Competition Law and Intelligent Infrastructure Allocation Systems

1. Introduction

Intelligent Infrastructure Allocation Systems (IIAS) are technology-enabled systems that use algorithms, artificial intelligence, real-time data, optimisation models and automated decision-making to allocate scarce infrastructure capacity among competing users.

Examples include:

• AI allocation of cloud-computing and GPU capacity;
• allocation of electricity-grid capacity;
• EV-charging-network access;
• telecommunications spectrum and network capacity;
• railway slots and freight terminals;
• airport slots and airport facilities;
• data-centre capacity;
• ports and logistics infrastructure;
• hydrogen pipelines and energy corridors;
• digital-platform APIs and interoperability infrastructure.

Competition law becomes relevant when the entity controlling such infrastructure also competes with the businesses seeking access to it. An intelligent allocation system can then become a mechanism for exclusion, discrimination, foreclosure, self-preferencing or strategic capacity allocation.

Under Indian competition law, Section 4 of the Competition Act, 2002 prohibits abuse of a dominant position, including denial of market access, discriminatory conditions and leveraging dominance into another market. The CCI expressly recognises denial of market access and refusal to deal as potential forms of abuse.

2. Meaning of Intelligent Infrastructure Allocation

An IIAS generally performs five functions:

A. Capacity identification

The system determines available infrastructure capacity.

For example:

Available GPU hours = 100,000
Available grid capacity = 500 MW
Available charging slots = 10,000 sessions

B. User classification

The algorithm may classify users according to:

• demand;
• priority;
• price;
• historical usage;
• contractual status;
• reliability;
• location;
• network contribution;
• predicted demand.

C. Automated allocation

The system determines who receives access, how much access and when.

D. Dynamic pricing

Access prices may change according to:

• congestion;
• demand;
• time;
• user category;
• capacity scarcity.

E. Continuous optimisation

AI may continuously modify allocation decisions based upon real-time data.

This creates an important competition-law question:

Can an apparently neutral algorithm become an instrument through which a dominant infrastructure operator excludes or disadvantages competitors?

3. Competition-Law Framework

The central legal provisions may arise under:

Indian law

Competition Act, 2002

Particularly:

• Section 3 — anti-competitive agreements;
• Section 4 — abuse of dominant position;
• Section 5 — combinations;
• Section 19 — inquiry;
• Section 26 — investigation;
• Section 27 — remedial orders.

Section 4 is especially important because dominance itself is not unlawful; the concern is abuse of dominance. The CCI describes dominance as a position of strength enabling an enterprise to operate independently of competitive forces or affect competitors or consumers in its favour.

EU law

Article 102 TFEU is particularly relevant to:

• refusal of access;
• discriminatory access;
• exclusionary infrastructure control;
• unfair conditions;
• leveraging;
• self-preferencing.

US law

Section 2 of the Sherman Act addresses monopolisation and exclusionary conduct.

4. Relevant Market

An IIAS can create competition concerns at several levels.

Example 1 — AI computing infrastructure

Upstream market: AI compute infrastructure.

Downstream market: AI model development.

A dominant cloud provider may control scarce GPUs and simultaneously compete with independent AI developers.

Example 2 — Electricity infrastructure

Upstream market: grid-access infrastructure.

Downstream market: electricity generation or storage.

Example 3 — EV charging

Upstream market: charging-network infrastructure.

Downstream market: EV charging services.

Example 4 — Airport infrastructure

Upstream market: airport facilities.

Downstream market: aircraft maintenance or ground-handling services.

Therefore, competition authorities must examine both the infrastructure market and the markets dependent upon it.

5. Essential-Facility Dimension

The most important doctrine is the essential facilities doctrine.

The basic idea is that a dominant undertaking controlling infrastructure that competitors cannot realistically duplicate may, in appropriate circumstances, have competition-law obligations concerning access.

The CCI has described four important considerations:

1. the facility is controlled by a dominant firm;
2. competitors cannot realistically reproduce it;
3. access is necessary to compete; and
4. access can feasibly be provided. 

This is particularly significant for intelligent infrastructure because the infrastructure may be partly physical and partly digital.

For example:

Electricity grid + AI allocation software + proprietary network data

may collectively constitute the infrastructure necessary for competing in an energy market.

6. Intelligent Allocation Can Produce Competition Problems

A. Discriminatory allocation

The system may allocate:

• 70% capacity to the dominant firm's subsidiary;
• 20% to affiliated firms;
• 10% to independent competitors.

Even if the algorithm technically applies the same rules, its inputs or optimisation criteria may generate discriminatory results.

B. Self-preferencing

A vertically integrated infrastructure owner may instruct its allocation system to prioritise its own downstream operations.

For example:

A cloud provider allocates scarce AI chips to its own AI models before allocating them to competing AI developers.

The concern is not merely algorithmic design but whether infrastructure control is being used to disadvantage downstream competitors.

C. Strategic capacity withholding

A dominant firm may deliberately reserve infrastructure capacity.

For example:

20% of grid capacity remains unused while competitors are told that no capacity is available.

The question becomes whether the unused capacity is technically necessary or represents exclusionary withholding.

D. Algorithmic exclusion

An algorithm may systematically downgrade competitors through variables such as:

• reliability scores;
• priority scores;
• congestion scores;
• risk classifications;
• historical utilisation;
• predicted profitability.

The allocation may therefore appear neutral while producing exclusionary effects.

E. Dynamic pricing discrimination

AI can charge different access prices to different users.

Competition authorities may examine whether the differential pricing reflects legitimate cost differences or is being used to:

• raise rivals' costs;
• discriminate against competitors;
• subsidise the dominant firm's downstream operations;
• exclude smaller entrants.

7. Important Case Laws

1. United States v. Terminal Railroad Association, 224 U.S. 383 (1912)

This is one of the classic infrastructure-access cases.

A group controlling railroad terminals in St. Louis effectively controlled access to an important transportation infrastructure.

The US Supreme Court found that the arrangement could unlawfully restrict competition by preventing competing railroads from obtaining reasonable access.

Principle

Control over a strategically necessary infrastructure can generate competition-law obligations where exclusion of competitors substantially restricts competitive opportunities.

Relevance to IIAS

An intelligent railway-slot allocation system could raise similar issues if its controller:

• prioritises affiliated railway operators;
• systematically denies competitors slots;
• uses proprietary data to disadvantage rivals;
• allocates capacity in a discriminatory manner.

8. MCI Communications Corp. v. AT&T, 708 F.2d 1081 (7th Cir. 1983)

This is a major US essential-facilities case.

The court considered access to AT&T's telecommunications network and articulated influential criteria for essential-facility analysis.

The case is important because telecommunications infrastructure was necessary for competitors attempting to provide competing services.

Principle

A refusal to provide access to infrastructure can become an antitrust problem where:

• control is held by a monopolist;
• competitors cannot reasonably duplicate the facility;
• access is necessary for competition;
• provision of access is feasible.

IIAS relevance

Modern telecommunications allocation systems can automatically determine:

• bandwidth;
• network priority;
• interconnection;
• spectrum utilisation;
• routing.

A dominant operator could theoretically use these systems to discriminate against competitors.

9. Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585 (1985)

The US Supreme Court examined a dominant ski operator's termination of a previously profitable cooperative arrangement with a competitor.

The case is important for the broader doctrine of exclusionary refusal to deal.

Principle

A dominant firm may face antitrust liability where its conduct represents exclusionary behaviour rather than legitimate competition on the merits, particularly where there is evidence of abandoning an established course of cooperation for anticompetitive reasons.

IIAS relevance

Suppose an infrastructure operator historically provides competitors access through an intelligent allocation platform and then abruptly changes the algorithm to make competitor access commercially impossible.

The historical pattern of cooperation could become relevant to assessing the conduct.

10. Verizon Communications Inc. v. Trinko, 540 U.S. 398 (2004)

Trinko is particularly important because it places limits on compulsory-access theories.

The Supreme Court did not recognise a broad general duty for monopolists to share infrastructure with competitors.

The Court also emphasised the importance of preserving incentives for firms to invest in infrastructure.

Principle

Competition law should not automatically transform every refusal to deal into an antitrust violation.

IIAS relevance

A company that spends billions developing:

• AI data centres;
• proprietary cloud infrastructure;
• charging networks;
• fibre networks;

should not automatically be required to provide competitors unlimited access.

Therefore, intelligent infrastructure cases require a balance between:

access + competition

and

investment incentives + property rights.

11. Oscar Bronner GmbH & Co. KG v. Mediaprint, C-7/97

Bronner is a foundational EU case concerning refusal of access to infrastructure.

The Court established a demanding framework for requiring a dominant undertaking to provide access to infrastructure developed for its own business.

The infrastructure must, among other things, be indispensable, and there must be no actual or potential substitute capable of providing a viable alternative.

Principle

Mere usefulness is not enough.

The infrastructure must satisfy a particularly strong form of indispensability.

IIAS relevance

A competitor cannot simply argue:

"The dominant firm's AI allocation platform is better than all alternatives."

The competition-law question is more demanding:

Is access actually indispensable for effective competition, or can competitors develop or obtain realistic alternatives?

12. IMS Health GmbH & Co. KG v. Commission, C-418/01

The IMS Health litigation concerned access to a data structure used for pharmaceutical sales information.

It is highly relevant to modern intelligent infrastructure because it demonstrates how intangible infrastructure and information systems can raise access questions.

Principle

Competition law can become relevant where proprietary information architecture becomes indispensable to downstream competition, but the threshold for compulsory access remains demanding.

IIAS relevance

Modern intelligent infrastructure may include:

• APIs;
• datasets;
• cloud interfaces;
• digital identity systems;
• interoperability protocols;
• AI model interfaces;
• data-routing systems.

Therefore, "infrastructure" should not necessarily be understood as only roads, pipelines or physical networks.

13. Deutsche Telekom AG v. Commission, C-152/19 P

This case concerned telecommunications infrastructure and pricing/access conditions.

The Court confirmed the importance of Article 102 TFEU in dealing with exclusionary conduct involving infrastructure-controlled markets.

Principle

Competition law may address conduct by dominant infrastructure operators that makes downstream competition difficult through the conditions imposed on access.

The Court also reiterated that the strict Bronner conditions arise in the specific context of refusal to provide access to infrastructure developed for the dominant undertaking's own business.

IIAS relevance

An intelligent network operator cannot necessarily avoid competition scrutiny merely because access technically exists.

The terms and conditions of access may themselves be problematic.

14. Google Shopping / Google and Alphabet v Commission, C-48/22 P

This modern case is particularly significant for intelligent systems.

The Court considered Google's treatment of competing comparison-shopping services through its general search infrastructure.

The Court distinguished a pure refusal-to-deal situation from conduct involving unfair conditions imposed on access to infrastructure that the dominant undertaking already makes available.

IIAS relevance

This distinction is extremely important.

An AI infrastructure operator may say:

"We do not refuse access."

But competition concerns may still arise if:

• the access algorithm systematically disadvantages competitors;
• affiliated services receive preferential treatment;
• competitors receive inferior infrastructure performance;
• access conditions are discriminatory.

Thus:

Access ≠ automatically fair access.

15. Google Android Auto / Alphabet, Case C-233/23

The Court's 2025 judgment is particularly relevant to digital infrastructure.

The Court clarified that the strict Bronner indispensability requirement does not automatically apply where a dominant undertaking has developed infrastructure with the intention of allowing third parties to use it.

This is important for platforms and intelligent infrastructure because a system designed from the outset as a multi-user infrastructure is different from a proprietary facility built exclusively for the dominant undertaking's internal use.

IIAS relevance

Consider:

AI platform + public API + third-party developer ecosystem.

If the platform is deliberately designed to host third parties, a competition authority may scrutinise discriminatory or exclusionary access conditions differently from a completely proprietary system.

16. Air Works India (Engineering) Pvt. Ltd. v. GMR Hyderabad International Airport Ltd.

This Indian case provides a useful infrastructure example.

The CCI considered access to airport infrastructure for third-party line-maintenance services and recognised the importance of physical airport access for competitors operating in the downstream market. The decision discussed factors including control, duplication, alternatives, denial of access and spare capacity.

IIAS relevance

Suppose an airport uses an AI system to allocate:

• aircraft parking;
• maintenance bays;
• gates;
• baggage facilities;
• ground-handling slots.

If the airport operator also participates in downstream services, the algorithm's allocation methodology could become a competition-law issue.

17. Arshiya Rail Infrastructure Ltd. v. Ministry of Railways

The Indian essential-facilities debate has also arisen in relation to railway infrastructure and container terminals.

The case illustrates an important limitation: infrastructure is not necessarily an essential facility merely because access to it is commercially advantageous.

Where rivals can realistically establish alternative facilities, compulsory-access arguments become weaker.

IIAS relevance

For an intelligent railway-allocation system, a competition authority would need to determine:

• whether alternative terminals exist;
• whether capacity can realistically be expanded;
• whether the infrastructure is indispensable;
• whether access can technically be provided;
• whether discriminatory allocation is occurring.

18. Legal Test for Intelligent Infrastructure Allocation

A useful analytical framework is:

Step 1 — Identify the infrastructure

Determine precisely what is being controlled.

It may be:

• physical infrastructure;
• digital infrastructure;
• data;
• API;
• computing capacity;
• network;
• algorithmic allocation system;
• combined physical-digital infrastructure.

Step 2 — Define the relevant market

Identify both:

Infrastructure market

and

downstream dependent market.

Step 3 — Establish dominance

Examine:

• market share;
• infrastructure control;
• entry barriers;
• network effects;
• switching costs;
• economies of scale;
• data advantages;
• vertical integration;
• technological advantages.

Step 4 — Examine allocation methodology

Ask:

Who designed the algorithm?

What variables does it use?

Are those variables objectively justified?

Can competitors audit the allocation?

Are affiliated companies treated differently?

19. Algorithmic Transparency

Competition authorities may need to examine the architecture of the allocation system.

Relevant evidence could include:

• source-code documentation;
• allocation rules;
• model specifications;
• training data;
• historical allocation records;
• API logs;
• capacity reservations;
• pricing records;
• priority scores;
• internal communications;
• model-change logs.

The absence of transparent criteria can make it difficult for competitors to determine whether discriminatory treatment exists.

However, competition law must also account for legitimate protection of:

• trade secrets;
• cybersecurity;
• intellectual property;
• confidential business information.

20. Data as a Competitive Input

Intelligent infrastructure produces enormous amounts of data.

For example, an AI-powered electricity allocation system may collect:

• grid demand;
• congestion;
• user consumption;
• production patterns;
• pricing;
• network reliability;
• capacity forecasts.

If the infrastructure operator also competes downstream, it could potentially obtain a substantial informational advantage.

This creates a data-leveraging problem:

Infrastructure control → privileged data → better algorithm → stronger downstream position → greater infrastructure control.

This can produce a self-reinforcing competitive advantage.

21. Self-Preferencing

One of the most important risks is:

Infrastructure → allocation algorithm → affiliated downstream business.

For example:

Dominant cloud provider

↓

Controls scarce GPU infrastructure

↓

AI allocation algorithm

↓

Affiliated AI model receives priority

↓

Independent AI developers receive delayed access

↓

Downstream competitors face higher costs

The legal inquiry should focus on whether the preferential allocation is objectively justified or whether infrastructure dominance is being used to protect or extend downstream market power.

22. Algorithmic Discrimination

Discrimination can take several forms.

Explicit discrimination

The algorithm expressly gives affiliated firms priority.

Indirect discrimination

Apparently neutral variables disproportionately disadvantage competitors.

Dynamic discrimination

The algorithm continually modifies allocation in ways that systematically disadvantage rivals.

Data-driven discrimination

Competitors' proprietary or commercially sensitive information is used to improve the dominant firm's downstream operations.

23. Capacity Hoarding

An intelligent system can also facilitate capacity hoarding.

For example:

A dominant operator reserves 80% of scarce infrastructure capacity but uses only 50%.

If competitors cannot obtain capacity, the competition authority may examine whether the unused reservation has a legitimate operational justification.

Relevant questions include:

• Is the reserve necessary?
• Is demand genuinely forecast to increase?
• Is the capacity technically unavailable?
• Is there a contractual commitment?
• Is the reservation disproportionately harming competitors?

24. Interoperability

Intelligent infrastructure increasingly operates through APIs and common protocols.

Competition concerns can arise if a dominant infrastructure operator:

• refuses interoperability;
• restricts API access;
• degrades interoperability;
• changes technical standards strategically;
• imposes discriminatory API conditions.

The EU telecommunications framework demonstrates that access obligations can sometimes operate independently of a traditional abuse-of-dominance finding, particularly where sectoral regulation promotes sharing of physical infrastructure.

Thus, competition law and sector regulation may operate together.

25. Infrastructure Allocation and Mergers

Intelligent infrastructure also creates merger-control concerns.

Consider:

Major cloud provider + GPU infrastructure provider

or

Grid operator + energy-storage platform

or

EV charging network + vehicle manufacturer.

The competition authority may examine whether the combined entity could:

• foreclose rivals;
• reserve infrastructure capacity;
• deny interoperability;
• increase rivals' costs;
• obtain sensitive competitor data;
• favour affiliated businesses.

Potential remedies include:

• access commitments;
• interoperability;
• non-discrimination;
• data separation;
• functional separation;
• capacity commitments;
• monitoring mechanisms.

26. Competition Concerns in Different Infrastructure Sectors

27. Intelligent Infrastructure and Predatory Pricing

An infrastructure operator may also use intelligent pricing algorithms to subsidise its downstream business.

For example:

Infrastructure access price to rivals = ₹100

while:

Affiliated downstream business effectively receives infrastructure = ₹20.

The competition authority may investigate whether the pricing structure amounts to:

• margin squeeze;
• discriminatory pricing;
• cross-subsidisation;
• exclusionary pricing.

The precise legal test depends upon the jurisdiction and market structure.

28. Efficiency Justifications

Not every differential allocation is anti-competitive.

An intelligent allocation system may legitimately prioritise users based on:

• safety;
• emergency requirements;
• technical constraints;
• reliability;
• congestion;
• latency;
• cybersecurity;
• contractual obligations;
• public-interest requirements.

Therefore:

Different treatment ≠ automatically unlawful discrimination.

The central question is whether the distinction is objectively justified, proportionate and competitively neutral, or whether it is being used to disadvantage rivals.

29. Remedies

Competition authorities could consider several remedies.

Structural remedies

• separation of infrastructure and downstream businesses;
• divestiture;
• ownership restrictions.

Behavioural remedies

• non-discriminatory access;
• transparent allocation criteria;
• fair pricing;
• capacity commitments;
• interoperability.

Algorithmic remedies

• independent algorithmic auditing;
• monitoring;
• logging;
• explainability requirements;
• restrictions on discriminatory variables;
• independent review of allocation decisions.

Data remedies

• data-access obligations;
• data separation;
• prohibition on using competitor-sensitive data;
• interoperability requirements.

30. Emerging AI Infrastructure Dimension

The issue is becoming particularly important in AI markets.

Modern AI competition depends upon scarce infrastructure such as:

• advanced GPUs;
• high-performance computing;
• data centres;
• electricity;
• networking;
• cooling;
• cloud capacity.

The OECD has specifically identified AI infrastructure—including computing resources such as advanced chips—as an emerging competition-policy concern because concentration in these inputs can affect competition in downstream AI markets.

This creates a potential chain:

Chips → computing capacity → cloud → AI models → applications

Control at an upstream stage can therefore influence competition at several downstream stages.

31. Distinction Between Traditional and Intelligent Infrastructure

The competition-law principles remain broadly recognisable, but the evidentiary and technical complexity increases substantially.

32. Key Legal Issues for Examination

An examination answer should identify these questions:

1. Is the infrastructure essential?
2. Who controls it?
3. Does the controller possess dominance?
4. Is there a realistic substitute?
5. Can competitors duplicate the infrastructure?
6. Is access technically feasible?
7. Is access actually refused?
8. If access exists, are its conditions discriminatory?
9. Does the allocation algorithm favour an affiliate?
10. Is capacity being strategically withheld?
11. Does the system increase rivals' costs?
12. Does the system exploit competitor-sensitive data?
13. Is there objective justification?
14. Are sector-specific access rules applicable?
15. Would compulsory access undermine investment incentives?
16. What remedy would preserve both competition and innovation?

33. Overall Legal Principle

The central competition-law proposition can be stated as follows:

An intelligent infrastructure allocation system is not unlawful merely because it is controlled by a dominant undertaking or because it produces differential allocation. Competition-law concern arises where control over indispensable or strategically important infrastructure is used, without adequate objective justification, to exclude competitors, discriminate against them, raise their costs, restrict market access, or leverage infrastructure dominance into downstream markets.

The modern jurisprudence also shows an important distinction between refusal to provide access to proprietary infrastructure and discriminatory or unfair conditions imposed on infrastructure that is already made available to third parties. Recent EU jurisprudence has expressly emphasised this distinction.

34. Conclusion

Intelligent Infrastructure Allocation Systems represent a new intersection between essential facilities, digital markets, algorithmic decision-making and abuse of dominance.

The traditional competition-law concern was:

Who controls the infrastructure?

The modern question is broader:

Who controls the infrastructure, who controls the data, who designs the allocation algorithm, what criteria does it apply, and whether that system can be used to determine which competitors obtain meaningful access to the market?

The jurisprudence from Terminal Railroad, MCI, Aspen Skiing, Trinko, Bronner, IMS Health, Deutsche Telekom, Google Shopping, Google Android Auto, Air Works and Arshiya Rail Infrastructure demonstrates the evolution from traditional physical infrastructure toward increasingly complex digital and hybrid infrastructure.

For future competition-law analysis, the most important concepts are therefore:

Dominance + indispensability + access + algorithmic neutrality + interoperability + data control + non-discrimination + objective justification + investment incentives.

These principles provide a framework for assessing intelligent allocation systems across AI infrastructure, cloud computing, electricity grids, telecommunications, EV charging, transport, ports, airports, data centres and digital platforms.