can function as the raw material of machine intelligence.

Consider two firms:

Firm A

10 million customers;

ten years of behavioural data;

real-time transaction data;

proprietary feedback loops.

Firm B

100,000 customers;

limited historical data;

no real-time information.

Firm A may be able to build substantially more accurate predictive systems.

This creates a potential data-intelligence barrier to entry.

7. Compute Inequality

Data alone is insufficient.

Advanced AI requires:

GPUs;

specialised processors;

cloud infrastructure;

electricity;

networking;

storage.

Consequently, access to computing capacity can itself become a competitive bottleneck.

If only a small number of undertakings control advanced compute infrastructure, smaller competitors may be unable to develop comparable AI capabilities.

This produces:

Compute inequality → AI capability inequality → market-power inequality.

8. Model Inequality

There may also be substantial differences between:

open-source models;

commercial models;

specialised foundation models;

frontier models.

A firm controlling a highly capable foundation model can potentially influence multiple downstream markets.

For example:

Foundation model

↓

AI assistant

↓

Search

↓

Advertising

↓

Commerce

↓

Financial services

The model therefore becomes an economic infrastructure asset.

9. Intelligence as a Bottleneck

The traditional economic bottleneck was often physical:

railways;

ports;

electricity;

telecommunications.

The AI economy can produce cognitive bottlenecks.

Examples include:

dominant AI APIs;

specialised models;

proprietary datasets;

AI agents;

prediction platforms.

A firm controlling such a bottleneck can potentially determine:

who receives access;

at what price;

under what conditions;

with what functionality.

10. Intelligence Inequality and Vertical Integration

Vertical integration can reinforce intelligence inequality.

Consider:

Cloud provider

↓

Foundation model

↓

Enterprise software

↓

AI application

↓

Distribution platform

The same undertaking controls multiple layers.

It can potentially obtain:

data advantages;

compute advantages;

distribution advantages;

cross-subsidisation advantages.

Competition law must therefore examine whether vertical integration produces ecosystem foreclosure.

11. Intelligence and Consumer Dependence

Consumers may become dependent upon AI systems for:

search;

financial decisions;

insurance;

healthcare;

education;

employment;

shopping.

If one AI platform becomes the primary decision-making interface, it can influence not only consumer behaviour but also which suppliers receive consumer demand.

The platform becomes an intermediary between:

economic suppliers and human decision-makers.

That can create considerable structural market power.

12. Intelligence Inequality and Labour Markets

The concept also applies to labour markets.

AI can create differences between:

workers who possess AI tools;

workers who do not;

firms using advanced AI;

firms relying upon conventional processes.

An employer with superior AI can potentially:

screen workers;

predict productivity;

optimise wages;

automate tasks;

monitor performance.

Competition-law concerns may arise where AI systems facilitate:

wage coordination;

no-poach arrangements;

labour-market segmentation;

exclusion of competing workers.

13. Intelligence Inequality and Political Economy

Economic intelligence can also influence political power indirectly.

Large technology firms may possess superior capacity to:

forecast markets;

identify consumer preferences;

model regulatory outcomes;

optimise lobbying strategies;

influence information distribution.

This raises a broader structural concern:

Economic intelligence concentration can become institutional power concentration.

Competition law therefore has an indirect relationship with economic democracy.

14. Intelligence Inequality and Innovation

There are two competing effects.

Positive effect

Large AI firms may have the resources necessary for:

frontier research;

large-scale experimentation;

advanced infrastructure;

high-risk innovation.

Negative effect

Once dominant, those firms may:

acquire emerging competitors;

restrict interoperability;

limit access to data;

control technical standards;

foreclose alternative innovation paths.

Competition policy must therefore distinguish:

innovation leadership

from

innovation foreclosure.

15. Intelligence Inequality and Merger Control

Acquisitions become particularly important.

A dominant AI firm acquiring a small company may not be acquiring current revenue.

It may be acquiring:

specialised researchers;

proprietary data;

model architecture;

intellectual property;

an emerging technology.

Therefore, conventional turnover thresholds may understate competitive significance.

This is closely connected to killer-acquisition theory.

16. Intelligence Inequality and Dynamic Competition

Traditional competition analysis often asks:

"Who is competing today?"

AI requires another question:

"Who could become a competitive threat tomorrow?"

A small AI company may possess:

an innovative model;

superior architecture;

specialist data;

a new agentic technology.

Its present market share may be tiny while its innovation potential is substantial.

17. Intelligence Inequality and Network Effects

AI systems can create powerful network effects.

More users produce:

more queries;

more behavioural data;

more feedback;

more model improvement.

Better models attract more users.

This creates:

Users → data → intelligence → better service → more users.

A platform that reaches sufficient scale may therefore obtain a self-reinforcing advantage.

18. Intelligence Inequality and Switching Costs

Switching can become difficult when consumers or businesses depend upon:

proprietary APIs;

customised models;

historical data;

AI workflows;

agent configurations;

proprietary integrations.

Even where competitors exist, switching may be economically costly.

Thus:

Intelligence infrastructure can produce technological lock-in.

19. Important Case Laws

The following cases provide useful foundations for analysing intelligence inequality, although the concept itself is an emerging economic theory rather than an established standalone legal doctrine.

1. United States v Microsoft

The Microsoft litigation is one of the most important authorities concerning technology, dominance and exclusionary conduct.

Microsoft's control over the operating-system environment created significant opportunities to disadvantage competing technologies.

Relevance

AI ecosystems may produce similar structures:

AI infrastructure → applications → distribution.

Control of the underlying layer can allow an undertaking to influence competition downstream.

2. Google Shopping

The Google Shopping case is highly relevant to data, algorithms and preferential access to consumers.

Relevance

A dominant AI platform could potentially use algorithmic control to favour:

its own services;

affiliated products;

preferred suppliers.

Intelligence inequality therefore becomes especially significant where the undertaking controls the algorithm determining consumer visibility.

3. Google Android

Google Android demonstrates how control over one technological layer can influence competition across adjacent markets.

Relevance

An AI ecosystem controlling:

operating systems;

assistants;

app stores;

search;

cloud services

may similarly leverage intelligence infrastructure into downstream markets.

4. Bronner v Mediaprint

Bronner is important for analysing access to infrastructure controlled by dominant firms.

Relevance

If advanced AI infrastructure becomes indispensable to downstream competitors, questions may arise concerning:

access;

interoperability;

technical interfaces;

discriminatory conditions.

The case also cautions that compulsory access is an exceptional remedy and requires a demanding legal test.

5. Commercial Solvents v Commission

Commercial Solvents established important principles concerning the use of dominance in one market to restrict competition in another.

Relevance

An undertaking dominant in:

AI compute

could potentially leverage that power into:

foundation models or AI applications.

The case therefore provides a useful foundation for analysing cross-layer AI leveraging.

6. Intel v Commission

Intel is important concerning exclusionary incentives and rebates.

Relevance

An AI infrastructure provider could potentially give favourable:

compute prices;

API access;

cloud credits;

model access

to customers who commit to its ecosystem.

Such arrangements could deepen intelligence inequality by making alternative AI providers less viable.

7. Hoffmann-La Roche v Commission

Hoffmann-La Roche is a foundational authority concerning loyalty-inducing practices by dominant undertakings.

Relevance

AI firms could create comparable dependency through:

exclusive AI contracts;

ecosystem discounts;

proprietary integrations;

data incentives.

The result may be reduced opportunities for competing AI suppliers.

8. Qualcomm

The Qualcomm litigation is relevant to technology markets, licensing, royalties and exclusionary strategies.

Relevance

AI technologies increasingly depend upon intellectual-property rights and technical standards.

Control over critical AI patents or licensing arrangements can therefore reinforce intelligence inequality by increasing competitors' entry costs.

9. United Brands

United Brands remains important for understanding the exercise of market power by dominant undertakings.

Relevance

The case provides a general analytical foundation for assessing whether superior economic resources have translated into abusive market behaviour.

In AI markets, superior intelligence alone should not be unlawful. The legal concern arises when that advantage is used in an exclusionary or exploitative manner.

10. MEO

MEO is relevant to discriminatory treatment by dominant firms.

Relevance

A dominant AI infrastructure provider could potentially provide different:

processing speeds;

API limits;

data access;

model capabilities

to similarly situated downstream businesses.

The relevant question would be whether the differentiation produces competitive disadvantage.

20. From Intelligence Advantage to Intelligence Monopoly

The progression can be represented as:

Data advantage

↓

Model advantage

↓

Prediction advantage

↓

Decision advantage

↓

Market-share advantage

↓

More data

↓

Structural intelligence advantage

This is potentially more durable than conventional economies of scale because the system continuously improves itself through market activity.

21. The Concept of an Intelligence Moat

An intelligence moat exists where a firm possesses mutually reinforcing advantages in:

data;

compute;

talent;

models;

users;

distribution;

feedback.

A competitor may theoretically be able to replicate each individual component.

But replicating all components simultaneously may be extremely difficult.

This is why AI competition may involve ecosystem barriers rather than traditional entry barriers.

22. Regulatory Responses

Possible competition-policy responses include:

1. Data portability

Allow users and businesses to transfer relevant data.

2. Interoperability

Require appropriate technical interfaces.

3. Non-discrimination

Prevent unjustified discriminatory access to AI infrastructure.

4. Merger scrutiny

Examine acquisitions involving emerging AI competitors.

5. Access remedies

Where legally justified, require access to critical infrastructure.

6. Algorithmic audits

Examine whether AI systems systematically favour affiliated businesses.

7. Structural separation

In exceptional circumstances, separate infrastructure from downstream commercial operations.

23. Competition Law Should Not Equalise Intelligence

An important limitation must be emphasised.

Competition law should not require every firm to possess identical intelligence capabilities.

Differences can legitimately arise from:

innovation;

investment;

entrepreneurship;

research;

risk-taking;

intellectual property.

Competition law protects the competitive process, not equality of outcomes.

Therefore:

Intelligence inequality is not itself an antitrust violation.

It becomes a competition concern when the inequality is produced or maintained through conduct that unlawfully restricts competitive opportunities.

24. A Proposed Intelligence-Competition Test

Regulators could examine five questions:

Question 1 — Intelligence asset

What scarce intelligence resource does the firm control?

Question 2 — Replicability

Can competitors realistically reproduce it?

Question 3 — Dependency

Do downstream firms depend upon it?

Question 4 — Leveraging

Is the resource being used to expand dominance into adjacent markets?

Question 5 — Foreclosure

Does the conduct materially reduce competitors' ability to compete?

This framework could supplement traditional market-power analysis.

25. Structural Effects

Intelligence inequality can ultimately affect the entire structure of the economy.

A possible hierarchy could emerge:

Compute owners

↓

Foundation-model providers

↓

AI platforms

↓

AI-enabled enterprises

↓

Workers and consumers

Those controlling the upper layers may capture disproportionate economic value.

This produces a form of vertical intelligence concentration.

26. Conclusion

Intelligence inequality is emerging as a potentially important structural feature of AI-intensive economies.

The essential transformation is:

Capital inequality → data inequality → intelligence inequality → market-power inequality.

AI makes superior information processing scalable, continuous and increasingly autonomous. Firms possessing superior data, compute, models and distribution can therefore acquire advantages that reinforce themselves through feedback loops.

The principal competition concerns include:

data concentration;

compute concentration;

foundation-model dominance;

ecosystem lock-in;

vertical leveraging;

algorithmic self-preferencing;

exclusive AI arrangements;

killer acquisitions;

interoperability restrictions;

control over essential AI infrastructure.

The cases of Microsoft, Google Shopping, Google Android, Bronner, Commercial Solvents, Intel, Hoffmann-La Roche, Qualcomm, United Brands and MEO provide important doctrinal building blocks.

The central principle is therefore:

Competition law should not attempt to eliminate differences in intelligence capability; it should prevent dominant control over intelligence infrastructure from being converted into unlawful exclusion, foreclosure or exploitation.

In an AI-driven economy, the decisive competition question may increasingly become not merely who owns the most capital, but who possesses the greatest capacity to know, predict, optimise and act—and whether competitors can realistically acquire equivalent capabilities.