AI-Driven Corporate Scaling and Exponential Concentration Effects

Introduction

AI-driven corporate scaling refers to the use of artificial intelligence, machine learning, automation, data analytics, algorithmic decision-making and AI-enabled infrastructure to expand a firm's output, customer base, geographic reach and market power without a proportionate increase in employees, physical assets or transaction costs.

The competition-law concern arises when AI creates exponential rather than linear scaling. A firm with superior data, computing infrastructure, algorithms, distribution networks and capital can continuously improve its products, attract more users, collect more data and reinvest the resulting advantages into further expansion. This can produce a reinforcing cycle:

AI capability → lower marginal cost → greater scale → more data/users → better AI → stronger network effects → higher entry barriers → greater concentration.

The issue is not that corporate growth itself is unlawful. Competition law becomes relevant when scaling produces or reinforces market power, facilitates exclusionary conduct, creates anticompetitive concentrations, or allows a dominant undertaking to extend its power into adjacent markets.

China's current platform-competition framework expressly recognizes risks involving data, algorithms, platform rules, dynamic pricing and coordinated behaviour. SAMR's 2026 Internet Platform Antitrust Compliance Guidelines specifically address the use of AI and algorithms in horizontal coordination, resale-price restrictions and other potentially anticompetitive conduct.

I. Meaning of AI-Driven Corporate Scaling

Traditional corporate expansion generally requires proportional increases in:

• employees;
• factories and offices;
• distribution infrastructure;
• sales personnel;
• physical capital;
• managerial capacity.

AI changes this relationship.

A highly automated company may serve millions of additional customers through:

• AI customer service;
• automated content generation;
• algorithmic pricing;
• automated logistics;
• predictive marketing;
• AI-powered coding;
• automated compliance;
• cloud infrastructure;
• autonomous decision systems;
• machine-learning recommendation engines.

Consequently, the marginal cost of serving an additional user can fall dramatically.

This creates the possibility of what may be called an AI scale advantage.

II. The Exponential Concentration Mechanism

The most important competition-law phenomenon is the interaction between several reinforcing advantages.

1. Data advantage

A larger company may obtain more:

• consumer data;
• transaction data;
• behavioural data;
• search data;
• product-performance information;
• feedback data.

That data can improve its AI systems.

2. Algorithmic advantage

Better algorithms can produce:

• better recommendations;
• better predictions;
• lower costs;
• more accurate advertising;
• better fraud detection;
• improved logistics.

3. Network effects

More users can make the platform more valuable.

For example:

More users → more sellers → more products → better recommendations → more users.

4. Economies of scale

AI infrastructure often involves large fixed investments but relatively low marginal costs.

Therefore:

Average cost falls as output increases.

5. Economies of scope

The same AI infrastructure may be deployed across multiple markets.

A company possessing:

• cloud infrastructure,
• payment systems,
• advertising technology,
• consumer data,
• logistics,

can use the same technological ecosystem across numerous sectors.

6. Capital reinforcement

Higher revenues provide resources for:

• acquiring competitors;
• acquiring startups;
• hiring specialised AI personnel;
• purchasing computing capacity;
• developing proprietary models.

Thus, AI scaling can become self-reinforcing.

III. Competition-Law Concern

The fundamental question is not:

"Is the company large?"

Instead, the relevant questions are:

1. How did the company obtain its market position?
2. Can competitors realistically enter or expand?
3. Does the company control an essential input?
4. Does it use its position in one market to exclude rivals in another?
5. Does AI facilitate exclusionary conduct?
6. Are acquisitions removing potential future competitors?
7. Are network effects and switching costs making market power durable?

IV. Principal Competition Risks

A. Data-driven entry barriers

An incumbent may possess datasets unavailable to new entrants.

A new competitor therefore faces a double problem:

No users → little data → weaker AI → inferior product → fewer users.

This can make entry difficult even where the underlying technology is theoretically available.

B. AI infrastructure concentration

Advanced AI systems require:

• computing power;
• cloud infrastructure;
• specialised chips;
• engineering talent;
• data;
• model-development capabilities.

If these inputs are concentrated among a few corporations, downstream AI competitors may become dependent upon incumbent firms.

The FTC's study of major cloud providers' partnerships and investments in generative-AI companies identified potential competition concerns involving access to computing resources and engineering talent, increased switching costs and access to commercially sensitive information.

C. Killer acquisitions and potential competition

AI startups can initially have:

• small revenues;
• few employees;
• limited market share.

Yet they may possess valuable technology or represent a future competitive threat.

Acquiring such a company may eliminate a potential competitor before it becomes a significant rival.

This makes conventional turnover-based merger thresholds potentially less informative in rapidly developing AI markets.

D. Ecosystem expansion

A large platform may use AI to enter adjacent markets.

For example:

Search → advertising → cloud → AI assistants → payments → commerce.

The competition concern arises if an incumbent uses advantages from its existing ecosystem to foreclose competitors in the new market.

V. Relevant Case Laws

The following cases illustrate the legal principles applicable to AI-driven scaling even where the underlying conduct did not necessarily involve modern generative AI.

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

Microsoft used its dominant position in operating systems while imposing restrictions affecting competing browsers.

The case is important for AI because it demonstrates that a firm possessing an entrenched technological platform can potentially use that platform to restrict competitive threats in adjacent markets.

Relevance to AI scaling

An AI ecosystem may similarly involve:

dominant infrastructure → adjacent AI product → distribution advantage → exclusionary conduct.

The Microsoft precedent therefore provides an important framework for examining whether technological integration is legitimate innovation or exclusionary leveraging.

2. United States v. Google LLC — D.D.C. 2024, remedies proceedings continuing thereafter

The Google search litigation concerns the maintenance of dominance through distribution arrangements and practices affecting access to search users.

The underlying principle is particularly relevant to AI ecosystems because distribution can be as important as technological superiority.

An AI service may technically compete with an incumbent but remain disadvantaged if:

• the incumbent controls the operating system;
• default settings favour its AI product;
• browsers favour its assistant;
• app stores favour its services.

Thus, control over distribution can reinforce technological concentration.

3. Google Shopping — European Commission, Case AT.39740

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

The principle is highly relevant to AI-generated search and recommendation systems.

An AI platform controlling the interface through which consumers discover information may potentially favour:

• its own products;
• affiliated services;
• proprietary AI applications;
• preferred commercial partners.

The case therefore illustrates the importance of self-preferencing and algorithmic ranking.

4. Google Android — European Commission, Case AT.40099

The European Commission's Android decision examined Google's practices concerning mobile operating systems, search and application distribution.

The case demonstrates how several complementary products can be combined into an ecosystem capable of reinforcing an incumbent's position.

AI relevance

AI companies may similarly integrate:

• operating systems;
• cloud;
• app stores;
• search;
• assistants;
• advertising;
• foundation models.

The competition issue becomes whether integration creates efficiencies or makes competing AI providers unable to obtain effective access to consumers.

5. Meta Platforms, Inc. v. FTC

The FTC's litigation concerning Meta's acquisitions of Instagram and WhatsApp illustrates the importance of potential competition and acquisitions in rapidly evolving digital markets.

The broader competition-law lesson is that the significance of an acquisition cannot always be assessed solely through the target's present market share.

AI relevance

An AI startup may currently have:

• minimal revenue;
• limited users;
• no significant market share,

while nevertheless possessing technology capable of becoming an important competitive constraint.

Consequently, acquisition analysis may need to consider:

• innovation competition;
• pipeline products;
• technological capabilities;
• future expansion;
• access to data;
• talent.

6. FTC v. Amazon.com Inc.

The FTC's antitrust action against Amazon concerns alleged practices relating to Amazon's marketplace and its relationships with sellers and competitors.

The case is significant for understanding how a large platform can simultaneously act as:

1. infrastructure provider;
2. marketplace operator;
3. competitor to marketplace participants.

AI relevance

AI intensifies this structural problem.

A platform controlling:

• consumer data;
• ranking algorithms;
• advertising;
• AI recommendation systems;

may possess information that its own downstream competitors do not.

This raises concerns about information asymmetry and vertical leveraging.

7. FTC v. Illumina, Inc. — Illumina/GRAIL

The Illumina/GRAIL litigation concerned a vertically related acquisition involving genomic sequencing technology and cancer-detection testing.

The case demonstrates the importance of analysing whether an acquisition can harm competition even where the parties are not straightforward horizontal competitors.

AI relevance

The same reasoning can apply to AI infrastructure.

For example:

AI infrastructure provider + downstream AI application

may create incentives and capabilities to restrict competitors' access to essential inputs.

8. FTC v. Qualcomm Inc. — 969 F.3d 974 (9th Cir. 2020)

The Qualcomm litigation concerned licensing practices and competition in cellular technology.

Although not an AI case, it is useful for examining the relationship between:

• technological standards;
• intellectual property;
• licensing;
• market power;
• downstream competition.

AI markets increasingly involve proprietary models, interfaces, patents and technical standards.

Therefore, control over technological inputs can become an important source of competitive leverage.

VI. Chinese Competition-Law Developments

China provides particularly important examples because its antitrust authorities have increasingly focused on platform ecosystems and algorithmic control.

1. Alibaba "Two Choices" Case

SAMR found Alibaba had abused its dominant position by requiring merchants to choose between Alibaba's platform and competing platforms.

The authority imposed a substantial administrative penalty and required rectification.

The case demonstrates how platform dominance can be reinforced through exclusionary contractual and technological mechanisms.

2. Meituan "Two Choices" Case

SAMR found Meituan had used its dominant position in online food-delivery platform services to impose exclusive arrangements.

The authority stated that Meituan used differential treatment, delayed merchant onboarding, deposits, data and algorithmic measures to support the exclusivity mechanism.

AI-scaling relevance

This illustrates a crucial point:

Algorithmic control can transform contractual power into scalable market power.

Instead of manually monitoring thousands of merchants, a platform can potentially automate:

• ranking;
• traffic allocation;
• penalties;
• pricing;
• visibility;
• merchant classification.

That substantially increases the reach of exclusionary strategies.

3. Tencent/China Literature–New Classics Media Concentration

SAMR identified Tencent-controlled China Literature's acquisition of New Classics Media as one of the transactions that had been implemented without prior notification.

The transaction illustrates the importance of merger control where large digital ecosystems expand into adjacent markets.

4. Alibaba Investment–Intime Retail Concentration

Alibaba's acquisition of control over Intime was also identified by SAMR as a previously unreported concentration.

SAMR explained that Alibaba Investment ultimately obtained control through successive acquisitions.

The case demonstrates how corporate scaling can occur through incremental acquisitions, rather than one immediately transformative transaction.

This is particularly relevant to AI because a major corporation can potentially acquire several small AI businesses over time.

5. Hive Box–China Post Smart Parcel Locker Concentration

SAMR also penalised the failure to notify the Hive Box acquisition of China Post Smart Parcel Locker.

Both businesses operated smart parcel-locker services.

The case demonstrates how consolidation can occur in technologically mediated infrastructure markets even where the businesses may appear relatively specialised.

6. Huya–DouYu Merger

China's prohibition of the Huya/DouYu concentration is particularly important.

SAMR described the transaction as the first prohibited concentration in China's platform economy and concluded that the transaction could eliminate or restrict competition in relevant markets.

The case is important because it demonstrates that platform scale itself can become a merger-control concern, particularly where network effects and market concentration reinforce each other.

VII. AI Makes Existing Competition Problems More Powerful

The distinctive feature of AI is not necessarily that it creates entirely new competition-law doctrines.

Rather, AI can amplify existing mechanisms of market power.

VIII. The "Exponential Concentration" Feedback Loop

A useful analytical model is:

Stage 1 — Initial advantage

A firm has superior:

• capital;
• data;
• computing;
• technology.

Stage 2 — Lower costs

AI reduces:

• labour requirements;
• transaction costs;
• customer-service costs;
• production costs.

Stage 3 — Greater market share

Lower costs allow the firm to:

• reduce prices;
• increase output;
• improve services;
• expand geographically.

Stage 4 — More data

More customers generate more information.

Stage 5 — Better AI

Additional information improves algorithms and models.

Stage 6 — Stronger ecosystem

The company can cross-sell:

• cloud;
• advertising;
• payments;
• software;
• AI services.

Stage 7 — Higher entry barriers

Competitors face:

• insufficient data;
• higher infrastructure costs;
• limited distribution;
• switching costs;
• lack of specialised talent.

Stage 8 — Further concentration

The original advantage becomes self-reinforcing.

Thus:

AI can transform ordinary economies of scale into cumulative economies of scale.

IX. AI and Merger Control

AI-driven corporate scaling makes merger control especially significant.

Authorities may need to consider:

1. Killer acquisitions

Whether a dominant company is systematically acquiring emerging competitors.

2. Talent acquisitions

An acquisition may provide access to:

• engineers;
• researchers;
• model developers.

3. Data acquisitions

The target may possess strategically valuable datasets.

4. Compute acquisitions

An acquisition may provide control over scarce computing resources.

5. Distribution acquisitions

A large platform may acquire an AI company primarily to control its distribution.

6. Vertical foreclosure

An infrastructure provider may acquire downstream AI applications and disadvantage rival developers.

X. AI Partnerships Can Also Produce Concentration

Concentration does not necessarily arise through outright acquisitions.

It can emerge through:

• exclusive cloud agreements;
• equity investments;
• revenue-sharing arrangements;
• long-term computing commitments;
• licensing agreements;
• preferential access arrangements.

The FTC's AI partnerships study specifically examined equity and revenue-sharing rights, consultation/control rights and exclusivity arrangements involving major cloud providers and AI developers.

Therefore, competition analysis should examine economic control, not merely formal ownership.

XI. Current EU Development

The European Union's Digital Markets Act provides an important additional framework.

The European Commission designated Alphabet, Amazon, Apple, ByteDance, Meta and Microsoft as gatekeepers in 2023.

In June 2026, the Commission announced preliminary views that AWS and Microsoft Azure should also be designated as gatekeepers for cloud computing services, noting entrenched positions, switching costs, ecosystem effects and the growing importance of AI tools and partnerships in cloud procurement.

In July 2026, the Commission announced two Google DMA non-compliance decisions concerning self-preferencing in Search and restrictions on steering users to alternative purchasing channels.

These developments illustrate a broader regulatory shift toward examining ecosystem power and gatekeeping, rather than focusing exclusively on traditional single-market market shares.

XII. China: AI-Specific Regulatory Direction

China's regulatory framework is also becoming increasingly explicit about AI and algorithms.

SAMR's 2026 Internet Platform Antitrust Compliance Guidelines identify risks from:

• AI-assisted exchange of competitively sensitive information;
• algorithmic coordination;
• dynamic pricing;
• data-based coordination;
• algorithmic allocation of traffic;
• automated resale-price restrictions. 

In addition, SAMR published AI-related unfair-competition cases in 2026, showing that Chinese enforcement is addressing AI not merely as a technological issue but also through competition and market-order regulation.

XIII. Remedies

Competition authorities may employ several remedies.

Structural remedies

• divestiture;
• prohibition of mergers;
• separation of business units.

Behavioural remedies

• non-discrimination;
• interoperability;
• data-access obligations;
• restrictions on exclusivity;
• transparent ranking;
• fair access to APIs.

Merger remedies

• licensing;
• firewall arrangements;
• continued access to infrastructure;
• prohibition of discriminatory treatment.

Regulatory remedies

• algorithmic auditing;
• reporting requirements;
• competition compliance programmes.

XIV. Corporate Compliance Framework

AI-intensive corporations should therefore establish an AI Competition Compliance Programme.

It should examine:

1. Market power — whether AI has materially strengthened the firm's position.
2. Data access — whether competitors can reasonably obtain essential data.
3. Algorithmic neutrality — whether ranking systems favour affiliated businesses.
4. Acquisitions — whether startup acquisitions eliminate potential competition.
5. Exclusivity — whether cloud, distribution or data contracts foreclose rivals.
6. Interoperability — whether competitors can technically connect to the ecosystem.
7. Pricing algorithms — whether AI facilitates coordination.
8. Sensitive information — whether algorithms exchange competitors' strategic information.
9. Vertical integration — whether upstream infrastructure is being used to disadvantage downstream competitors.
10. AI partnerships — whether investment arrangements effectively create control without formal ownership.

XV. Overall Legal Principle

The central competition-law principle can be expressed as follows:

AI-driven growth is generally compatible with competition law when it results from innovation, efficiency and legitimate economies of scale; the competition concern arises when accumulated AI advantages become mechanisms for durable exclusion, foreclosure, coordinated conduct or anticompetitive concentration.

The distinction between efficient scaling and anticompetitive concentration is therefore critical.

Efficient scaling

Innovation → lower costs → better products → consumer benefit → legitimate expansion

Potentially problematic scaling

Data advantage → network effects → exclusion → acquisition of potential rivals → higher entry barriers → durable market power

Conclusion

AI-driven corporate scaling can fundamentally alter the relationship between firm size and market power. Traditional economies of scale can be amplified by data feedback loops, network effects, cloud infrastructure, automated decision-making and AI-enabled distribution.

The major competition-law risks are therefore not limited to conventional monopoly pricing. They include:

• data-based barriers to entry;
• AI infrastructure dependence;
• algorithmic exclusion;
• self-preferencing;
• platform leveraging;
• killer acquisitions;
• talent and technology acquisitions;
• exclusive AI/cloud partnerships;
• algorithmic coordination;
• ecosystem foreclosure; and
• exponential concentration of economic power.

The Microsoft, Google, Meta, Amazon, Illumina, Qualcomm, Alibaba, Meituan and Huya–DouYu matters collectively demonstrate the underlying legal principles. The emerging AI environment does not necessarily require abandoning established competition-law doctrines; instead, it requires applying those doctrines to data, algorithms, computing infrastructure, ecosystems and rapidly scaling AI firms.