AI-Directed M&A Ecosystems and Continuous Consolidation Markets

AI-DIRECTED M&A ECOSYSTEMS AND CONTINUOUS CONSOLIDATION MARKETS

1. Introduction

Artificial intelligence is increasingly capable of influencing not only the operation of businesses but also their merger and acquisition strategies. An AI-directed M&A ecosystem may use algorithms, machine learning, predictive analytics, automated valuation systems, competitive-intelligence tools and large datasets to identify acquisition targets, predict competitive threats, assess synergies and determine the timing or structure of acquisitions.

The competition-law concern becomes more significant where acquisitions occur continuously rather than as a single transaction. A dominant undertaking may repeatedly acquire startups, data providers, software companies, infrastructure providers, complementary applications or emerging competitors. Each individual transaction may appear relatively small, while the cumulative effect may be to eliminate potential competitors, increase control over data or compute resources, create interoperability dependencies and raise barriers to entry.

The legal issue is therefore not simply:

"Is this particular AI-related acquisition anticompetitive?"

It may also be:

"Does a pattern of acquisitions progressively transform an open competitive ecosystem into a concentrated market controlled by one undertaking or a small group of undertakings?"

Modern merger enforcement increasingly considers innovation, ecosystems, vertical foreclosure, potential competition, access to critical inputs and dynamic competitive effects. The European Commission's 2026 review of its merger guidelines expressly recognizes major changes arising from digitalisation, globalisation and other structural developments.

2. Meaning of AI-Directed M&A

AI-directed M&A refers to situations in which AI systems materially assist or influence:

1. target identification;
2. acquisition screening;
3. valuation;
4. competitive-threat prediction;
5. due diligence;
6. market mapping;
7. customer and data analysis;
8. innovation forecasting;
9. post-acquisition integration;
10. acquisition sequencing;
11. portfolio optimization; and
12. decisions concerning whether a potential competitor should be acquired, partnered with or otherwise neutralised.

AI therefore becomes part of the decision architecture of corporate consolidation.

The technology itself is not unlawful. The competition-law concern arises where the resulting acquisition strategy substantially lessens competition or facilitates exclusionary conduct.

3. Continuous Consolidation

Continuous consolidation occurs where an undertaking repeatedly undertakes acquisitions over time.

For example:

AI infrastructure company

↓

Acquires data company

↓

Acquires model-development company

↓

Acquires cloud orchestration company

↓

Acquires AI application company

↓

Acquires distribution platform

↓

Acquires monitoring/certification company

↓

Creates vertically and horizontally integrated AI ecosystem

The individual transactions may involve different relevant markets.

However, cumulatively they may provide control over:

• data;
• computing capacity;
• foundation models;
• application programming interfaces;
• distribution;
• cloud infrastructure;
• talent;
• customer relationships;
• intellectual property;
• interoperability standards; and
• downstream applications.

This creates a cumulative-concentration problem.

4. Why AI Makes Continuous M&A Particularly Significant

A. Data accumulation

AI businesses depend heavily upon data.

Acquiring multiple firms can therefore create a progressively larger data advantage.

An acquisition can provide:

• proprietary datasets;
• behavioural information;
• customer histories;
• transaction data;
• training data;
• feedback loops;
• proprietary labels; and
• usage information.

The competitive advantage may increase with each acquisition.

B. Network effects

AI platforms may benefit from:

• more users;
• more data;
• better models;
• more developers;
• more applications; and
• more integrations.

This may create a feedback loop:

Users → Data → Better AI → More Users → More Data

An acquisition can accelerate that feedback loop.

C. Talent acquisition

A technology acquisition can effectively transfer:

• engineers;
• researchers;
• patents;
• technical know-how;
• founders;
• customer relationships; and
• specialised AI teams.

Repeated acquisitions can therefore reduce the pool of independent innovators.

D. Elimination of future competitors

The acquired company may not currently compete directly with the acquirer.

It may nevertheless become a future competitor.

This creates the potential-competition problem.

A startup developing an emerging AI architecture may be insignificant today but strategically important several years later.

5. Killer Acquisitions and AI Markets

A "killer acquisition" generally refers to an acquisition in which an incumbent purchases an emerging competitor or innovation in circumstances where the acquisition may eliminate a future competitive constraint.

The concern is particularly relevant to AI because innovation cycles can be extremely rapid.

A large platform may acquire:

• a promising model developer;
• an AI search company;
• an AI cybersecurity firm;
• an autonomous-agent company;
• a specialised medical-AI developer;
• an AI-chip designer; or
• a data-rich application.

The target's current revenue may be modest even though its future competitive significance is substantial.

Consequently, conventional turnover-based merger thresholds may sometimes fail to capture the strategic importance of the transaction.

6. AI-Based Target Selection and Competitive Intelligence

An acquirer may deploy AI to analyse thousands of firms and identify which startups present the greatest competitive threat.

The system may analyse:

• patent portfolios;
• hiring patterns;
• developer activity;
• venture financing;
• customer migration;
• product releases;
• GitHub activity;
• pricing;
• model performance;
• cloud consumption;
• user growth; and
• technological roadmaps.

This creates a new competition-law question:

Can an algorithmically optimised acquisition programme systematically remove emerging competitive threats?

Competition authorities may therefore need to examine not merely the individual transaction but the strategic pattern of acquisitions.

7. Serial Acquisitions

Serial acquisitions occur when one undertaking repeatedly purchases businesses in related markets.

They can produce:

Horizontal concentration

Competitors are successively acquired.

Vertical integration

Suppliers, infrastructure providers and downstream distributors are acquired.

Conglomerate expansion

Companies operating in adjacent markets are incorporated into a single ecosystem.

Data concentration

Different datasets become available to one firm.

Ecosystem concentration

Multiple complementary products become controlled by one platform.

8. The "Portfolio Effect"

A major competition concern is that each acquisition can increase the value of the next acquisition.

For example:

Acquisition 1: AI model

Acquisition 2: AI data provider

Acquisition 3: cloud infrastructure

Acquisition 4: AI application

Acquisition 5: distribution platform

=

Integrated AI ecosystem

The combined ecosystem may possess competitive advantages that none of the acquired companies individually possessed.

This can create an acquisition-driven cumulative advantage.

9. Relevant Competition-Law Theories

A. Substantial lessening of competition

The principal concern is whether the transaction substantially reduces competitive constraints.

B. Elimination of potential competition

An incumbent may acquire a firm that could otherwise become a meaningful competitor.

C. Innovation foreclosure

The acquisition may reduce incentives to develop competing technology.

D. Input foreclosure

An integrated firm may deny competitors access to:

• data;
• APIs;
• computing resources;
• technical infrastructure;
• models;
• distribution channels; or
• essential interfaces.

E. Customer foreclosure

A dominant ecosystem may steer customers toward its own acquired products.

F. Data foreclosure

A merged undertaking may gain exclusive access to datasets necessary for effective AI development.

G. Interoperability foreclosure

An integrated ecosystem may make interoperability more difficult for independent competitors.

H. Raising rivals' costs

Control of infrastructure or critical inputs can increase competitors' costs.

10. Case Law 1 — Illumina/GRAIL

FTC v. Illumina, Inc. / GRAIL, Inc.; In the Matter of Illumina and GRAIL

This is one of the most significant modern merger cases concerning innovation and vertical integration.

Illumina supplied DNA sequencing technology while GRAIL developed multi-cancer early-detection testing. The FTC alleged that Illumina's acquisition could diminish competition and innovation in the emerging cancer-detection market.

The FTC ultimately ordered divestiture, and the Fifth Circuit found substantial evidence supporting the Commission's determination, although it remanded on a separate issue concerning the treatment of Illumina's rebuttal evidence. Illumina subsequently announced that it would divest GRAIL.

Relevance to AI-directed M&A

The case demonstrates that a competition authority may examine an acquisition involving an emerging technology even where the parties are not conventional horizontal competitors.

For AI markets, the analogous situation could involve:

AI infrastructure provider + emerging AI application

or

foundation-model company + specialised AI developer.

The central concern becomes whether control of an upstream technological input can disadvantage downstream innovation.

Principle

A vertical acquisition can raise serious competition concerns where the acquiring firm has the ability and incentive to disadvantage downstream rivals.

11. Case Law 2 — FTC v. Meta/Within

FTC v. Meta Platforms, Inc. / Within Unlimited, Inc.

Meta sought to acquire Within, developer of the VR fitness application Supernatural.

The FTC alleged that the transaction could reduce competition and innovation in markets for VR fitness applications. The FTC specifically characterised Meta as having significant positions in VR hardware, an app store and VR applications, while Within was an important developer of VR fitness software.

Relevance to AI ecosystems

The case illustrates the importance of examining acquisitions by firms that already control several layers of a technological ecosystem.

An analogous AI ecosystem could contain:

AI hardware

↓

operating system

↓

model

↓

application store

↓

AI applications

An acquisition at the application level may have competitive significance because of the acquirer's control over the surrounding ecosystem.

Principle

The competitive effect of an acquisition cannot necessarily be evaluated by examining the target as an isolated business.

12. Case Law 3 — Microsoft/Activision Blizzard

The Microsoft/Activision transaction illustrates the significance of ecosystem competition, vertical integration and access.

The European Commission examined whether the transaction could reduce competition in console and PC gaming, multi-game subscription services and cloud-game streaming.

Relevance to AI

AI ecosystems similarly contain multiple interconnected layers.

For example:

Cloud

→ compute

→ foundation model

→ application

→ distribution

→ subscription service

A large acquisition may therefore affect competition beyond the immediate market in which the target operates.

Principle

Merger analysis may need to consider competitive effects across interconnected markets where an undertaking has substantial ecosystem power.

13. Case Law 4 — Adobe/Figma

Adobe/Figma

Adobe's proposed acquisition of Figma represented a major transaction involving complementary software ecosystems.

The European Commission opened an in-depth investigation and identified concerns regarding the potential elimination of competition and innovation. The transaction was ultimately abandoned in December 2023. The Commission's merger materials record the abandonment following the Phase II investigation.

Relevance to AI

The case is particularly useful for understanding acquisitions involving:

• innovative software;
• developer ecosystems;
• rapidly expanding technologies;
• future competitive constraints; and
• products that may become important platforms.

An AI incumbent acquiring a rapidly growing developer platform could raise comparable issues.

Principle

A target's current size is not necessarily the only measure of its competitive importance.

14. Case Law 5 — Nvidia/Arm

FTC v. Nvidia Corp. / Arm Ltd.

The proposed Nvidia acquisition of Arm attracted substantial competition scrutiny because Arm's technology was widely used by competing semiconductor businesses.

The FTC identified concerns involving access to Arm's technology and the possibility that Nvidia could use control of an important technological input to disadvantage competitors. The matter illustrates the importance of neutral infrastructure and input access in concentrated technology ecosystems. The FTC case record identifies the Nvidia/Arm matter among its merger proceedings.

Relevance to AI

AI depends upon semiconductor and computing infrastructure.

An analogous transaction could involve:

AI chip designer + dominant AI platform

or

specialised accelerator technology + major cloud provider.

The resulting entity could potentially control an important input required by downstream competitors.

Principle

Control of a strategically important technological input can make a merger competition-sensitive even when the parties operate at different levels of the supply chain.

15. Case Law 6 — Amgen/Horizon

FTC v. Amgen Inc. / Horizon Therapeutics plc

The FTC challenged Amgen's acquisition of Horizon, arguing that Amgen could use its existing portfolio of blockbuster products and bargaining power to disadvantage competing products and reinforce Horizon's positions in certain markets. The matter ultimately resulted in a settlement and final order.

Relevance to AI

This illustrates an important concept for continuous consolidation:

The competitive significance of a transaction may arise from the acquirer's existing portfolio rather than merely from overlap with the target.

In AI, a company controlling:

• cloud services;
• foundation models;
• data;
• advertising;
• distribution; and
• enterprise software

could potentially leverage those assets after acquiring an adjacent AI company.

Principle

Merger analysis can consider how an acquirer's existing portfolio changes the competitive consequences of acquiring another business.

16. Case Law 7 — Booking/eTraveli

Booking Holdings / eTraveli

The European Commission prohibited Booking's acquisition of eTraveli after concluding that the transaction could strengthen Booking's position in the market for hotel booking services by increasing its visibility and traffic acquisition advantages.

The Commission's merger decision materials record the transaction as prohibited.

Relevance to AI ecosystems

The case illustrates the importance of ecosystem reinforcement.

An AI platform may acquire a complementary service not because the target is itself a major competitor but because the target can strengthen the incumbent's position in its principal market.

For example:

AI assistant + travel platform

or

AI search + specialised content platform

could generate ecosystem advantages beyond the target's standalone market position.

17. Cumulative Effects of Serial AI Acquisitions

Suppose a dominant AI firm makes the following acquisitions:

The final competitive position may be substantially different from the position produced by any single acquisition.

This creates a cumulative merger-control problem.

18. Continuous Consolidation and the "Stack Effect"

AI markets can be analysed as a technological stack:

Layer 1 — Semiconductor

GPUs, NPUs and AI accelerators.

Layer 2 — Compute

Cloud and data-centre infrastructure.

Layer 3 — Data

Training, behavioural and proprietary datasets.

Layer 4 — Foundation models

Large language, vision, multimodal and specialised models.

Layer 5 — Orchestration

Agents, APIs and workflow systems.

Layer 6 — Applications

Healthcare, finance, legal, education and industrial AI.

Layer 7 — Distribution

Search, browsers, operating systems, app stores and enterprise platforms.

Repeated acquisitions across these layers can produce vertical and conglomerate integration.

19. AI as an M&A Decision Engine

AI can itself determine acquisition priorities.

An acquisition algorithm could assign targets based on:

• probability of becoming a competitor;
• patent strength;
• user growth;
• talent concentration;
• technological uniqueness;
• customer overlap;
• data value;
• acquisition price;
• integration cost;
• regulatory risk; and
• expected strategic value.

This creates a novel competition-law question:

If an AI system systematically identifies and recommends acquisition of the most threatening startups, does the resulting acquisition pattern demonstrate strategic elimination of potential competition?

The answer would depend on evidence. The existence of AI alone would not establish an infringement.

Relevant evidence could include:

• acquisition records;
• internal strategy documents;
• model outputs;
• board presentations;
• target-ranking systems;
• post-acquisition conduct;
• treatment of competing products;
• investment decisions concerning acquired technologies; and
• communications concerning future competitive threats.

20. AI and Acquisition of Potential Competitors

Competition authorities may examine whether the target could have become an important competitor absent the acquisition.

Important indicators include:

1. technological differentiation;
2. investment levels;
3. customer adoption;
4. research pipeline;
5. patent portfolio;
6. hiring;
7. venture funding;
8. product roadmap;
9. ability to scale;
10. likelihood of independent market entry.

A small AI startup can therefore have substantial competitive significance even if its current revenues are low.

21. Data Concentration Through M&A

Repeated acquisitions can create a proprietary data reservoir.

For example:

Company A

→ consumer data

Company B

→ enterprise data

Company C

→ behavioural data

Company D

→ specialised scientific data

After acquisition:

Combined AI Data Lake

The merged company may have an advantage in:

• model training;
• personalization;
• prediction;
• advertising;
• product development;
• customer targeting; and
• automated decision-making.

The competition question becomes whether competitors can realistically reproduce the same data advantage.

22. Compute Concentration

AI requires substantial computational resources.

M&A can therefore produce concentration in:

• GPUs;
• AI accelerators;
• data centres;
• cloud infrastructure;
• model-serving infrastructure; and
• specialised inference systems.

A vertically integrated undertaking could potentially control several stages of the compute supply chain.

This can raise:

• input foreclosure;
• discriminatory access;
• interoperability;
• raising-rivals'-costs; and
• innovation-foreclosure concerns.

23. Algorithmic M&A and Tacit Coordination

AI-directed M&A also raises concerns beyond unilateral consolidation.

Multiple companies could use sophisticated systems to:

• identify acquisition targets;
• monitor competitors;
• coordinate investment;
• forecast market responses; and
• react to rivals' acquisitions.

Competition law would distinguish legitimate parallel business conduct from unlawful coordination.

The critical issue would be evidence of an agreement, concerted practice or other legally relevant coordination—not merely the fact that different companies use similar algorithms.

24. Market Partitioning Through Acquisitions

Acquisitions may also divide technological ecosystems.

For example:

• one company controls AI healthcare;
• another controls AI finance;
• another controls AI education;
• another controls AI advertising.

If the same financial or technological networks repeatedly acquire companies within specific verticals, the market may become divided among entrenched ecosystems.

Potential concerns include:

• exclusion of independent entrants;
• interoperability restrictions;
• exclusive data arrangements;
• customer lock-in;
• cross-platform discrimination; and
• reduced innovation.

25. Acquisition of Standards and Certification Systems

An AI company might acquire:

• certification software;
• auditing platforms;
• AI safety testing companies;
• benchmarking services;
• compliance platforms.

This can create a different form of market power.

If the acquirer controls both:

AI technology

and

the mechanism used to certify that technology

it may influence competitive access to the market.

The competition concern becomes stronger where certification is effectively necessary for market participation.

26. AI M&A and Intellectual Property

Acquisitions can consolidate:

• patents;
• copyrights;
• trade secrets;
• model weights;
• training methodologies;
• proprietary datasets;
• technical documentation.

Competition authorities may therefore examine whether acquisition of intellectual property prevents rivals from developing competing products.

This is especially important where the acquired IP is difficult to replicate.

27. Innovation Competition

Traditional merger analysis often examines:

• prices;
• output;
• market shares.

AI markets require significant attention to innovation competition.

Relevant questions include:

1. Will independent research continue?
2. Will alternative models be developed?
3. Will startups continue entering?
4. Will competing architectures survive?
5. Will research teams remain independent?
6. Will acquired products continue to compete with the acquirer's products?
7. Will the transaction reduce experimentation?

The Illumina/GRAIL litigation illustrates how innovation effects can become central to merger enforcement.

28. Killer Acquisition Versus Efficiency Acquisition

Not every startup acquisition is anticompetitive.

An acquisition may generate legitimate efficiencies through:

• complementary technology;
• reduced duplication;
• improved research;
• better infrastructure;
• greater investment;
• faster commercialization;
• improved security; and
• lower production costs.

The legal question is therefore whether the claimed efficiencies are:

• genuine;
• merger-specific;
• verifiable; and
• sufficiently beneficial to competition or consumers under the applicable legal framework.

29. Merger Threshold Problems

AI startups can have:

• low turnover;
• high valuation;
• substantial venture funding;
• important patents;
• strategic datasets;
• significant user growth.

Therefore, turnover-based merger thresholds may not always capture economically important transactions.

Competition authorities have consequently become increasingly interested in alternative mechanisms for identifying strategically significant acquisitions.

30. Remedies

Where competition concerns are established, possible remedies include:

Structural remedies

• divestiture;
• sale of acquired assets;
• separation of business units.

Behavioural remedies

• non-discrimination obligations;
• access commitments;
• interoperability;
• licensing;
• data access;
• firewall requirements.

Ecosystem remedies

• API access;
• portability;
• neutrality obligations;
• prohibition of self-preferencing.

Innovation remedies

• continued investment;
• preservation of research teams;
• licensing of relevant technology.

Structural remedies are generally more directly capable of restoring independent competitive structures, while behavioural remedies require continuing monitoring.

31. Evidence in AI-Directed M&A Investigations

Competition authorities may examine unusual evidence sources, including:

• acquisition algorithms;
• AI-generated target lists;
• internal model prompts;
• strategic dashboards;
• acquisition scoring systems;
• board materials;
• investment committee records;
• communications between executives;
• product roadmaps;
• technical documentation;
• model-development records;
• patent filings;
• developer activity; and
• post-acquisition product decisions.

The AI system itself may therefore become relevant evidence concerning the rationale and expected competitive consequences of an acquisition.

32. Possible Competition-Law Test

A useful analytical framework is:

STEP 1 — Identify the transaction

What exactly is being acquired?

STEP 2 — Identify the ecosystem

What other products, platforms and infrastructures does the acquirer control?

STEP 3 — Identify potential competition

Could the target become an important competitor?

STEP 4 — Identify strategic assets

Does the target possess:

• data?
• compute?
• IP?
• talent?
• distribution?
• users?

STEP 5 — Analyse foreclosure

Could competitors be denied:

• inputs?
• customers?
• interoperability?
• data?
• distribution?

STEP 6 — Analyse innovation

Will independent innovation decline?

STEP 7 — Examine acquisition history

Has the acquirer repeatedly purchased similar or complementary businesses?

STEP 8 — Examine AI decision systems

Did AI-driven analysis systematically identify competitive threats?

STEP 9 — Assess efficiencies

Are claimed efficiencies merger-specific and verifiable?

STEP 10 — Determine appropriate remedy

Would divestiture, access, interoperability or another remedy preserve effective competition?

33. Continuous Consolidation Matrix

34. Key Legal Lessons From the Case Law

The six principal cases establish several recurring themes.

1. Illumina/GRAIL

Vertical integration can threaten innovation and downstream competition where the acquiring firm controls an important technological input.

2. Meta/Within

An acquisition can be significant because of the acquirer's existing ecosystem position and the target's role in emerging technological competition.

3. Microsoft/Activision

Competition analysis can extend across interconnected technological markets and distribution ecosystems.

4. Adobe/Figma

Innovation-oriented software acquisitions may receive scrutiny even where traditional market-share analysis does not tell the whole story.

5. Nvidia/Arm

Control of an important technological input can create foreclosure concerns where competitors depend upon that input.

6. Amgen/Horizon

An acquirer's existing portfolio can materially change the competitive consequences of an acquisition.

7. Booking/eTraveli

Acquisition of a complementary business can reinforce an incumbent's ecosystem position even where the target is not simply a direct competitor.

35. Conclusion

AI-directed M&A ecosystems and continuous consolidation markets represent a shift from transaction-by-transaction competition analysis toward increasingly dynamic ecosystem analysis.

The principal competition concern is not that AI is being used to make acquisition decisions. Rather, it is whether AI-enabled acquisition strategies allow powerful firms to systematically identify, acquire and integrate emerging competitive constraints, thereby accumulating data, talent, IP, infrastructure, distribution and network effects.

The most important legal questions are therefore:

1. Who is being acquired?
2. Why is that target strategically important?
3. Could the target become an independent competitor?
4. What critical assets does it control?
5. What other ecosystem assets does the acquirer already control?
6. Is this acquisition part of a broader series of acquisitions?
7. Does the combined ecosystem foreclose rivals?
8. Does the transaction reduce innovation?
9. Can claimed efficiencies be independently verified?
10. Would the transaction strengthen durable entry barriers?

The modern merger cases involving Illumina/GRAIL, Meta/Within, Microsoft/Activision, Adobe/Figma, Nvidia/Arm, Amgen/Horizon and Booking/eTraveli demonstrate why AI acquisitions should increasingly be examined through the lenses of potential competition, innovation, vertical foreclosure, ecosystem power, strategic inputs and cumulative consolidation, rather than solely through current market shares. The European Commission's ongoing merger-guideline review also reflects the broader shift toward accounting for digitalisation and changing competitive realities.

Thus, the central competition-law challenge posed by AI-directed M&A is the transition from one-off merger control to assessment of the cumulative architecture of corporate control over an evolving technological ecosystem.

Add applicable legal frameworks and jurisdictional tests

Add applicable legal frameworks and jurisdictional tests

Correct the case-law count and legal characterisations

Separate merger control from post-merger conduct