This Is Antitrust Risk

Coordination among competitors runs close to a legal third rail, and it should. Stand back far enough from a network that structures demand across many buyers and routes it among many suppliers, and you are looking at exactly the thing the antitrust laws were written to watch: competitors brought into contact around the terms of trade. A founder's counsel who raises it is doing the job correctly, and the question deserves a better answer than an assurance of good intentions. I have spent this entire series arguing for coordination and against collusion, and the line between the two is not a rhetorical convenience. It is enforceable, and people go to prison for crossing it. Claim: a coordination layer stays lawful only when it structures demand and never touches price — no price-fixing, no bid-rigging, no group boycotts, no coercive exclusivity — with neutral, auditable routing and a hard wall between organizing the work and setting what it costs; it turns into real antitrust risk the moment it pools competitors' forward pricing or output intentions, or starts steering who is allowed to deal with whom.

Here is the line as plainly as I know how to draw it. Helping a buyer describe a job well enough to quote it, and helping capable shops find that job, is coordination. Collecting what competitors intend to charge, or arranging in advance who wins, is collusion. Same wires, same participants, opposite sides of a felony.

The bright lines

They are not subtle, and they carry criminal weight. Section 1 of the Sherman Act declares "every contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade" illegal, and a violation is a felony punishable by corporate fines up to $100 million and individual imprisonment up to ten years. The Department of Justice treats price-fixing and bid-rigging as per se violations, which in its own words means they "cannot be justified" by arguing that the agreed prices were reasonable or that the agreement prevented ruinous competition. Price-fixing is any agreement among competitors to raise, fix, or maintain price. Bid-rigging is competitors settling in advance who submits the winning bid. A system that let suppliers see and converge on each other's numbers would not be operating near that line. It would be over it, with a log file waiting for the prosecutor.

The objection at its strongest

The objection has teeth right now, because regulators have started treating the software itself as the instrument of the conspiracy. In 2024 the Department of Justice and eight states sued RealPage under Sections 1 and 2 of the Sherman Act, alleging that competing landlords fed nonpublic, competitively sensitive pricing data into a shared algorithm that helped align their rents. The Attorney General's framing is the sentence every platform builder should keep somewhere visible: "Using software as the sharing mechanism does not immunize this scheme from Sherman Act liability." The Deputy Attorney General was blunter — "training a machine to break the law is still breaking the law." Those are allegations, not findings, and they should be read that way. What matters for anyone building is the enforcement theory, and it arrived after the agencies had already withdrawn decades-old guidance that once blessed certain competitor information exchanges: DOJ rescinded the old health-care safety-zone statements in February 2023 as "overly permissive" on information sharing, and the FTC followed that July. If your legal comfort was inherited from that era, it has expired.

The line the Court drew a century ago

The doctrine already marks the ground a coordination layer has to stand on, and it marked it long before anyone had a server. In American Column & Lumber (1921) the Supreme Court struck down a trade-association data-exchange plan that curtailed production and raised prices, calling it an unlawful combination. Four years later, in Maple Flooring, the Court upheld an association that "openly and fairly" gathered and disseminated information on past costs, production volumes, and prices already transacted, so long as it reached no agreement on future prices or output. That is the governance thesis of this series, handed down in 1925. Aggregated, historical, evidence-based information with no agreement on forward price or output is lawful; channeling competitors toward a coordinated number is not. Group boycotts sit under the same discipline, and the layer must never become the venue for one — especially if it ever holds market power, since a network with power that steers its members away from a disfavored supplier is inviting the per se treatment the Court reserves for exactly that conduct.

The governed design

So you build the bright lines into the architecture rather than the terms of service. Routing criteria are neutral and auditable, so a regulator or a member can see why a given supplier appeared on a given job, and none of those criteria is another competitor's forward price. Suppliers price independently and never see each other's quotes; the system structures the request, not the number. It coordinates no output, brokers no market division, and enforces no coercive exclusivity — buyers and suppliers stay free to transact off-network, and nothing punishes them for doing it. Revenue comes from structuring work and verifying trust rather than from touching the spread, which removes the incentive to nudge a price before anyone has to resist it. The control point is the separation between demand-structuring and price-setting, and it has to be verifiable from the audit log. A policy is what you say. A log is what you did.

Operational test

This is testable by audit, which is the only kind of test worth much here. Can an outside reviewer reconstruct the routing criteria for any award and confirm that no competitor's forward pricing or output plan was an input? Do suppliers set their own prices with no visibility into rivals' quotes? Is there any mechanism anywhere — a feature, a report, a customer-success habit that grew up on its own — by which the network could steer members to boycott a supplier or carve up a market? Are buyers and suppliers demonstrably free to deal off-platform? Drift on any one of those and you are accumulating exposure regardless of what you intended.

Implications

If coordination slides quietly into price alignment, the critic wins the argument and the layer deserves the prosecution it will get, because a cartel with an API is still a cartel. Hold the century-old line — structure demand, refuse to touch price — and the same machinery does what the antitrust laws actually favor, which is make a market more competitive by helping capable suppliers get seen by buyers who never knew they existed. This one is load-bearing well beyond any single company. A coordination layer that ends up in an antitrust complaint does not just lose a case; it hands every skeptic of reindustrialization a decade of free arguments. The practical failure mode is collusion. The next objection drops the law entirely and comes at this from engineering pragmatism, and it is the one every standards veteran raises first: fine, it is legal and it is right, but standards take too long.

Questions to Ask

1. Can we reconstruct, for any award, why each supplier was routed, using only neutral criteria?
2. Does any competitor's forward price or output intention ever enter our system?
3. Do suppliers price independently, without visibility into each other's quotes?
4. Is there any path by which the network coordinates a boycott or divides a market?
5. Are buyers and suppliers free to transact off-platform without penalty?
6. Where does structuring demand end and setting price begin, and is that boundary in the audit log?