Autonomous Procurement Agents and Buyer-Side Market Power Concentration

Introduction

Autonomous procurement agents are AI systems that can independently perform substantial parts of a firm's purchasing process: identifying suppliers, comparing bids, negotiating prices and contract terms, allocating purchase volumes, switching suppliers, forecasting demand, and executing or recommending transactions with limited human intervention.

From a competition-law perspective, the central concern is not merely that procurement becomes automated. The more significant issue arises when autonomous agents aggregate purchasing decisions across a large firm, platform, corporate group, or consortium, thereby increasing the buyer's ability to influence supplier prices and non-price terms.

This can create or strengthen monopsony or oligopsony power—the buyer-side analogue of monopoly power. The U.S. Supreme Court has expressly recognized monopsony as market power on the buying side of a market.

The competition problem can therefore be represented as:

AI procurement autonomy → purchasing aggregation → increased buyer leverage → supplier dependence → reduced supplier output/investment/innovation → possible downstream competitive harm.

The technology itself is not necessarily unlawful. The legal question is whether the resulting conduct substantially restricts competition, facilitates exclusionary conduct, creates unlawful coordination, or produces anticompetitive monopsony effects.

I. Meaning of Buyer-Side Market Power

1. Monopoly versus monopsony

A monopoly exists principally on the selling side:

Seller → controls output/price → Buyers

A monopsony operates in the opposite direction:

Multiple Sellers → compete for → Dominant Buyer

A powerful buyer may be able to:

• force suppliers to reduce prices;
• impose unfavorable payment terms;
• demand exclusivity;
• reduce purchased quantities;
• impose costly compliance requirements;
• obtain commercially sensitive information;
• discriminate among suppliers;
• exclude particular suppliers;
• reduce supplier margins;
• acquire strategically important suppliers;
• influence innovation incentives.

The Supreme Court in Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co. described monopsony as market power on the buy side and recognized that predatory bidding can be a form of exclusionary conduct.

II. How Autonomous Procurement Agents Can Increase Buyer Power

An autonomous procurement system can affect buyer power in several ways.

A. Procurement aggregation

Suppose a corporation previously had 100 purchasing managers negotiating independently with suppliers.

An AI procurement system may consolidate those decisions into one enterprise-wide system.

Instead of:

100 purchasing units → 100 negotiations

the system produces:

1 procurement architecture → thousands of purchasing decisions

This substantially increases the buyer's bargaining leverage.

B. Supplier substitution becomes algorithmic

An autonomous agent can continuously calculate:

• price;
• quality;
• delivery reliability;
• switching cost;
• supplier capacity;
• geographic location;
• historical performance;
• defect rates;
• credit risk;
• alternative suppliers.

It may therefore determine that Supplier A can be replaced by Supplier B almost instantaneously.

The supplier may consequently face a much more credible threat:

"Accept the buyer's terms or lose the contract."

This can transform ordinary bargaining power into more substantial buyer-side market power.

C. Dynamic purchasing

Traditional procurement usually occurs periodically.

Autonomous systems can continuously monitor:

• supplier prices;
• commodity prices;
• inventory;
• production capacity;
• transportation costs;
• competitor behavior;
• exchange rates;
• demand forecasts.

The agent can repeatedly alter purchasing allocations.

This potentially gives a large buyer real-time monopsony leverage.

III. Concentration Through Autonomous Procurement

The most important structural concern arises where several purchasing activities become centralized.

Consider:

Firm A: 10% purchasing share
Firm B: 8%
Firm C: 7%
Firm D: 5%

Individually, none may have substantial buyer power.

But a common procurement platform, purchasing consortium, merger, or coordinated procurement architecture could potentially bring a much larger share under a single decision-making mechanism.

The resulting concentration may create:

Monopsony

One dominant buyer.

Oligopsony

A small number of powerful buyers.

Algorithmically reinforced oligopsony

Several large buyers use highly sophisticated procurement systems that repeatedly monitor and respond to supplier behavior.

The last category presents particularly difficult competition questions because the purchasing environment may become highly automated without an obvious traditional human agreement.

IV. Buyer Power Is Not Automatically Anticompetitive

A critical distinction must be maintained.

Large purchasing volume ≠ unlawful monopsony.

A buyer can become more efficient because of:

• economies of scale;
• lower transaction costs;
• better logistics;
• reduced inventory costs;
• improved forecasting;
• lower search costs;
• better supplier matching;
• reduced procurement fraud.

These may constitute genuine efficiencies.

The competition concern arises when purchasing power is used to reduce competitive supply rather than merely reduce procurement costs.

The distinction was emphasized by the U.S. Department of Justice in its discussion of buyer power: a reduction in input prices may result either from genuine efficiencies or from monopsony power. Where the latter causes a reduction in the quantity of inputs purchased, competitive harm can result.

V. Principal Competition Risks

1. Excessive supplier price suppression

An autonomous agent may be programmed to minimize procurement expenditure.

If the optimization function effectively becomes:

Minimize supplier price at every opportunity

the agent may systematically push suppliers below sustainable competitive returns.

Short-term purchasing savings could therefore produce long-term:

• supplier exit;
• capacity reduction;
• reduced investment;
• lower quality;
• reduced innovation;
• fewer suppliers.

This is the classic monopsony concern.

2. Supplier exclusion

An autonomous agent might repeatedly select only the lowest-cost suppliers.

This could create a feedback loop:

Supplier receives fewer orders → loses economies of scale → costs increase → AI ranks supplier lower → receives even fewer orders.

Eventually:

supplier exit → fewer suppliers → greater buyer concentration.

Thus an algorithm designed merely to optimize procurement could potentially contribute to endogenous supplier concentration.

VI. Data Advantage and Buyer Power

Autonomous procurement systems can accumulate enormous amounts of supplier information.

For example:

• minimum acceptable price;
• production capacity;
• reservation price;
• inventory;
• delivery constraints;
• previous bids;
• contract concessions;
• financial difficulties;
• switching costs.

If a buyer has access to this information across thousands of transactions, the procurement agent can engage in sophisticated individualized bargaining.

The resulting power may be greater than the buyer's aggregate market share alone suggests.

VII. Algorithmic Price Discrimination Against Suppliers

Suppose an autonomous procurement agent determines that:

• Supplier A has many alternative customers;
• Supplier B is economically dependent on the buyer;
• Supplier C has excess capacity;
• Supplier D cannot easily switch customers.

The system may offer:

This is analogous to the buyer-side concerns identified in the DOJ's Aetna-Prudential analysis, where buyer power could be increased by the ability to negotiate separately with suppliers and exploit differences in supplier dependence and switching costs.

VIII. Autonomous Procurement and Tacit Coordination

A more difficult problem arises when multiple buyers use autonomous procurement agents.

Suppose:

• Buyer A's AI observes supplier prices;
• Buyer B's AI observes similar information;
• both continuously react to market conditions.

Even without an explicit agreement, sophisticated algorithms may make procurement strategies increasingly interdependent.

Potential risks include:

• synchronized purchasing reductions;
• common supplier exclusion;
• coordinated refusal to deal;
• parallel reductions in purchase prices;
• common procurement standards;
• information exchange;
• allocation of suppliers.

The legal analysis would still require evidence of the relevant anticompetitive agreement or effects. Similar algorithmic behavior alone should not automatically be treated as proof of collusion.

IX. Autonomous Agents and Procurement Information Exchange

Procurement agents can also exchange or process information at unprecedented scale.

An enterprise could possess data concerning:

• thousands of supplier bids;
• competing supplier quotations;
• expected capacity;
• future price movements;
• contract expiration dates.

If competing buyers gain access to each other's competitively sensitive purchasing information, competition-law risks increase.

This is particularly relevant to information-exchange theories under the rule of reason.

The Second Circuit's analysis in Todd v. Exxon Corp. is relevant because it considered information exchange in a market where coordination could be facilitated by the nature of the information and the structure of the market. Later courts continue to use Todd in analyzing information-exchange restraints.

X. Autonomous Procurement and Merger Control

Autonomous procurement makes buyer-side merger analysis increasingly important.

Traditionally, merger analysis often focuses on:

Will the merged company raise prices to consumers?

For procurement-intensive industries, another question is:

Will the merged company acquire excessive power over suppliers?

Modern competition analysis expressly recognizes this possibility.

The European Commission's merger guidance states that mergers can create a significant impediment to effective competition on purchasing markets where purchasing markets are concentrated and sellers are fragmented, potentially creating or strengthening monopsony or oligopsony power. It also recognizes possible reductions in supplier investment, innovation and output.

XI. Relevant Market Definition

A competition authority must determine the relevant purchasing market.

For autonomous procurement, this may involve:

Product market

What inputs are being purchased?

Examples:

• semiconductor components;
• agricultural products;
• logistics services;
• cloud computing;
• professional services;
• pharmaceuticals;
• industrial components.

Geographic market

Where can suppliers realistically sell?

Buyer alternatives

Which buyers are realistic substitutes from the supplier's perspective?

Supplier alternatives

Can suppliers switch to:

• other purchasers;
• exports;
• alternative industries;
• direct-to-consumer sales;
• substitute products?

Switching costs

High switching costs can dramatically increase buyer power.

XII. Six Important Case Laws

1. Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co., 549 U.S. 312 (2007)

This is the most important U.S. Supreme Court authority for the modern concept of buyer-side market power.

The Court recognized monopsony as the buying-side counterpart of monopoly and addressed predatory bidding.

The case concerned allegations that Weyerhaeuser bid up the price of sawlogs to disadvantage competing purchasers and thereby obtain greater monopsony power.

Principle

Buyer-side conduct can constitute anticompetitive conduct even though it involves increasing, rather than decreasing, the price paid for inputs.

Relevance to autonomous procurement

An AI agent could theoretically be instructed to:

1. identify rival buyers;
2. determine their dependence on a particular input;
3. bid aggressively;
4. cause rival buyers to exit;
5. subsequently exploit enhanced purchasing power.

Thus, autonomous procurement should not be assessed solely through the question:

"Does the AI obtain low prices?"

The relevant question is:

How does the procurement strategy affect competitive conditions among buyers and suppliers?

2. Mandeville Island Farms, Inc. v. American Crystal Sugar Co., 334 U.S. 219 (1948)

This case involved sugar refiners that allegedly agreed on uniform prices to be paid to sugar-beet growers.

The Supreme Court held that price fixing by purchasers could violate the Sherman Act even though the injured parties were sellers rather than consumers.

The Court emphasized that the refiners controlled the practical purchasing outlets available to growers.

Principle

Antitrust law is not restricted to protecting consumers from seller-side price increases.

Buyer-side price fixing can itself constitute an unlawful restraint.

Autonomous-agent relevance

If competing buyers deployed agents that independently—or pursuant to a common arrangement—fixed maximum purchase prices, the fact that the conduct occurs on the procurement side would not immunize it.

3. National Collegiate Athletic Association v. Alston, 594 U.S. 69 (2021)

Alston is an important modern monopsony authority.

The Supreme Court addressed restrictions imposed by the NCAA on compensation and education-related benefits for certain student-athletes.

The Court applied ordinary antitrust principles and rejected the argument that the NCAA's special structure placed the challenged restraints outside normal competition analysis.

The underlying record recognized NCAA monopsony power in the relevant labor market, although the Supreme Court did not itself resolve the market-definition issue because the parties did not contest it.

Principle

Buyer-side restraints can violate Section 1 even where the buyers operate through a coordinated institutional structure.

Autonomous-agent relevance

A procurement consortium using a common autonomous agent should therefore not assume that technological or organizational integration removes antitrust scrutiny.

4. Todd v. Exxon Corp., 275 F.3d 191 (2d Cir. 2001)

Todd is important for the relationship between information exchange and buyer-side market power.

The case involved alleged exchanges of compensation information in an employment market. The Second Circuit emphasized factors such as:

• market power;
• susceptibility of the market to coordination;
• nature of information exchanged;
• market-wide effects. 

Principle

Information that appears commercially useful may become anticompetitive when exchanged in a concentrated market capable of coordinated behavior.

Autonomous-agent relevance

Procurement AI can process vastly more information than human procurement teams.

Consequently, competition authorities may examine:

• what data the agent receives;
• who supplied it;
• whether competitors can access it;
• whether supplier-specific information is shared;
• whether the system facilitates coordinated procurement behavior.

5. In re Dairy Farmers of America, Inc. Cheese Antitrust Litigation, 767 F. Supp. 2d 880 (N.D. Ill. 2011)

The litigation involved allegations that Dairy Farmers of America and others manipulated purchases in cheese and milk-futures markets.

The court discussed monopsony as market power on the buying side and examined allegations that large purchases were used to obtain control over relevant markets and influence prices.

Principle

Large-scale purchasing can become competition-law relevant where purchasing behavior is allegedly used to obtain or exercise market power rather than merely to satisfy legitimate procurement needs.

Autonomous-agent relevance

An autonomous agent could execute thousands of transactions rapidly.

That creates a novel evidentiary question:

Were purchases genuinely made to satisfy business demand, or did the procurement strategy deliberately manipulate the purchasing market?

The answer would depend on evidence concerning purpose, market effects, alternative explanations and the agent's instructions.

6. FTC v. Staples, Inc. / Staples–Office Depot

The Staples–Office Depot litigation illustrates another important dimension: buyer concentration and the loss of competition in procurement-intensive business markets.

The FTC challenged the proposed merger because Staples and Office Depot competed closely for large business customers, with the agency alleging that elimination of this competition would lead to higher prices and reduced quality. The federal district court granted a preliminary injunction, and the transaction was abandoned.

Although the case primarily concerned downstream competition rather than a pure monopsony claim, it demonstrates the importance of contract-level competition and sophisticated purchasing arrangements.

Autonomous-agent relevance

An autonomous procurement system can make competition increasingly occur through:

• individualized bids;
• contract terms;
• supplier-specific pricing;
• service bundles;
• delivery commitments;
• quality metrics.

Therefore, simple market-share analysis may fail to capture the actual competitive constraints affecting procurement contracts.

XIII. Additional Relevant Authority: Aetna–Prudential

The DOJ's Aetna-Prudential merger analysis provides particularly useful buyer-power principles.

The DOJ examined whether consolidation would allow Aetna to depress prices paid to physicians in Houston and Dallas.

It emphasized:

• supplier switching costs;
• individual negotiations;
• supplier dependence;
• buyer ability to discriminate among suppliers;
• reduction in quantity and quality of services.

This is highly relevant to autonomous procurement because AI can make individualized supplier bargaining far more scalable.

XIV. Efficiency Defence

Autonomous procurement can generate legitimate efficiencies.

For example:

Transaction-cost efficiency

AI reduces procurement administration.

Search efficiency

AI rapidly identifies suitable suppliers.

Inventory efficiency

Better forecasting reduces over-ordering.

Logistics efficiency

Purchasing can be synchronized with transportation.

Quality efficiency

Supplier performance can be continuously evaluated.

Fraud prevention

Automated verification can reduce procurement fraud.

These benefits should be distinguished from savings generated by suppressing competitive supply.

XV. The "False Efficiency" Problem

Suppose an autonomous agent reports:

"Procurement costs decreased 15%."

That does not necessarily establish a procompetitive efficiency.

The 15% reduction could result from:

Scenario A — genuine efficiency

Better forecasting → lower inventory → lower logistics cost → lower total cost.

or:

Scenario B — monopsony

Buyer threatens suppliers → suppliers accept lower prices → suppliers reduce output → weaker suppliers exit.

The numerical procurement saving is identical, but the competitive consequences are completely different.

The DOJ's Aetna-Prudential analysis expressly distinguishes genuine efficiencies from reductions in input prices caused by monopsony power.

XVI. Long-Term Supplier Effects

Autonomous procurement may create a particularly important dynamic competition problem.

Stage 1

AI aggressively selects the lowest-cost suppliers.

Stage 2

Higher-cost suppliers lose contracts.

Stage 3

Those suppliers reduce investment.

Stage 4

Some suppliers exit.

Stage 5

Remaining suppliers become fewer and more dependent.

Stage 6

Buyer obtains even greater bargaining power.

Stage 7

AI demands further price reductions.

This creates a potential buyer-power feedback loop:

Procurement optimization → supplier exit → supplier concentration → greater buyer power → further procurement pressure.

The European Commission's purchasing-market guidance specifically recognizes that monopsony/oligopsony power can reduce suppliers' incentives to invest and innovate and can reduce the number of suppliers in the market.

XVII. Autonomous Procurement and Small Suppliers

The problem may be particularly acute for SMEs.

A major buyer's autonomous agent may impose:

• electronic invoicing;
• cybersecurity requirements;
• insurance requirements;
• minimum production capacity;
• rapid delivery;
• complex data reporting;
• extended payment periods.

Each requirement may appear individually reasonable.

But collectively, they can create substantial barriers to participation.

Therefore, competition authorities may need to examine not only price effects, but also:

• supplier entry;
• supplier exit;
• innovation;
• quality;
• diversity of supply;
• contractual dependency.

XVIII. Exclusive Procurement by Autonomous Agents

An AI system may recommend:

"Use one supplier for 85% of requirements."

The rationale may be:

• lower price;
• greater reliability;
• volume discounts;
• standardized quality.

However, very high allocation can create dependency.

If the buyer subsequently requires exclusivity, the arrangement may raise questions involving:

• exclusive dealing;
• foreclosure;
• raising rivals' costs;
• supplier access;
• vertical restraints;
• abuse of dominance.

The legality depends on market structure, duration, coverage, justification and competitive effects.

XIX. Autonomous Procurement and Vertical Integration

A powerful purchaser may eventually acquire strategically important suppliers.

The sequence can be:

AI identifies supplier dependency

↓

Buyer increases procurement share

↓

Supplier becomes financially dependent

↓

Buyer acquires supplier

↓

Rival buyers lose access

↓

Vertical foreclosure becomes possible

Thus procurement algorithms can potentially influence not only purchasing prices but also market structure.

XX. Competition-Law Liability for Autonomous Agents

An autonomous agent itself is generally a technological instrument rather than the legal person responsible for competition-law conduct.

Potential responsibility may instead attach to:

1. Corporate management

Where management knowingly designs or authorizes an anticompetitive procurement strategy.

2. Compliance personnel

Where known risks are ignored despite adequate warning.

3. Procurement executives

Where they intentionally use the system to coordinate or exclude competitors.

4. Corporate entity

The company may bear responsibility for conduct carried out through its systems.

5. Multiple firms

Where agents are connected through a common procurement arrangement or coordinated architecture.

The central issue is therefore not:

"Did the AI make the decision?"

but:

"What human or corporate conduct caused the system to operate in this way, and what were the competitive effects?"

XXI. Evidence and Algorithmic Auditability

Competition authorities may increasingly require examination of:

• source code;
• procurement objectives;
• optimization functions;
• training data;
• supplier-ranking models;
• transaction logs;
• bid histories;
• override records;
• human instructions;
• model outputs;
• supplier exclusion rules;
• communications concerning algorithm design.

This creates an important distinction between:

AI autonomy and AI opacity.

A company cannot necessarily rely on the complexity of an AI system to explain away procurement conduct.

XXII. Compliance Framework

Companies deploying autonomous procurement agents should consider a competition-law control architecture.

1. Market-power assessment

Determine:

• buyer market share;
• supplier alternatives;
• supplier concentration;
• switching costs;
• entry barriers.

2. Procurement-agent rules

Prohibit the agent from:

• coordinating with competing buyers;
• exchanging confidential competitor information;
• intentionally excluding suppliers without legitimate justification;
• manipulating supplier markets.

3. Human oversight

Require approval for:

• major supplier exclusions;
• unusual purchasing patterns;
• coordinated procurement;
• substantial changes in supplier allocation.

4. Audit logs

Maintain records of:

• agent instructions;
• decisions;
• supplier rankings;
• bids;
• pricing recommendations;
• overrides.

5. Dynamic monitoring

Monitor whether AI optimization produces:

• supplier exit;
• excessive concentration;
• systematic price suppression;
• declining output;
• declining innovation.

XXIII. Competition-Law Test

A useful analytical framework is:

Step 1 — Identify the purchasing market

What input or service is being purchased?

Step 2 — Measure buyer concentration

Who are the major buyers?

Step 3 — Examine supplier alternatives

Can suppliers realistically switch buyers?

Step 4 — Determine buyer power

Can the autonomous procurement system materially affect supplier terms?

Step 5 — Identify conduct

Is the agent:

• negotiating?
• excluding?
• allocating?
• coordinating?
• imposing exclusivity?
• exchanging information?

Step 6 — Examine effects

Does the conduct cause:

• supplier exit?
• lower output?
• reduced innovation?
• reduced quality?
• exclusion of rival buyers?
• higher downstream prices?

Step 7 — Consider efficiencies

Are savings caused by genuine technological efficiencies rather than suppression of competitive supply?

Step 8 — Evaluate remedies

Possible remedies could include:

• data-access restrictions;
• procurement firewalls;
• supplier non-discrimination;
• interoperability;
• human approval requirements;
• limits on exclusivity;
• algorithmic auditing;
• divestiture in merger cases.

XXIV. Six-Case Synthesis

XXV. Conclusion

Autonomous procurement agents can transform buyer power from a relatively episodic bargaining advantage into a continuous, data-driven form of market power.

The principal competition-law concern is not that AI obtains lower prices. Lower procurement prices may reflect legitimate efficiencies.

The concern arises when autonomous procurement becomes capable of systematically suppressing competitive supply, excluding suppliers or rival buyers, facilitating coordination, or reinforcing concentration.

The most significant risk can therefore be conceptualized as:

Autonomous procurement + concentrated purchasing + supplier dependence + algorithmic optimization = potential monopsony/oligopsony risk.

The existing case law already establishes the foundational principles. Weyerhaeuser recognizes monopsony as the buying-side analogue of monopoly; Mandeville Island Farms confirms that purchaser-side price fixing can fall within antitrust law; Alston demonstrates that buyer-side restraints can be tested under ordinary competition principles; Todd provides an important framework for information-exchange concerns; and the Dairy Farmers litigation illustrates how purchasing behavior itself can become relevant to market-power analysis.