AI-Driven Corporate Scaling and Exponential Concentration Effects

Introduction

AI-driven corporate scaling refers to the ability of firms to use artificial intelligence, automation, data, cloud computing, algorithms, and foundation models to expand output, market reach, decision-making capacity, and customer acquisition without proportionately increasing labour or other traditional costs.

The competition-law concern is that AI can produce exponential concentration effects. A firm that obtains an initial advantage in data, computing capacity, distribution, talent, capital, or users may use AI to convert that advantage into:

More users → more data → better AI → lower costs/better products → more users → greater revenues → more computing and talent → stronger AI.

This is sometimes described as an AI/data/network-effect flywheel. The FTC has specifically warned that generative-AI network effects can reinforce incumbent positions and that AI mergers or partnerships may allow large firms controlling critical inputs to consolidate market power.

Importantly, large scale by itself is not unlawful. Competition law generally becomes concerned when scale is accompanied by exclusionary conduct, acquisition of nascent competitors, discriminatory access to essential inputs, tying/bundling, self-preferencing, exclusive arrangements, or other conduct that protects or extends market power.

1. Meaning of Exponential Concentration

Traditional economies of scale normally mean:

As output increases, average cost decreases.

AI can intensify this into a much stronger feedback mechanism.

Traditional scaling

More production → lower unit cost → greater sales

AI-driven scaling

More users
↓
More data and feedback
↓
Better models and predictions
↓
Better products / lower marginal cost
↓
More users and transactions
↓
More revenue and computing resources
↓
More AI investment
↓
Greater market advantage

The consequence can be non-linear concentration.

A company that is only somewhat larger at the beginning can, under certain conditions, become substantially more difficult to challenge because its advantages reinforce each other.

2. Major Sources of AI-Driven Concentration

A. Data advantages

AI systems can improve with access to large quantities of high-quality data.

A firm with:

• search data;
• consumer behaviour data;
• transaction data;
• location data;
• advertising data;
• industrial data;
• financial data; or
• user-generated content

may be able to improve its AI systems faster than smaller competitors.

The U.S. Department of Justice has specifically described data-driven feedback effects in digital markets: greater consumer exposure can generate more data, which can improve the product, which attracts additional users.

Competition concern

The relevant question is not simply:

"Does the firm possess lots of data?"

It is:

"Does control over that data create an advantage that rivals cannot realistically reproduce, and is that advantage reinforced through exclusionary conduct?"

3. Computing and Infrastructure Concentration

Modern AI requires substantial:

• GPUs;
• AI accelerators;
• cloud infrastructure;
• electricity;
• data centres;
• networking;
• specialised engineering;
• model-training infrastructure.

Consequently, firms controlling computing infrastructure can potentially occupy a powerful position upstream of AI developers.

The FTC's investigation into major cloud/AI partnerships highlighted possible effects involving access to computing resources, engineering talent, switching costs, and access to sensitive technical information.

This creates a possible vertical concentration structure:

Cloud provider → computing → foundation model → AI application → distribution platform → consumer

If one corporate ecosystem controls several layers, rivals may become dependent upon it.

4. Network Effects

AI products can exhibit direct and indirect network effects.

For example:

More users

→ more interactions

→ more feedback

→ more training/evaluation information

→ potentially better product

→ greater user adoption.

The FTC has expressly identified this possibility in generative AI, warning that network effects could allow first movers to establish significant advantages and make entry more difficult.

5. Economies of Scope

AI can also create economies of scope.

A company possessing one large AI infrastructure can use it across:

• search;
• advertising;
• cloud computing;
• smartphones;
• productivity software;
• commerce;
• payments;
• autonomous systems;
• cybersecurity;
• healthcare;
• financial services.

Therefore, the same underlying AI investment can support multiple markets.

This creates a competition-law question concerning leveraging:

Can market power in one AI-related market be transferred into adjacent markets?

6. AI and Acquisitions of Nascent Competitors

One of the most important risks is the acquisition of small AI companies before they become significant competitors.

A dominant company might acquire:

• a promising model developer;
• an AI search startup;
• a specialised AI application;
• an AI safety company;
• an AI infrastructure developer;
• an important data provider.

The acquisition may remove a future competitive threat even though the target has little current revenue.

The FTC has expressly warned that incumbents may acquire nascent AI competitors instead of competing with them and may also acquire complementary applications or critical inputs.

7. Six Major Case Laws

1. United States v. Google LLC — Search Monopoly

Court: U.S. District Court for the District of Columbia
Year: 2024 liability decision

This case is highly relevant to AI-driven concentration because it illustrates the data-distribution-scale feedback loop.

The court found Google liable for unlawfully maintaining monopolies in general search services and general search advertising.

The government's case focused heavily on Google's distribution agreements and its ability to maintain an enormous user base.

The later remedy proceedings also specifically considered how Google's existing search monopoly could interact with the emerging AI ecosystem. The DOJ's remedy materials expressly addressed the possibility of Google using AI to further entrench its existing search position.

Principle

A dominant digital platform may not use exclusionary arrangements to protect an established network advantage and thereby prevent rivals from achieving sufficient scale.

AI relevance

AI search can potentially be integrated into an existing search ecosystem.

Thus:

Search users → search data → better AI/search → more users → stronger distribution advantage.

The case demonstrates why competition authorities may examine AI not as an isolated technology but as an extension of existing platform power.

2. FTC v. Amazon.com, Inc.

Court: U.S. District Court for the Western District of Washington
Filed: 2023

The FTC alleged that Amazon unlawfully maintained monopoly power through a collection of exclusionary practices.

Among the alleged mechanisms were:

• anti-discounting practices;
• manipulation of seller visibility;
• tying Prime eligibility to fulfilment;
• self-preferencing;
• increasing seller dependency;
• raising barriers to competing platforms.

The FTC's amended complaint specifically describes strong economies of scale and network effects, including a feedback loop whereby shoppers attract sellers and sellers increase product selection, which in turn attracts additional shoppers.

Principle

Network effects and economies of scale can make an established platform increasingly difficult to challenge.

AI relevance

AI can intensify the same mechanism through:

• algorithmic recommendations;
• automated pricing;
• personalised search;
• advertising optimisation;
• seller analytics;
• generative shopping assistants.

Therefore, an AI-enhanced marketplace may potentially make an existing network advantage even stronger.

3. United States v. Microsoft Corp.

Court: U.S. District Court for the District of Columbia
Year: 2001

The Microsoft litigation remains a foundational case concerning technology-platform power.

Microsoft was found to have unlawfully maintained its monopoly in the PC operating-system market through exclusionary conduct involving browser distribution and relationships with computer manufacturers and other market participants.

Principle

A dominant technology platform cannot use control over a strategically important platform to exclude or disadvantage emerging competitive threats.

AI relevance

The analogy is particularly important for:

• AI operating systems;
• AI assistants;
• cloud platforms;
• AI-enabled productivity suites;
• default AI applications.

If an AI provider controls a major distribution platform, it may possess the ability to determine which competing AI services receive meaningful access to users.

4. European Commission — Google Android

Case: Google Android
Authority: European Commission
Decision: 2018

The European Commission found that Google imposed contractual restrictions concerning Android devices that restricted competition, including arrangements involving:

• pre-installation;
• search;
• browser applications;
• distribution incentives.

The case illustrates the importance of default status and distribution scale in digital markets.

Principle

A firm possessing substantial platform power cannot necessarily use contractual restrictions to extend that power into neighbouring markets.

AI relevance

The same issue becomes significant where AI assistants compete for default placement on smartphones.

In 2026, the European Commission issued binding DMA measures concerning Google's Android ecosystem, including interoperability for competing AI services and access to search data.

This demonstrates how the competition problem has evolved from traditional search/browser defaults toward AI-assistant access and interoperability.

5. Google Shopping — European Commission

Case: Google Search (Shopping)
Authority: European Commission
Decision: 2017

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

The case is important for the concept of self-preferencing.

Principle

A vertically integrated platform may create competition concerns when it uses control over an important platform or infrastructure to favour its own downstream service.

AI relevance

AI systems increasingly determine:

• which products are recommended;
• which websites are displayed;
• which merchants receive visibility;
• which applications are suggested;
• which answers consumers receive.

Thus, AI-generated recommendations can become a new form of digital gatekeeping.

The EU's 2026 DMA enforcement against Google also addressed self-preferencing in Search, illustrating the continuing importance of this issue.

6. FTC v. Facebook, Inc. / Meta Platforms

Court: U.S. District Court for the District of Columbia
Litigation: 2020 onward

The FTC's case against Facebook concerned alleged maintenance of monopoly power through conduct involving acquisitions and restrictions affecting competitive threats.

The case is relevant to AI-driven concentration because social-network markets demonstrate the importance of:

• network effects;
• user scale;
• data;
• acquisitions;
• ecosystem integration;
• barriers to entry.

Principle

Where a platform's value increases with user participation, achieving sufficient scale can become an important competitive barrier.

AI relevance

Generative AI can combine with social platforms to create a new feedback mechanism:

Users → interactions → data → model improvement → engagement → more users.

This potentially creates a powerful ecosystem advantage when AI is integrated into a large existing platform.

7. United States v. Google — Digital Advertising

A further important body of litigation concerns Google's digital advertising technology stack.

The DOJ alleged that Google unlawfully maintained monopolies in parts of the digital advertising technology ecosystem through conduct involving publisher ad servers, advertiser tools, and ad exchanges. The litigation demonstrates the importance of controlling multiple vertically connected layers of a digital ecosystem.

AI relevance

AI can potentially connect:

data → targeting → advertising → content generation → ranking → measurement

A firm controlling several stages may gain both information advantages and opportunities to favour its own products.

8. Competition-Law Analysis of AI-Driven Scaling

A. Dominance

Large scale is not automatically unlawful.

Competition authorities generally ask whether the undertaking possesses substantial market power.

Relevant factors include:

• market share;
• entry barriers;
• switching costs;
• network effects;
• access to data;
• computing capacity;
• vertical integration;
• distribution control;
• technological advantages;
• customer dependency.

B. Exclusionary Conduct

AI concentration becomes a competition problem where scale is protected through conduct such as:

1. Exclusive dealing

Preventing customers or distributors from using rival AI systems.

2. Tying

Conditioning access to one product upon use of another AI service.

3. Self-preferencing

Using an AI platform to favour the firm's own downstream services.

4. Refusal of access

Restricting rivals' access to:

• data;
• APIs;
• cloud infrastructure;
• interoperability;
• essential interfaces.

5. Predatory conduct

Using enormous financial resources generated by an incumbent ecosystem to sustain below-cost strategies aimed at eliminating competitors.

6. Acquisitions

Purchasing emerging AI competitors before they reach meaningful scale.

9. The AI "Scale Flywheel"

A particularly important analytical model is:

              LARGE USER BASE                    ↓              MORE DATA                    ↓          BETTER AI PERFORMANCE                    ↓          BETTER USER EXPERIENCE                    ↓             MORE USERS                    ↓              MORE REVENUE                    ↓        MORE COMPUTE + TALENT                    ↓          FASTER AI DEVELOPMENT                    ↓             GREATER SCALE                    ↓        STRONGER MARKET POSITION                    ↺

This creates a positive feedback loop.

The competition-law difficulty is that the market may eventually cease to operate primarily through ordinary price competition and instead become characterised by competition over scale itself.

10. Exponential Concentration Through Vertical Integration

AI concentration can occur at multiple layers:

The greatest competition concern arises when a single corporate ecosystem controls several interconnected layers.

For example:

Cloud → Foundation Model → Operating System → AI Assistant → Search → Advertising

may produce significantly greater leverage than control over any single layer.

11. Partnerships and Minority Investments

Concentration does not necessarily require a traditional merger.

AI markets increasingly involve:

• minority investments;
• exclusive partnerships;
• cloud commitments;
• revenue-sharing arrangements;
• licensing;
• preferred access;
• board rights;
• technology integration.

The FTC's 2025 study of partnerships between major cloud providers and AI developers specifically examined equity interests, revenue-sharing arrangements, consultation rights, control rights and exclusivity concerns.

Therefore, competition analysis may need to look beyond conventional M&A.

12. Effects on Market Entry

AI-driven concentration can increase barriers to entry through:

Capital requirements

Training advanced models can require substantial expenditure.

Data requirements

Entrants may lack equivalent datasets.

Compute requirements

Access to advanced computing infrastructure may be constrained.

Talent

Specialised AI researchers and engineers are scarce.

Distribution

Established firms already possess millions or billions of users.

Brand and trust

Consumers may prefer AI services integrated into familiar platforms.

Switching costs

Businesses may become dependent upon:

• APIs;
• cloud infrastructure;
• proprietary model formats;
• enterprise software;
• data pipelines.

13. Effects on Innovation

Concentration can have two different economic effects, and competition analysis must distinguish them.

Potential efficiency effect

Large firms may be able to:

• invest heavily in research;
• develop sophisticated models;
• build large-scale infrastructure;
• reduce costs;
• improve reliability;
• deploy AI globally.

Potential competition effect

Excessive concentration may:

• reduce independent innovation;
• discourage entry;
• increase dependency;
• limit interoperability;
• suppress alternative business models;
• make acquisition of competitors more likely.

Consequently, scale itself should not be treated as inherently anticompetitive.

The crucial question is whether the scale advantage results from superior competition or is being protected through exclusionary mechanisms.

14. Remedies

Possible competition-law remedies include:

Structural remedies

• divestiture;
• separation of vertically integrated businesses;
• restrictions on acquisitions.

Behavioural remedies

• non-discrimination requirements;
• interoperability;
• API access;
• data access;
• prohibition of exclusivity.

Merger remedies

• prohibition of problematic acquisitions;
• licensing commitments;
• access commitments;
• firewalls.

Data-related remedies

• portability;
• interoperability;
• access to certain datasets;
• restrictions on combining datasets.

The Google search remedy proceedings illustrate how modern remedies can include data-access and interoperability measures. In September 2025, the DOJ announced remedies requiring Google to make certain search-index and user-interaction data available to qualifying rivals and restricting certain exclusive distribution arrangements.

15. Key Case-Law Principles

16. Emerging Competition Issues

The most significant future issues are likely to involve:

1. AI foundation-model concentration
2. Cloud–AI vertical integration
3. GPU and compute bottlenecks
4. Acquisition of AI startups
5. AI partnerships and minority investments
6. Data accumulation
7. AI assistant defaults
8. Self-preferencing by AI platforms
9. AI-powered exclusionary pricing
10. Interoperability restrictions
11. API discrimination
12. AI-generated search and recommendation gatekeeping
13. Algorithmic coordination
14. Exclusive access to specialised AI talent
15. Cross-market leveraging by AI ecosystems

Conclusion

AI-driven corporate scaling can transform ordinary economies of scale into powerful cumulative concentration mechanisms. Data, computing, users, capital, talent and distribution can reinforce one another, creating an AI-specific feedback loop in which an initial advantage becomes progressively harder for competitors to overcome.

The principal competition-law lesson from Microsoft, Google, Amazon, Google Shopping, Google Android and Facebook/Meta is that scale is not itself the prohibited conduct. The central legal inquiry is whether a firm uses its scale, platform position, data, infrastructure or ecosystem control to exclude competitors, prevent entry, acquire nascent threats, restrict interoperability, discriminate against rivals, or extend dominance into neighbouring markets.

The emerging AI environment therefore requires competition analysis at three interconnected levels:

Input concentration
→ compute, chips, data, talent

Model concentration
→ foundation models, AI infrastructure, cloud

Distribution concentration
→ operating systems, search, applications, marketplaces and AI assistants.

Where all three layers become concentrated within interconnected corporate ecosystems, the possibility of exponential rather than merely linear concentration becomes particularly significant. The FTC has expressly identified network effects, platform effects, M&A and control of critical AI inputs as competition concerns, while recent Google proceedings in the U.S. and EU demonstrate that regulators are already considering data access, interoperability and AI distribution as part of modern digital-competition enforcement.