Infrastructure Physics As Determinant Of Market Power

Introduction

Infrastructure physics refers to the physical, engineering and technological constraints that determine how an infrastructure system can be built, operated, scaled and accessed. In competition law, these constraints can become a source of market power because infrastructure is often subject to characteristics such as high fixed costs, limited capacity, network effects, bottlenecks, geographic immobility, interoperability constraints, economies of scale, switching costs and long replacement cycles.

Market power therefore cannot always be understood merely by examining prices or market shares. In infrastructure-intensive markets, the physical structure of the system may determine who can enter, who can access essential inputs, how quickly competitors can expand, and whether customers can realistically switch.

Examples include:

• electricity transmission grids;
• telecommunications networks;
• ports, airports and railways;
• pipelines and gas infrastructure;
• broadband fibre networks;
• cloud and data-centre infrastructure;
• semiconductor fabrication;
• payment and digital infrastructure;
• EV charging networks; and
• large-scale AI compute infrastructure.

The central competition-law proposition is:

Where physical or technological constraints make duplication economically or practically difficult, control over infrastructure can translate into durable market power.

1. Meaning Of Infrastructure Physics

Infrastructure physics is broader than simply "physical infrastructure."

It encompasses the technical properties of an infrastructure system that constrain competitive behaviour.

Major components

A. Capacity constraints

Infrastructure frequently has finite capacity.

For example, a transmission line, railway track, airport runway or data-centre connection cannot accommodate unlimited users simultaneously.

A dominant infrastructure operator can exploit capacity scarcity through:

• discriminatory allocation;
• congestion pricing;
• priority access;
• capacity withholding;
• delayed connections; or
• preferential treatment of affiliated businesses.

B. Economies of scale

Infrastructure often involves enormous initial investment.

Examples include:

• fibre networks;
• electricity grids;
• pipelines;
• ports;
• cloud data centres;
• semiconductor fabs.

The average cost may decline substantially as utilisation increases.

Consequently:

AC=F+VCQAC=\frac{F+VC}{Q}

where:

• F = fixed infrastructure cost;
• VC = variable cost;
• Q = output.

High fixed costs can make multiple competing networks inefficient.

This can create a natural-monopoly structure.

C. Geographic immobility

A digital service can theoretically expand globally, but many infrastructure assets cannot.

A railway track, electricity transmission line, port terminal, pipeline or fibre route is tied to a particular geography.

This means that competition may need to be assessed at the level of:

• route;
• corridor;
• access point;
• interchange;
• terminal;
• grid node; or
• geographic catchment.

D. Network topology

Infrastructure systems frequently operate as networks rather than collections of independent assets.

A network's structure may determine competitive significance.

For example:

Generation → transmission → distribution → consumer

or:

Data centre → backbone → access network → end user

Control over a strategically located node can therefore be more important than overall market share.

E. Interoperability constraints

Infrastructure frequently requires compatibility.

Examples include:

• telecommunications standards;
• payment systems;
• charging protocols;
• rail networks;
• electricity grids;
• cloud APIs;
• operating systems.

A firm controlling a compatibility layer may obtain market power even if it does not control the entire infrastructure system.

2. Why Infrastructure Physics Creates Market Power

Traditional competition analysis frequently begins with:

Market definition → market share → barriers to entry → competitive effects.

Infrastructure markets often require an additional layer:

Physical constraints → technical bottleneck → access dependency → strategic control → market power.

The key issue is not simply:

"How large is the firm?"

It is:

"Can rivals realistically reproduce the infrastructure necessary to compete?"

If the answer is no, infrastructure control may constitute a significant competitive advantage.

3. Sunk Costs And Irreversibility

Infrastructure investment is often sunk.

Once a railway, pipeline, fibre route or data centre has been constructed, its value may be highly specific to the relevant market.

This creates asymmetric competitive conditions.

An incumbent may have:

• an established network;
• existing rights of way;
• regulatory permissions;
• installed connections;
• historical customer relationships; and
• accumulated technical data.

A new entrant must often recreate these assets from scratch.

This creates a structural entry barrier.

4. Bottlenecks And Essential Infrastructure

A particularly important form of infrastructure market power arises when an asset constitutes a bottleneck.

A bottleneck exists where competitors require access to a particular infrastructure component and there is no reasonably equivalent alternative.

Examples:

• a single port terminal;
• a railway connection;
• an electricity interconnector;
• a gas pipeline;
• a telecommunications network;
• a payment network;
• a cloud infrastructure layer.

Competition law may become concerned where the infrastructure owner:

1. controls the bottleneck;
2. possesses market power;
3. denies or restricts access;
4. discriminates between users; or
5. imposes commercially unreasonable conditions.

5. Infrastructure Physics And Essential Facilities Doctrine

The essential facilities doctrine is particularly relevant.

Its basic logic is:

Where an infrastructure asset is indispensable for effective competition and cannot reasonably be duplicated, exclusionary control over that asset may raise competition-law concerns.

However, not every important infrastructure facility is legally "essential."

Courts generally examine factors such as:

• indispensability;
• practical availability of alternatives;
• duplication feasibility;
• economic cost of duplication;
• technical constraints;
• geographic constraints;
• regulatory barriers; and
• effects on downstream competition.

6. Infrastructure Physics And Market Definition

Physical infrastructure can affect the relevant market itself.

Consider an airport.

Passengers may technically be able to travel from another airport, but if the alternative airport is:

• 150 km away;
• poorly connected;
• capacity constrained; or
• unsuitable for particular routes,

it may not be an effective substitute.

Thus, physical distance and connectivity can influence substitutability.

The same applies to:

• ports;
• electricity grids;
• railway stations;
• broadband networks;
• pipelines; and
• data centres.

7. Infrastructure Physics And Entry Barriers

Infrastructure creates several distinct entry barriers.

Thus, engineering constraints can become economic barriers to entry.

8. Infrastructure Physics And Discriminatory Access

A dominant infrastructure provider may not need to refuse access completely.

It can instead manipulate:

Timing

• delayed connection;
• slow technical approval;
• prolonged testing.

Capacity

• insufficient allocation;
• congestion claims;
• artificial scarcity.

Quality

• degraded access;
• inferior interoperability;
• slower connectivity.

Price

• excessive access fees;
• discriminatory tariffs;
• margin squeeze.

Information

• withholding technical specifications;
• asymmetric information;
• selective disclosure.

These strategies can preserve infrastructure-based dominance while appearing formally open to competitors.

9. Infrastructure Physics And Vertical Integration

Infrastructure control becomes particularly significant when the infrastructure owner operates downstream.

For example:

Infrastructure owner

↓

Wholesale access

↓

Downstream service

If the infrastructure operator also competes downstream, it may have an incentive to disadvantage rivals.

Potential mechanisms include:

• preferential capacity;
• discriminatory connection;
• higher access charges;
• lower service quality;
• faster deployment for affiliated operations;
• access to superior technical information.

This creates a classic vertical foreclosure problem.

10. Infrastructure Physics And Margin Squeeze

Infrastructure can generate a margin-squeeze problem.

Suppose:

Paccess>Pretail−CdownstreamP_{access} > P_{retail}-C_{downstream}

where:

• PaccessP_{access} = infrastructure access price;
• PretailP_{retail} = downstream retail price;
• CdownstreamC_{downstream} = downstream operating cost.

If the infrastructure owner sets access charges so high that an equally efficient downstream competitor cannot profitably compete, infrastructure control may be used to foreclose competition.

This has historically been important in:

• telecommunications;
• energy;
• rail;
• broadband; and
• digital infrastructure.

11. Infrastructure Physics And Network Effects

Physical infrastructure can interact with network effects.

For example, a telecommunications network becomes more valuable as more users connect to it.

This can create:

More users → greater utilisation → lower average cost → greater attractiveness → more users

This positive feedback loop can make infrastructure dominance self-reinforcing.

Digital infrastructure can intensify this because physical networks and software ecosystems can operate together.

12. Infrastructure Physics In Digital Markets

The concept is increasingly relevant to digital competition.

Modern digital markets depend on physical infrastructure such as:

• data centres;
• fibre-optic networks;
• submarine cables;
• cloud servers;
• GPUs;
• electricity supply;
• cooling systems;
• storage infrastructure.

Thus, apparently "virtual" markets may ultimately depend upon physical scarcity.

For AI:

AI Market Power≈Compute+Data+Energy+Network+DistributionAI\ Market\ Power \approx Compute + Data + Energy + Network + Distribution

A company controlling scarce compute infrastructure can potentially influence downstream AI markets even if the AI application itself is technically reproducible.

13. Infrastructure Physics And AI Compute

AI infrastructure presents a particularly interesting competition-law problem.

Large-scale AI systems require:

• GPUs/accelerators;
• high-bandwidth networking;
• electricity;
• cooling;
• data-centre capacity;
• specialised chips;
• cloud infrastructure.

If these resources are concentrated among a few providers, physical scarcity can become a competitive bottleneck.

Potential concerns include:

• preferential access to GPUs;
• long-term exclusive capacity agreements;
• discriminatory cloud pricing;
• refusal to supply compute;
• tying compute to AI services;
• capacity reservation;
• interoperability restrictions.

Therefore:

AI competition may increasingly depend upon physical infrastructure competition.

14. Case Law

1. United States v. Terminal Railroad Association of St. Louis

This is one of the foundational infrastructure cases.

A group of railroads controlled the terminal facilities necessary for efficient railway access into St. Louis.

The Supreme Court was concerned that control over the terminal system could exclude competing railroads.

Principle

Control over an infrastructure bottleneck can generate substantial market power where rivals cannot practically compete without access.

Importance

The case illustrates the relationship between:

physical bottleneck → indispensability → exclusionary power.

It is a foundational reference for essential-facilities analysis.

15. United States v. Otter Tail Power Co.

Otter Tail controlled electricity transmission facilities and also competed in electricity distribution.

The Supreme Court considered its refusal to provide transmission services to municipalities seeking to enter electricity distribution.

Principle

Control of transmission infrastructure can create substantial competitive power when downstream competitors depend upon that infrastructure.

Competition significance

The case demonstrates the risks created by vertical integration plus infrastructure control.

The infrastructure operator could potentially use its upstream position to protect its downstream market.

16. MCI Communications Corp. v. AT&T

The case concerned telecommunications infrastructure and AT&T's control over critical telecommunications facilities.

MCI argued that AT&T's refusal to provide access to its network impaired competition.

The Seventh Circuit developed an influential framework for essential-facilities analysis.

Key factors

Among the important considerations were:

1. control of the essential facility;
2. competitor inability reasonably to duplicate the facility;
3. denial of access; and
4. feasibility of providing access.

Significance

This case explicitly connects physical infrastructure duplicability with competition-law power.

If duplication is technically or economically unrealistic, infrastructure control becomes much more significant.

17. Bronner v Mediaprint

The Court of Justice of the European Union considered access to a newspaper home-delivery distribution system.

The issue was whether refusal to provide access to an existing distribution network could constitute an abuse of dominance.

The Court adopted a strict approach to indispensability.

Principle

An infrastructure or distribution system is not "essential" merely because access would make competition easier or more convenient.

The alternative must be practically unavailable, and duplication must not be realistically possible.

Infrastructure-physics relevance

The case demonstrates that competition law must distinguish:

commercial advantage

from

physical indispensability.

18. Oscar Bronner And Duplication Economics

Bronner is particularly important because it recognises the importance of economic feasibility.

The relevant question is not simply whether a competitor would incur costs in creating its own network.

The issue is whether establishing an alternative would be:

• technically feasible;
• economically viable; and
• realistically achievable.

This is directly connected to infrastructure physics.

Where duplication requires enormous capital, land, regulatory approvals and years of construction, the practical competitive alternative may be substantially weaker.

19. IMS Health GmbH & Co. KG v NDC Health

The CJEU considered access to a pharmaceutical sales-data system organised around a particular geographical structure.

The Court examined whether the refusal to license a protected system could amount to abusive conduct.

Principle

Compulsory access to an infrastructure or protected system requires exceptional circumstances.

Among them is the need for the facility or resource to be genuinely indispensable.

Infrastructure relevance

The case illustrates how technical architecture and standardisation can create structural dependence.

A market participant may become dependent not merely on an asset, but on the architecture through which the market operates.

20. Bronner–IMS Health–Magill Framework

The European jurisprudence therefore establishes an important principle:

Control + indispensability + absence of realistic alternatives + exclusionary effect

is substantially more significant than merely possessing an attractive or commercially successful infrastructure.

This prevents competition law from converting every refusal to deal into compulsory infrastructure sharing.

21. Deutsche Telekom v Commission

In Deutsche Telekom, the European Commission and EU courts examined wholesale access to telecommunications infrastructure and downstream pricing.

The case is particularly important for the margin squeeze doctrine.

Infrastructure relevance

The incumbent controlled essential telecommunications infrastructure while downstream competitors depended on wholesale access.

If wholesale prices and retail prices were structured so that competitors could not profitably compete, infrastructure control could produce foreclosure.

Principle

A dominant infrastructure operator cannot necessarily use control over upstream infrastructure to make downstream competition economically impossible.

22. Telefónica v Commission

The Telefónica broadband case similarly concerned access conditions and pricing in telecommunications.

The EU courts considered whether wholesale pricing could restrict effective competition in downstream broadband markets.

Infrastructure-physics relevance

Telecommunications networks demonstrate that market power may arise from:

• network deployment costs;
• limited infrastructure alternatives;
• access dependency;
• geographic coverage; and
• sunk investment.

The case therefore illustrates how the physical architecture of communications networks can influence competitive conditions.

23. Google Shopping And Infrastructure-Like Digital Bottlenecks

Although Google Shopping is not a traditional physical-infrastructure case, it demonstrates how competition analysis can extend the infrastructure concept to strategically important digital access points.

A search engine can operate as a gateway through which consumers access downstream markets.

The economic structure can resemble infrastructure:

Gateway → traffic → downstream services

The relevant infrastructure is therefore partly digital and informational rather than purely physical.

This is increasingly important for platform competition.

24. Infrastructure Physics And Natural Monopoly

A natural monopoly generally emerges where one infrastructure network can serve the market more efficiently than multiple parallel networks.

Examples historically include:

• electricity grids;
• gas pipelines;
• water systems;
• railway infrastructure;
• local telecommunications networks.

The competitive problem is therefore unusual.

Competition may be possible for the market, rather than within the infrastructure.

For example:

One firm may operate the network while multiple firms compete over downstream services.

This is why infrastructure sectors frequently require:

• access regulation;
• interoperability;
• non-discrimination;
• price regulation;
• structural separation; or
• independent network governance.

25. Infrastructure Physics And Competition For The Market

Where infrastructure cannot efficiently be duplicated, competition may shift from:

competition within the network

to:

competition to control the network.

This creates risks of:

• bidding manipulation;
• concession capture;
• infrastructure acquisition;
• exclusive rights;
• long-term contracts;
• political influence;
• regulatory capture.

Competition law may therefore intersect with:

• public procurement law;
• sector regulation;
• merger control;
• concession law; and
• state-aid/subsidy rules.

26. Infrastructure Physics And Merger Control

Infrastructure concentration can make mergers particularly significant.

Suppose two firms operate competing infrastructure networks.

Even if their current market shares appear moderate, their merger could eliminate the only realistic alternative network.

The authority should therefore consider:

Structural questions

• How many independent networks exist?
• Can another network realistically be constructed?
• How long would construction take?
• Is land available?
• Are permits available?
• Is capacity expandable?
• Are there alternative routes?

Dynamic questions

• Can technology make duplication cheaper?
• Will new infrastructure emerge?
• Can customers multi-home?
• Can interoperability reduce dependency?

This prevents market-share analysis from overlooking infrastructure realities.

27. Infrastructure Physics And Entry Timing

Infrastructure markets frequently have long lead times.

A competitor may theoretically be able to enter but require:

• five years to obtain permits;
• several years to construct the network;
• substantial capital;
• access to land;
• connection agreements.

Therefore, theoretical entry does not necessarily constitute effective competitive constraint.

Competition authorities should distinguish:

Theoretical entry≠Timely entry\text{Theoretical entry} \neq \text{Timely entry}

This is especially important in infrastructure merger and dominance analysis.

28. Infrastructure Physics And Capacity Withholding

A dominant infrastructure operator can sometimes create scarcity without physically destroying capacity.

For example:

Available capacity = 100

but the operator allocates:

Competitors = 20

Affiliate = 70

Reserved/unused = 10

The unused capacity may itself become a competitive instrument.

This raises questions concerning:

• strategic capacity reservation;
• foreclosure;
• discriminatory allocation;
• exclusionary contracting.

29. Infrastructure Physics And Long-Term Contracts

Long-term infrastructure contracts can transform physical scarcity into contractual market power.

Examples include:

• exclusive port agreements;
• pipeline capacity reservations;
• cloud-compute commitments;
• fibre access contracts;
• airport slots;
• semiconductor supply agreements.

The competitive problem becomes:

physical scarcity + contractual foreclosure

rather than physical scarcity alone.

30. Infrastructure Physics And Switching Costs

Infrastructure often creates high switching costs.

A business connected to one network may need to incur substantial costs to move to another.

These may include:

• equipment replacement;
• software migration;
• retraining;
• new technical certification;
• contractual termination costs;
• downtime;
• interoperability testing.

Consequently, infrastructure dominance can persist even where alternative providers technically exist.

31. Infrastructure Physics And Data Centres

Data centres illustrate the convergence of physical and digital market power.

A hyperscale data centre requires:

• land;
• electricity;
• cooling;
• fibre connectivity;
• specialised hardware;
• network interconnection.

The availability of these resources varies geographically.

A firm may therefore obtain local or regional power through control of:

• high-capacity electricity connections;
• fibre routes;
• specialised data-centre campuses;
• scarce GPU clusters.

The resulting market power is partly determined by physical infrastructure scarcity.

32. Infrastructure Physics And Energy Constraints

AI and cloud markets increasingly depend on electricity.

Suppose:

Compute Capacity≤Available EnergyEnergy per ComputationCompute\ Capacity \leq \frac{Available\ Energy}{Energy\ per\ Computation}

If electricity connection capacity is scarce, expansion of compute infrastructure may become physically constrained.

Thus, electricity infrastructure can indirectly determine:

• AI compute supply;
• cloud capacity;
• latency;
• model-training capacity;
• data-centre expansion.

This produces a new form of competition issue:

energy infrastructure as an upstream input into digital market power.

33. Infrastructure Physics And Regulatory Bottlenecks

Sometimes the infrastructure itself is not scarce, but regulatory access is.

Examples include:

• rights of way;
• spectrum;
• permits;
• grid connections;
• airport slots;
• railway paths;
• environmental approvals.

A firm with an established infrastructure footprint may therefore possess a regulatory advantage over entrants.

Competition analysis should distinguish:

natural physical scarcity

from

artificial scarcity created through regulatory or contractual arrangements.

34. Infrastructure Physics And Strategic Bottlenecks

Not every infrastructure asset has equal competitive importance.

A strategically located asset can have disproportionate significance.

For example:

• a single bridge connecting two logistics corridors;
• a particular electricity interconnector;
• a submarine cable landing station;
• a major payment switch;
• a critical railway junction;
• a semiconductor fabrication node.

Thus:

Strategic Importance≠Asset SizeStrategic\ Importance \neq Asset\ Size

A small node may control a disproportionately large amount of economic activity.

35. Competition-Law Test

A useful analytical framework is:

Step 1 — Identify the infrastructure

What physical or technological asset is involved?

Step 2 — Identify the constraint

What makes duplication difficult?

Step 3 — Determine dependency

Which competitors depend on the infrastructure?

Step 4 — Assess alternatives

Are technically and economically realistic alternatives available?

Step 5 — Assess capacity

Is capacity scarce, expandable or deliberately restricted?

Step 6 — Examine conduct

Has the infrastructure owner:

• denied access?
• discriminated?
• tied access?
• imposed excessive prices?
• delayed connection?
• reserved capacity?
• degraded interoperability?

Step 7 — Assess competitive effects

Could the conduct:

• exclude rivals;
• raise their costs;
• reduce innovation;
• increase prices;
• reduce quality;
• slow entry?

Step 8 — Examine justification

Are restrictions justified by:

• safety;
• engineering requirements;
• congestion;
• investment incentives;
• security;
• interoperability;
• legitimate capacity constraints?

36. Six Core Case-Law Lessons

37. Emerging Competition-Law Issues

The concept is becoming increasingly important in new markets.

A. AI compute

GPU clusters and data-centre capacity may become bottlenecks.

B. EV charging

Charging networks can generate geographic and network effects.

C. Semiconductor fabrication

Advanced fabs require enormous capital and specialised technical capabilities.

D. Submarine cables

Cable landing stations and international fibre routes can create strategic bottlenecks.

E. Cloud infrastructure

Compute, storage, networking and data-egress infrastructure can create switching barriers.

F. Digital payments

Payment infrastructure may become a gateway through which downstream transactions must pass.

G. Smart grids

Grid-control systems can become infrastructure bottlenecks for distributed energy competitors.

38. Key Legal Principle

The deepest competition-law lesson is that market power can be embedded in the architecture of an infrastructure system itself.

A firm does not necessarily need to engage in obviously exclusionary conduct from the beginning.

It may acquire structural power because it controls:

the route, node, network, capacity, interface, standard, connection or physical resource upon which competitors depend.

Accordingly, competition authorities should examine not only who sells the product, but also who controls the physical and technical conditions under which the product can be supplied.

Conclusion

Infrastructure physics as a determinant of market power represents a shift from purely firm-centric competition analysis toward structural and systems-based analysis.

The most important variables are:

scarcity + capacity + topology + geographic constraints + sunk costs + network effects + interoperability + switching costs + duplication feasibility.

Where these factors combine, infrastructure may become a bottleneck through which an entire market must pass.

The principal case law—from Terminal Railroad and Otter Tail to MCI, Bronner, IMS Health, Deutsche Telekom and Telefónica—demonstrates that competition law has long recognised that control over infrastructure can produce extraordinary competitive power.

The modern development is that the infrastructure is no longer necessarily a railway or electricity grid. It may be a cloud region, GPU cluster, fibre route, data centre, payment network, semiconductor fab, AI compute layer or digital interoperability interface.

Thus, in modern competition law:

Physical and technical architecture can itself become an economic source of market power.