Competition Law and Medical AI Ecosystem Concentration

Introduction

Medical AI ecosystem concentration refers to a situation in which a small number of firms control several interconnected layers of healthcare artificial intelligence, such as:

• medical datasets and electronic health records;
• cloud computing and AI infrastructure;
• foundation models and machine-learning systems;
• diagnostic and clinical-decision software;
• hospital information systems;
• medical imaging platforms;
• APIs and interoperability infrastructure;
• wearable and remote-monitoring data;
• distribution channels and app stores; and
• acquisition pipelines for AI-healthcare startups.

Competition law becomes particularly important because a firm does not necessarily need a very large share of the final medical-AI application market to exercise significant competitive influence. Control over a critical upstream input—such as clinical data, computing infrastructure, APIs, hospital distribution, or an installed healthcare platform—can create vertical leverage into downstream AI markets.

The most directly relevant precedents are currently found in digital health, healthcare technology, data-driven healthcare and adjacent technology markets rather than in a large body of reported cases specifically involving "medical AI." The European Commission's Google/Fitbit decision, for example, expressly considered digital healthcare, health-data accumulation and API foreclosure.

I. Legal Framework

1. Relevant Market

A medical-AI investigation may require several overlapping relevant markets rather than one broad "AI healthcare" market.

Possible markets include:

1. AI-assisted radiology;
2. AI pathology;
3. AI diagnostic support;
4. clinical decision-support systems;
5. AI-enabled patient monitoring;
6. healthcare data analytics;
7. AI drug-discovery platforms;
8. medical transcription and documentation;
9. AI-enabled EHR services;
10. medical-imaging datasets;
11. healthcare cloud infrastructure;
12. AI model-training infrastructure; and
13. APIs connecting AI applications to healthcare data.

The relevant market may therefore be layered.

For example:

Clinical data → cloud infrastructure → foundation model → AI diagnostic application → hospital/EHR distribution → patient/physician use.

A company controlling several consecutive layers may possess competitive advantages that cannot be understood simply by examining its share of the final AI application market.

II. Sources of Medical-AI Ecosystem Concentration

1. Data concentration

Medical AI requires large quantities of:

• electronic health records;
• diagnostic images;
• pathology slides;
• genomic information;
• clinical notes;
• laboratory results;
• wearable-device data;
• longitudinal patient histories; and
• labelled clinical datasets.

Data can become a competition concern where competitors cannot obtain comparable datasets on reasonable terms.

However, possession of a large dataset does not automatically establish market power. Authorities generally need to consider substitutability, uniqueness, interoperability, access possibilities, quality and the ability of rivals to replicate the input.

The Google/Fitbit investigation illustrates this distinction. The European Commission examined whether Fitbit's health and fitness data could strengthen Google's position in digital healthcare and whether competitors could be foreclosed from accessing Fitbit data through its Web API.

III. Six Important Case Laws

1. Google / Fitbit — European Commission, Case M.9660

Facts

Google proposed acquiring Fitbit, a company collecting substantial health and fitness information through wearable devices.

The European Commission investigated several possible competition effects, including:

• accumulation of health data;
• digital healthcare;
• wearable-device markets;
• API access;
• vertical foreclosure; and
• Google's ability to combine datasets.

The Commission specifically considered whether Google could restrict competitors' access to Fitbit's Web API and thereby disadvantage digital-healthcare providers.

Competition principle

The case is particularly important for medical AI because data access can constitute an important competitive input even where the merging parties are not direct competitors.

The Commission recognized that third-party digital-healthcare applications depended on access to Fitbit data through APIs and examined whether the merged entity would possess both the ability and incentive to foreclose that access.

Google ultimately accepted commitments concerning the use of Fitbit data and access arrangements.

Relevance to Medical AI

The same theory can apply where:

dominant EHR/cloud/wearable platform → controls health dataset → downstream medical-AI developers depend upon API → platform restricts or discriminates against access.

This makes Google/Fitbit one of the most directly relevant precedents for medical-AI data concentration.

2. Illumina / GRAIL — FTC and European Union

Facts

Illumina, a major provider of DNA-sequencing systems, sought to acquire GRAIL, which developed a multi-cancer early-detection test using sequencing technology.

The FTC alleged that Illumina's control of sequencing technology could allow it to disadvantage competing cancer-detection developers.

The FTC ultimately ordered divestiture, and the Fifth Circuit found substantial evidence supporting the FTC's determination that the transaction threatened competition. Illumina subsequently announced that it would divest GRAIL.

The EU proceedings also produced an important judgment concerning the Commission's jurisdiction to review the transaction under Article 22 of the EU Merger Regulation.

Competition principle

The case demonstrates the importance of vertical foreclosure.

The concern was not merely:

"Illumina and GRAIL are competitors."

Instead, the theory involved control of an upstream technological input and a downstream healthcare innovation.

Medical-AI application

A comparable structure could arise where:

AI infrastructure provider → controls essential medical-computing/data input → acquires medical-AI developer.

Potential concerns include:

• preferential access to computing;
• discriminatory pricing;
• withholding technical information;
• delayed API access;
• degradation of interoperability;
• preferential model deployment;
• exclusion of rival AI developers; and
• reduced innovation.

Illumina/GRAIL therefore provides a significant precedent for examining vertical integration between an infrastructure provider and an innovative healthcare technology company.

3. FTC v. Surescripts

Facts

Surescripts operated electronic-prescription networks connecting healthcare providers, pharmacies and other participants.

The FTC alleged that Surescripts maintained monopolies in electronic-prescription routing and eligibility through exclusionary agreements, loyalty arrangements and conduct making it more difficult for customers to use competing platforms.

The litigation produced a ruling recognizing Surescripts' very high market share and the importance of network effects; the subsequent settlement prohibited specified exclusionary practices.

Competition principle

Surescripts is especially important for network effects and multihoming.

Healthcare platforms can become stronger as more:

• hospitals;
• doctors;
• pharmacies;
• insurers;
• EHR systems; and
• patients

join the network.

This creates a feedback loop:

more users → more data/connections → more value → greater adoption → greater entry barriers.

Medical-AI relevance

A medical-AI platform integrated with thousands of hospitals could potentially develop a similar ecosystem:

hospitals → EHR integration → patient data → AI model improvement → physician adoption → additional hospitals.

Competition authorities may therefore examine whether contractual arrangements prevent hospitals or healthcare professionals from multihoming between competing AI platforms.

4. FTC v. U.S. Anesthesia Partners / Welsh Carson

Facts

The FTC challenged conduct involving U.S. Anesthesia Partners and private-equity firm Welsh Carson concerning consolidation in Texas anesthesia markets.

The FTC alleged that the defendants used acquisitions and contracting strategies to suppress competition and consolidate anesthesiology services. The litigation remains a significant healthcare-concentration proceeding.

The district court's proceedings have addressed the structure of the Texas anesthesia market and the alleged monopolization strategy.

Competition principle

The case demonstrates that competition authorities may examine serial acquisitions and cumulative consolidation, rather than examining each transaction in isolation.

Medical-AI application

The same concept is significant for AI healthcare because a major technology or healthcare company may acquire numerous small AI businesses:

• radiology AI startup;
• pathology AI startup;
• clinical documentation company;
• patient-monitoring AI company;
• medical-imaging company;
• drug-discovery AI company.

Each acquisition may appear individually small.

But cumulative acquisitions can create ecosystem concentration.

Authorities may therefore examine:

"What competitive constraints existed before the sequence of acquisitions, and what remains afterward?"

5. United States v. Aetna Inc. / Humana Inc.

Facts

The DOJ challenged the proposed merger between Aetna and Humana.

The government alleged that the merger would substantially lessen competition in Medicare Advantage and certain commercial health-insurance markets. The district court ultimately prevented the merger.

Competition principle

The case demonstrates the importance of examining:

• localized markets;
• concentration;
• head-to-head competition;
• entry;
• substitution; and
• loss of competitive constraints.

Medical-AI relevance

Medical AI increasingly interacts with insurers through:

• risk assessment;
• claims analytics;
• fraud detection;
• utilization management;
• disease prediction;
• clinical decision support; and
• personalized healthcare.

A merger between an insurer and an AI healthcare platform could therefore produce conglomerate and vertical concerns, even if the firms do not compete directly in AI software.

For example:

insurer + medical-AI platform + patient-data infrastructure

could potentially give the combined company access to information or distribution channels unavailable to independent AI developers.

The Aetna/Humana framework illustrates why competition analysis should examine the competitive relationships between businesses across connected healthcare markets rather than focusing only on the immediate product sold.

6. Google Search / Digital-Platform Antitrust Litigation

The broader Google antitrust litigation is relevant because medical AI increasingly operates through general-purpose digital ecosystems.

The competition question can involve:

• search;
• operating systems;
• cloud;
• app distribution;
• advertising;
• data;
• AI models; and
• specialized healthcare applications.

The central lesson for medical AI is the distinction between competition within a market and leverage across markets.

A company may have substantial power at one layer and use that position to reinforce another layer.

For example:

operating system → app distribution → healthcare application → health data → AI model.

The legal relevance of this ecosystem approach becomes particularly significant when AI applications depend upon distribution controlled by the same firm that supplies infrastructure.

IV. Medical AI Ecosystem as a Multi-Layer Market

A useful analytical model is:

The Google/Fitbit decision is particularly instructive because the Commission examined both horizontal data effects and vertical access foreclosure.

V. Key Antitrust Theories

1. Data foreclosure

A dominant healthcare platform could theoretically restrict rivals' access to:

• patient records;
• imaging datasets;
• genomic information;
• wearable data;
• clinical annotations; or
• real-world evidence.

The crucial questions would be:

1. Is the dataset commercially or competitively important?
2. Is it unique or reasonably replicable?
3. Do rivals have alternative sources?
4. Is access technically feasible?
5. Does the dominant firm control access?
6. Does foreclosure substantially affect downstream competition?

2. API foreclosure

APIs are especially important because they determine whether an AI developer can interact with:

• EHR systems;
• medical devices;
• wearables;
• hospital databases;
• laboratory systems; and
• patient-management platforms.

Google/Fitbit expressly illustrates regulatory concern over the possibility of restricting access to a health-data API.

3. Cloud-compute concentration

Medical AI can require substantial:

• GPU capacity;
• storage;
• high-performance computing;
• model-training infrastructure; and
• inference capacity.

If the same firm controls cloud infrastructure and a competing medical-AI application, potential theories include:

• discriminatory pricing;
• preferential compute allocation;
• technical degradation;
• delayed access;
• tying;
• bundling; and
• refusal to supply.

4. EHR and AI integration

EHR systems may function as important gateways.

An EHR provider controlling:

patient data + clinical workflow + physician interface + AI marketplace

could potentially give preferential treatment to its own AI tools.

Potential competition issues include:

• self-preferencing;
• discriminatory API access;
• interoperability restrictions;
• tying;
• exclusive arrangements;
• switching costs; and
• restrictions on data portability.

Surescripts demonstrates how control of healthcare technology infrastructure can interact with network effects and contractual exclusion.

VI. Self-Preferencing

Suppose a hospital platform operates an AI marketplace containing:

• its own diagnostic AI; and
• competing independent AI systems.

If the platform systematically gives its own product:

• higher ranking;
• default status;
• superior API access;
• lower transaction fees;
• better computing resources; or
• privileged access to clinical data,

competition authorities could examine whether the conduct amounts to unlawful exclusion.

The relevant issue would not simply be whether the platform's own AI performs better. The inquiry would concern whether the platform is using control of an adjacent bottleneck to disadvantage competing products.

VII. Tying and Bundling

Medical-AI concentration can also create tying concerns.

For example:

EHR software + mandatory proprietary AI

or:

cloud services + proprietary medical-AI model

or:

hospital information system + exclusive AI diagnostic service.

The analysis would generally consider:

1. whether two distinct products exist;
2. whether the firm possesses sufficient power in the tying product;
3. whether customers are effectively forced to take the tied product;
4. whether rivals are foreclosed; and
5. whether legitimate efficiencies justify the arrangement.

VIII. Merger Control and Killer-Acquisition Concerns

Medical AI is characterized by numerous small startups.

A dominant ecosystem operator might acquire:

• promising diagnostic-AI companies;
• medical-imaging AI developers;
• clinical-documentation startups;
• healthcare data companies;
• AI model developers; and
• interoperability providers.

Traditional turnover thresholds may sometimes fail to capture acquisitions of startups whose present revenue is low but whose future competitive significance is high.

Illumina/GRAIL demonstrates how competition authorities can focus on innovation competition and future competitive constraints, rather than simply present sales.

IX. Innovation Competition

Medical AI creates an unusual competition-law issue: the principal competitive harm may be reduced innovation rather than immediately higher prices.

Potential indicators include:

• fewer independent AI models;
• reduced clinical experimentation;
• fewer alternative diagnostic systems;
• slower development of new algorithms;
• reduced interoperability;
• diminished research collaboration; and
• loss of independent sources of training data.

The FTC's healthcare competition work expressly recognizes that anticompetitive conduct can affect healthcare costs, quality and innovation.

X. Interoperability as a Competition Remedy

Where concentration is caused by ecosystem control, structural divestiture is not necessarily the only possible remedy.

Potential remedies may include:

1. API access obligations

Competitors receive continuing access to necessary interfaces.

2. Data portability

Healthcare customers can transfer relevant information between systems.

3. Non-discrimination

The platform cannot discriminate between its own AI and rival AI providers.

4. Data silos

Certain datasets cannot be combined with datasets used in competing commercial activities.

5. Interoperability

Technical standards prevent artificial switching barriers.

6. Contractual restrictions

Exclusive or loyalty arrangements can be prohibited.

7. Divestiture

Where behavioural remedies cannot adequately preserve competition, structural separation may be considered.

Google/Fitbit is a particularly useful example of behavioural commitments addressing data and interoperability concerns, whereas Illumina/GRAIL illustrates the potential role of structural separation in a vertical healthcare technology transaction.

XI. Application of the Six Cases

XII. Indian Competition-Law Perspective

For India, the principal framework is the Competition Act, 2002, administered by the Competition Commission of India (CCI).

Medical-AI concentration could potentially engage:

• Section 3 — anti-competitive agreements;
• Section 4 — abuse of dominant position;
• Section 5 — combinations;
• Section 6 — regulation of combinations;
• relevant theories concerning refusal of access;
• discriminatory conditions;
• tying/bundling;
• denial of market access; and
• leveraging dominance from one market into another.

A medical-AI ecosystem could therefore generate a Section 4 issue where an enterprise has substantial power in an upstream market—such as EHR infrastructure, healthcare data, cloud services or hospital software—and uses that position to restrict competition in a downstream AI market.

XIII. Competition-Law Test for Medical AI

A useful examination framework is:

Step 1 — Identify the ecosystem

↓

Step 2 — Define the relevant product and geographic markets

↓

Step 3 — Identify the bottleneck

Data / Compute / EHR / API / Cloud / Distribution

↓

Step 4 — Determine market power

Market share + network effects + switching costs + entry barriers

↓

Step 5 — Identify conduct

Exclusivity / tying / bundling / self-preferencing / refusal / discrimination

↓

Step 6 — Establish foreclosure

Can competitors realistically obtain equivalent inputs?

↓

Step 7 — Examine effects

Prices + quality + innovation + choice + entry

↓

Step 8 — Consider efficiencies

Accuracy / safety / interoperability / cost savings / clinical benefits

↓

Step 9 — Select proportionate remedy

Access / interoperability / non-discrimination / data separation / divestiture

XIV. Important Distinction: Data Size ≠ Monopoly

Possession of a large medical dataset should not automatically be treated as unlawful concentration.

The competition analysis should distinguish:

Large dataset

from

non-replicable strategic input

and from

legally or commercially controllable bottleneck.

Google/Fitbit demonstrates this nuanced approach: the Commission examined the importance of Fitbit data, alternative sources of health data, potential foreclosure and the competitive structure of digital healthcare rather than assuming that merely possessing health data established dominance.

Conclusion

Competition law concerning medical AI ecosystem concentration is fundamentally moving beyond traditional market-share analysis.

The principal concern is increasingly the architecture of control:

Data + Compute + Model + EHR + API + Distribution + Healthcare Network

Control over multiple layers can create opportunities for vertical foreclosure, self-preferencing, data exclusion, interoperability restrictions, tying, discriminatory access and acquisition-driven concentration.

The most directly instructive precedents include Google/Fitbit, for health-data and API access; Illumina/GRAIL, for vertical healthcare-technology integration and innovation; and Surescripts, for network effects, exclusion and healthcare technology infrastructure. The healthcare-consolidation cases involving U.S. Anesthesia Partners and Aetna/Humana provide additional frameworks for analysing cumulative concentration and loss of competitive constraints.