AI-Driven Capital Budgeting Systems and Investment Concentration

Introduction

AI-driven capital budgeting systems use machine-learning models, predictive analytics, optimization algorithms, and automated decision rules to determine where capital should be invested, how much should be invested, when investment should occur, and which projects should be rejected or deferred.

Traditional capital budgeting generally evaluates projects through NPV, IRR, payback period, risk-adjusted returns, and management judgment. AI systems can instead process large quantities of financial, operational, market, customer, supply-chain, ESG, and competitor data and continuously reallocate capital.

The competition-law concern arises when AI-controlled investment decisions produce investment concentration. Concentration can occur when:

1. the same AI infrastructure is used by many investors;
2. algorithms repeatedly direct capital toward the same firms;
3. institutional investors acquire significant interests in competing businesses;
4. AI systematically excludes smaller or new entrants from financing;
5. investment platforms use common datasets or models to coordinate investment behaviour;
6. AI-driven portfolio optimization creates common ownership across competitors; or
7. dominant financial-data or AI providers become gatekeepers for capital allocation.

The issue is therefore not simply whether AI makes better investment decisions. The central competition question is whether AI-mediated capital allocation changes market structure or facilitates conduct that restricts competition.

Recent competition-policy literature identifies AI's ability to create new gatekeepers, reinforce data feedback loops, and increase concentration as an emerging competition-policy concern.

I. Meaning of AI-Driven Capital Budgeting

An AI capital-budgeting system may perform the following functions:

1. Project selection

The system ranks projects according to predicted:

• return on investment;
• probability of success;
• market demand;
• financing requirements;
• regulatory risk;
• supply-chain risk;
• competitor behaviour;
• expected cash flows.

2. Dynamic capital allocation

Instead of approving a five-year investment plan once a year, an AI system may continuously move capital between:

• subsidiaries;
• geographic markets;
• product lines;
• acquisitions;
• infrastructure projects;
• R&D;
• suppliers;
• financial assets.

3. Risk-based allocation

AI may assign risk scores to potential investments and automatically reduce capital exposure to projects that appear less attractive.

4. Portfolio-wide optimization

A common algorithm can simultaneously evaluate thousands of investments and determine an optimal portfolio.

This becomes competition-sensitive when numerous competing firms use the same AI provider, model, data source, or optimization architecture.

II. How Investment Concentration Can Develop

A. Algorithmic convergence

Different investors may independently use AI systems trained on similar data.

Even without an express agreement, their investment decisions may converge.

For example:

Investor A's AI recommends investing heavily in semiconductor manufacturer X.

Investor B's AI uses similar market data and makes the same recommendation.

Investor C's AI does likewise.

The result may be a substantial concentration of capital in X even though the investors never directly coordinated.

Competition law must distinguish between lawful independent parallel conduct and coordination involving information exchange, common algorithms, or a facilitating intermediary.

III. Common Ownership Problem

The more important structural issue is common ownership.

Suppose an AI-managed investment fund owns substantial shares in:

• Company A;
• Company B;
• Company C;

and all three compete in the same product market.

The AI system may optimize the investor's aggregate portfolio rather than maximizing the competitive strength of any single company.

This can create a theoretical incentive to avoid aggressive competition between portfolio companies.

The issue has become particularly significant for large institutional investors. In 2025, the U.S. FTC and DOJ filed a statement of interest in litigation alleging that BlackRock, State Street and Vanguard used shareholdings in competing coal companies in ways that allegedly discouraged output. The agencies specifically addressed the potential application of Section 7 of the Clayton Act to common ownership. The allegations remain distinct from a judicial finding that the alleged conduct occurred.

IV. AI as a Capital-Allocation Gatekeeper

AI may create a new form of financial gatekeeping.

If a dominant AI-financial platform controls:

• investment analytics;
• credit scoring;
• project-risk assessment;
• financial forecasting;
• investment recommendations;
• access to institutional investors;

businesses may become dependent on that platform.

A startup could therefore face two separate barriers:

Capital requirement → AI evaluation → investor decision → market entry

If the AI system systematically assigns lower investment scores to firms outside an established ecosystem, capital concentration can become self-reinforcing.

V. Relevant Competition-Law Theories

1. Abuse of dominance

A dominant AI or financial-data provider could potentially engage in:

• discriminatory access;
• exclusionary ranking;
• self-preferencing;
• tying;
• refusal to supply essential data;
• discriminatory pricing;
• exclusion of competing investment platforms.

The relevant question would be whether the conduct harms the competitive process rather than merely disadvantaging an individual investment.

2. Anti-competitive agreements

AI systems can create risks where competing investors:

• share competitively sensitive information;
• use a common algorithm;
• coordinate investment strategies;
• delegate decisions to the same intermediary;
• agree upon investment restrictions;
• exchange future investment intentions.

An algorithm does not automatically eliminate the requirement to establish the relevant elements of an agreement or concerted practice.

3. Hub-and-spoke coordination

A particularly important scenario is:

Investor A → AI intermediary ← Investor B

The intermediary may receive confidential information from multiple competing investors and use it to generate recommendations.

If the system knowingly facilitates coordination, the intermediary can potentially become the "hub" connecting competing "spokes."

4. Merger control

AI-driven capital budgeting can also influence merger concentration.

An investment platform may use AI to identify acquisition targets continuously.

Repeated acquisitions may result in:

AI target identification → acquisition → integration → market concentration → additional acquisition capacity

If the acquisitions individually appear small but collectively reshape a market, merger-control authorities may examine the cumulative structural effects.

China's amended Anti-Monopoly Law expressly addresses the use of data, algorithms, technology, capital advantages and platform rules as potential means of engaging in monopolistic conduct.

VI. At Least 6 Relevant Case Laws

There is currently no major reported judgment squarely deciding the proposition that an AI-driven corporate capital-budgeting system itself constitutes an antitrust infringement. The following cases are therefore important doctrinal analogues concerning algorithmic coordination, information exchange, common ownership, market concentration, and exclusionary investment strategies.

1. United States v. RealPage, Inc.

Jurisdiction: United States
Principal issue: Algorithmic pricing and information exchange

RealPage is highly relevant because it demonstrates how an algorithm can potentially transform information supplied by competing businesses into coordinated market behaviour.

The competition concern is analogous to AI capital budgeting:

Competitor information → common algorithm → recommendations → parallel market behaviour

The important lesson is that an intermediary's algorithm does not necessarily make competitively sensitive coordination lawful merely because individual businesses receive automated recommendations.

Recent reporting identifies RealPage litigation as a major development in U.S. scrutiny of algorithmic coordination.

Relevance

If competing institutional investors feed future investment intentions, target preferences, or capital-allocation information into the same AI system, authorities could examine whether the system facilitates coordination.

2. United States v. Topkins

Jurisdiction: United States
Principal issue: Algorithmic implementation of price coordination

Topkins involved online sellers using algorithms to implement an agreement concerning prices.

The important principle is that automated implementation does not immunize an underlying anticompetitive agreement.

Application to AI capital budgeting

If competing investors first agree on investment strategies and then use AI to implement those strategies, the algorithm may be merely the mechanism through which the agreement operates.

The relevant legal question remains the underlying coordination.

3. Eturas v. Lietuvos Respublikos Konkurencijos Taryba

Court: Court of Justice of the European Union
Case: C-74/14
Principal issue: Common electronic platform and concerted practices

Travel agencies used a common electronic booking system. The system transmitted a message restricting discounts.

The CJEU considered whether knowledge of the system's restrictive message and continued participation could support an inference of participation in a concerted practice.

Relevance to AI capital budgeting

This is particularly useful where multiple investment institutions use a common AI platform.

The analytical chain could be:

Common platform → common information → common algorithmic recommendation → awareness → continued participation

But the existence of a common AI tool alone would not establish liability; evidence concerning knowledge, communication and participation remains important.

4. T-Mobile Netherlands BV v. Raad van bestuur van de Nederlandse Mededingingsautoriteit

Court: CJEU
Case: C-8/08
Principal issue: Information exchange and concerted practice

The case concerned exchanges of information between competitors.

The Court emphasized the potential competitive significance of exchanges that reduce uncertainty concerning competitors' future conduct.

Application

AI capital-budgeting systems can process:

• future investment plans;
• expected production capacity;
• acquisition intentions;
• R&D expenditure;
• expansion plans.

If competing firms exchange such forward-looking information through a common AI system, the system could reduce strategic uncertainty.

5. Wood Pulp II

Court: CJEU
Case: Joined Cases C-89/85 etc.
Principal issue: Parallel conduct and concerted practices

The case is important for distinguishing legitimate parallel behaviour from behaviour that can support an inference of concertation.

AI relevance

Suppose ten investors' AI systems independently recommend the same industry because all rely on publicly available information.

Similarity of investment decisions alone should not automatically equal collusion.

This distinction is critical because AI systems naturally generate similar outputs when trained on similar data.

6. United States v. Philadelphia National Bank

Court: U.S. Supreme Court
Citation: 374 U.S. 321 (1963)
Principal issue: Market concentration and structural merger analysis

The case is a foundational U.S. merger decision emphasizing the importance of market structure and concentration.

AI-capital-budgeting relevance

AI can transform investment decisions from isolated transactions into a structural force.

For example:

AI repeatedly selects incumbent firms → capital becomes concentrated → incumbents expand faster → smaller competitors receive less capital → entry declines

Philadelphia National Bank therefore provides a useful structural foundation for examining whether capital allocation contributes to increasing concentration.

7. United States v. Brown Shoe Co.

Court: U.S. Supreme Court
Citation: 370 U.S. 294 (1962)
Principal issue: Vertical integration and market concentration

Brown Shoe is important because merger analysis under U.S. antitrust law can consider structural effects beyond immediate price effects.

Application to AI investment systems

An AI system might systematically allocate capital toward vertically integrated ecosystems:

manufacturer → distributor → logistics provider → retailer

If repeated acquisitions or investment decisions increase vertical foreclosure risks, traditional merger principles may become relevant even though the underlying capital-allocation decisions were algorithmically generated.

8. United States v. Philadelphia National Bank and Brown Shoe — Combined Structural Significance

Taken together, these cases establish an important analytical proposition:

Competition law can examine how investment and acquisition decisions alter market structure, not merely whether an individual transaction immediately increases prices.

That principle becomes more significant when AI enables investors to make hundreds of investment and acquisition decisions at machine speed.

VII. Common Ownership and AI

Common ownership deserves separate treatment.

Assume:

If AI systems manage these portfolios collectively, investors may become economically exposed to the entire industry rather than to individual firms.

Potential competition concerns include:

1. Reduced incentives to compete

An investor may benefit from higher profits across the industry even if one portfolio company loses market share.

2. Coordinated shareholder influence

Large shareholders can potentially influence:

• board appointments;
• strategic plans;
• capital expenditure;
• expansion;
• acquisitions;
• pricing strategy.

3. Investment withdrawal threats

A large investor could potentially reward or penalize portfolio companies through capital allocation.

4. Capital discipline becoming a competitive constraint

AI may recommend that a portfolio company avoid aggressive expansion because expansion would reduce the profits of another portfolio company.

This is one of the central theoretical competition risks associated with common ownership.

The U.S. government's 2025 filing in the BlackRock litigation specifically recognized that institutional investors' ownership positions in competing firms can raise Section 7 issues where those holdings are allegedly used for anticompetitive purposes.

VIII. Investment Exclusion and New-Entry Barriers

AI-driven capital budgeting may disadvantage startups even without explicit discrimination.

Suppose an AI model is trained primarily on historical data.

Established companies possess:

• longer operating histories;
• larger customer datasets;
• stronger credit histories;
• more predictable revenues;
• greater collateral;
• established supply chains.

A startup may consequently receive a lower investment score.

This can produce a feedback loop:

Historical success → higher AI score → more capital → greater growth → more data → even higher AI score

At the same time:

Historical absence → lower AI score → less capital → weaker growth → less data → continued low score

This can create a capital-market version of a data-network-effect barrier to entry.

Research on AI and competition policy specifically identifies data feedback loops and concentration in AI services as emerging competition concerns.

IX. AI and Private Equity / Venture Capital

AI-driven capital allocation can have particularly strong effects in private-equity and venture-capital markets.

An AI platform may rank hundreds of potential investments and recommend only a small group.

Repeated use may cause:

AI screening → investment concentration → successful portfolio companies → more training data → improved model → greater concentration

This can generate algorithmic investment path dependency.

A small group of companies could repeatedly receive capital because the algorithm interprets previous investment success as evidence supporting additional investment.

X. Risk of Capital Hoarding

AI can also optimize for risk-adjusted returns rather than competitive diversity.

Consequently, capital may migrate toward:

• established technology firms;
• dominant platforms;
• large infrastructure operators;
• highly rated borrowers;
• companies possessing large datasets.

The result may be capital hoarding by incumbents.

Competition law may become relevant where the capital-allocation mechanism is connected with exclusionary conduct, discriminatory access, or coordinated investment strategies.

XI. Algorithmic Investment Coordination

A particularly difficult scenario is:

Several competing investors independently delegate investment decisions to the same AI provider.

The provider receives:

• portfolio information;
• investment intentions;
• risk tolerances;
• acquisition targets;
• expected returns;
• sector preferences.

The AI then recommends similar strategies to each investor.

The legal analysis should ask:

1. What information does the AI provider receive?
2. Is the information competitively sensitive?
3. Is information aggregated or identifiable?
4. Does the provider know the identity of competing investors?
5. Are future investment intentions disclosed?
6. Does the algorithm deliberately coordinate recommendations?
7. Is there a contractual or other mechanism facilitating coordination?
8. Are investors aware of the system's cross-client information use?

The critical distinction is between independent algorithmic optimization and algorithmically facilitated coordination.

XII. China-Specific Competition Perspective

China's amended Anti-Monopoly Law is particularly relevant because it expressly recognizes that undertakings must not use data, algorithms, technology, capital advantages and platform rules to engage in monopolistic conduct.

China's developing approach therefore creates a useful framework for analyzing AI-driven investment concentration.

Potential areas include:

A. Platform-controlled capital allocation

A large platform could use its financial ecosystem to favour businesses operating within its own ecosystem.

B. Algorithmic discrimination

AI could assign systematically different investment conditions to comparable enterprises.

C. Capital-based exclusion

A dominant undertaking could use financial resources to acquire or constrain competitors.

D. Algorithmically assisted acquisitions

AI could identify emerging competitors before they become significant and facilitate serial acquisitions.

E. Data advantages

Investment decisions could be based upon proprietary platform data unavailable to competing investors.

China's 2026 Internet-Platform Antitrust Compliance Guidelines also specifically identify algorithmic collusion as a monopoly risk, although publicly reported Chinese enforcement decisions centred primarily on AI-driven algorithmic collusion remain limited.

XIII. Regulatory and Compliance Issues

AI-driven investment systems should therefore incorporate:

1. Data separation

Competitor-sensitive information should not be unnecessarily pooled.

2. Model governance

Firms should document:

• training data;
• model objectives;
• optimization functions;
• decision variables;
• human overrides.

3. Conflict-of-interest controls

Where one investor owns competing businesses, the AI should identify common-ownership conflicts.

4. Explainability

Investment recommendations should be auditable.

5. Human review

Material acquisitions and capital reallocations should receive human competition-law review.

6. Algorithmic firewalls

Information relating to competing portfolio companies should be appropriately segregated.

7. Acquisition monitoring

AI systems should track cumulative acquisitions rather than examining each acquisition in isolation.

XIV. Competition-Law Risk Matrix

XV. Key Legal Principles Emerging From the Case Law

The cases collectively support several important principles.

Principle 1 — Technology does not immunize conduct

The use of an algorithm does not automatically transform potentially anticompetitive conduct into lawful conduct.

Principle 2 — Parallel outcomes are not automatically collusion

Similar AI recommendations can result from similar data and optimization objectives without an agreement.

Principle 3 — Information exchange matters

Forward-looking strategic information can be particularly competition-sensitive.

Principle 4 — Market structure matters

Competition law can examine whether capital allocation contributes to increased concentration.

Principle 5 — Common ownership deserves scrutiny

Ownership across competing firms can create incentives different from those of a single-firm investor.

Principle 6 — AI can accelerate structural effects

A human investment committee may make dozens of decisions annually; an AI system can evaluate thousands of investments continuously.

Principle 7 — The intermediary can become legally significant

An AI provider operating as a common intermediary between competitors may create a hub-and-spoke risk depending upon the facts.

XVI. Hypothetical Example

Assume an AI investment platform manages ₹10 trillion for 100 institutional investors.

The system identifies five companies as optimal investments in a particular industry.

Within three years:

• 70 investors acquire shares in Company A;
• 65 acquire shares in Company B;
• 58 acquire shares in Company C;
• the same AI provider supplies investment recommendations;
• the provider receives detailed portfolio data from all investors;
• smaller competitors receive substantially less capital.

The legal investigation would not simply ask:

"Did the AI choose the same companies?"

It would ask:

1. Why did the AI select those companies?
2. What data trained the model?
3. Did the provider share investor information?
4. Were investors aware of that sharing?
5. Did investors coordinate through the provider?
6. Did common ownership reduce competitive incentives?
7. Did the system exclude emerging competitors?
8. Did the resulting concentration affect market entry?
9. Were acquisitions accelerated by the AI?
10. Did the AI provider possess market power in investment analytics?

This demonstrates why AI-driven capital budgeting is fundamentally a market-structure issue as well as a technology issue.

XVII. Conclusion

AI-driven capital budgeting can dramatically improve the speed and sophistication of investment decisions, but it can also alter the distribution of capital throughout an economy.

The principal competition-law concern is not the mere use of AI. It is the possibility that AI transforms independent investment decisions into coordinated, exclusionary, or structurally concentrating behaviour.

The most important risks are:

AI optimization → common investment choices → common ownership → reduced competitive incentives → capital concentration → higher barriers to entry.

The relevant legal framework therefore combines antitrust agreements doctrine, information-exchange principles, merger control, abuse-of-dominance rules, common-ownership analysis, and digital-platform regulation.

For present purposes, the strongest doctrinal authorities are RealPage, Topkins, Eturas, T-Mobile Netherlands, Wood Pulp II, Philadelphia National Bank, and Brown Shoe. They do not establish that AI capital budgeting is itself unlawful; rather, they provide the legal principles for determining when algorithmic investment systems could facilitate coordination, exclusion, or excessive market concentration. Current regulatory developments reinforce that distinction: authorities are increasingly examining algorithms, common ownership and capital concentration, while direct enforcement specifically involving AI-controlled capital budgeting remains comparatively undeveloped.