Algorithmic Price Fixing After Cornish-Adebiyi: The Third Circuit Revives Hub-and-Spoke Claims Against Shared Pricing Software

Introduction

On July 29, 2026, the United States Court of Appeals for the Third Circuit issued its decision in Cornish-Adebiyi v. Caesars Entertainment, Inc., No. 24-3006, reversing the District of New Jersey’s dismissal of a putative class action alleging that Atlantic City casino-hotels and their algorithmic software provider, Cendyn Group, LLC, conspired to fix hotel room prices in violation of Section 1 of the Sherman Act. The plaintiffs allege that the casino-hotel defendants fed non-public, real-time room pricing and occupancy data into Cendyn’s “Rainmaker” dynamic pricing software, which then allegedly pooled that data across competing properties to generate coordinated price recommendations that the hotels followed approximately 90% of the time. The Complaint alleges this arrangement is a hub-and-spoke conspiracy—with Rainmaker as the hub and the casino-hotels as the spokes—resulting in supracompetitive room rates during a period of declining occupancy. The District Court (Judge Karen M. Williams) dismissed the Consolidated Amended Complaint under Rule 12(b)(6), finding the complaint “lacks a rim”—i.e., a horizontal agreement among the casino-hotels.

The Third Circuit’s reversal arrives on the heels of the Ninth Circuit’s August 2025 opinion in Gibson v. Cendyn Group, LLC, 148 F.4th 1069 (9th Cir. 2025), which affirmed dismissal of a similar Sherman Act challenge to Las Vegas hotel room pricing via the same software. At first glance, the two decisions appear to create a circuit split on the viability of algorithmic hub-and-spoke price-fixing claims. However, the decisions may not be in conflict: the divergent outcomes may lie in the presence in one and absence in the other of allegations that the pricing software pooled non-public competitor data—a factual allegation pleaded in Cornish-Adebiyi but not in Gibson.

What this decision means for your business

The intense focus on algorithmic pricing tools by both the private antitrust bar and enforcement authorities continues to require a 360-degree risk assessment of third-party software tools (including proprietary algorithms and AI solutions) that assist in competitive decision-making. The Third Circuit’s decision in Cornish-Adebiyi reinforces some of the key risk factors that must be considered. Most importantly: does your software use non-public data pooled from other subscribers—potentially, your competitors—as an input into the model? And do you know what your terms of service say about this? See below for a checklist of steps you can take to better understand your company’s risk profile.

Background

Plaintiffs in Cornish-Adebiyi are consumers who rented guest rooms from five Atlantic City casino-hotels—Caesars, Harrah’s, Tropicana, MGM/Borgata, and Hard Rock—during a class period beginning no later than June 28, 2018. They alleged that these competitors each subscribed to Cendyn Group’s “Rainmaker” algorithmic pricing software which, according to Plaintiffs, operated as follows:

• Each casino-hotel provided its current, non-public room pricing and occupancy data to Rainmaker.
• Rainmaker processed each hotel’s proprietary data alongside competitors’ data and generated “optimal” room rate recommendations.
• These recommendations were automatically uploaded into each hotel’s property management system.
• Casino-hotels allegedly accepted Rainmaker’s pricing recommendations approximately 90% of the time.
• Cendyn constrained overrides by requiring special permissions and scoring hotels on deviation frequency.

Plaintiffs further alleged that during the class period, room rates increased 22–25% while occupancy declined 5–8%—a pattern that Plaintiffs claimed was contrary to the unique economic incentives of casino-hotels, where rooms have historically served as loss leaders to attract patrons to the casino floor.

In its dismissal, the District Court relied heavily on the District of Nevada’s decision in Gibson v. Cendyn Group, LLC, which dismissed a similar complaint on similar facts. The Ninth Circuit subsequently affirmed Gibson, and the Supreme Court denied certiorari in April 2026.

Court’s Holding in the Third Circuit Case

The Third Circuit reversed and remanded, holding that the plaintiffs’ allegations plausibly support an inference that the casino-hotel defendants conspired to fix room prices through Cendyn’s Rainmaker software, in violation of Section 1 of the Sherman Act. The court concluded that the Consolidated Amended Complaint states a claim for a hub-and-spoke price-fixing conspiracy sufficient to survive a motion to dismiss.

The court began by explaining how pricing algorithms generally work, a basic but useful definition for the kinds of technologies that may be implicated by its holding:

Dynamic pricing algorithms are prescriptive AI systems that automate pricing for sellers in marketplaces. These algorithmic programs are typically trained on data specific to a vendor or market to suggest (and in some cases implement) prices. … [T]hese algorithms have become a common feature of many markets, including ide-sharing apps, airlines, and major online retailers and thus shape the marketplace for millions of Americans every day. (Cleaned up.)

The court then addressed six key issues:

1. Parallel Conduct

The Third Circuit identified two theories of parallel conduct that it found sufficient to support the conspiracy inference. First, the casino-hotels’ alleged contemporaneous and continuous use of Rainmaker as a “shared pricing agent,” to which they supposedly delegated pricing decisions, constituted parallel behavior. Second, the alleged synchronous movement of rising room prices and falling occupancy during the class period— according to plaintiffs, a departure from both historical practice and the purported economic logic of the casino-hotel industry—provided further evidence of parallelism. The court held that these theories, at least in combination, were sufficient at the pleading stage.

2. Plus Factors

Beyond parallel conduct, the court identified several “plus factors” that, when considered together, moved the plaintiffs’ allegations from merely consistent with independent action to plausibly suggestive of a conspiracy:

1. Motive to conspire: The court found that Post-2008 financial recession hardship in the Atlantic City casino-hotel market, combined with high market concentration and significant barriers to entry, created a strong economic incentive for coordinated pricing.
2. Actions contrary to economic self-interest: As alleged by plaintiffs, the casino-hotels failed to lower room rates despite declining occupancy, although casino-hotels have an extra incentive to fill rooms to drive casino floor revenue, making sustained high pricing amid falling occupancy contrary to the expected competitive “race to the bottom.”
3. Non-economic evidence of a traditional conspiracy: The allegations of a de facto exchange of competitively sensitive data through the software; supposed opportunities to collude at industry events; each hotel’s alleged knowledge of its competitors’ Rainmaker relationships; the purported sudden change in business practices; and a former Cendyn VP of Data Science/Analytics allegedly encouraging hotels to avoid the “race to the bottom” were all suggestive of a conspiracy.

3. Pleading Standards: No Requirement to Explain the Algorithm’s Mechanics

The Third Circuit held that plaintiffs need not plead the technical details of how Cendyn’s proprietary algorithm functions in order to survive a motion to dismiss. The court reasoned that requiring such detail would be “tantamount to expecting Plaintiffs to explain how Cendyn’s proprietary software works without affording the discovery required to do that” and that such specificity is “neither required nor appropriate” at the pleading stage.

4. Retained Override Authority is Not a Defense

The court rejected the defendants’ argument that the casino-hotels’ retained ability to override Rainmaker’s pricing recommendations defeated the conspiracy claim. Instead, the court likened the pricing recommendation to an agreed “starting point” for competitive pricing decisions. Given the alleged 90% compliance rate and practical constraints on exercising overrides, such as limited override permissions and override-frequency scoring, the court held that the retained override authority did not preclude a plausible inference of agreement.

5. Staggered Adoption Does Not Defeat Parallelism

The District Court had relied on Burtch v. Milberg Factors, Inc., 662 F.3d 212 (3d Cir. 2011), to hold that the casino-hotels’ adoption of Rainmaker over a lengthy period (2004–2018) defeated an inference of parallel conduct. The Third Circuit distinguished Burtch, which involved 18 months of divergent (not parallel) conduct. Here, the relevant parallel conduct was not the timing of adoption but rather the casino-hotels’ continuous deployment of and delegation to Rainmaker, combined with the alleged sudden synchronous price and occupancy movements during the class period. The court further noted that AI pricing tools can adapt and evolve over time, such that the “opportune time” for algorithmic collusion to take effect could arise well after initial adoption.

6. The “Guy Named Bob” Test

In a passage that is likely to resonate in future algorithmic-pricing litigation, the Third Circuit invoked former FTC Acting Chair Maureen K. Ohlhausen’s formulation: “Is it ok for a guy named Bob to collect confidential price strategy information from all the participants in a market, and then tell everybody how they should price? If it isn’t ok for a guy named Bob to do it, then it probably isn’t ok for an algorithm to do it either.” The court concluded that the plaintiffs plausibly alleged that what occurred here would constitute illegal collusion if performed by a human intermediary—and that the analysis does not change simply because the alleged “hub” is named “Rainmaker” rather than “Bob.”

The court acknowledged the amicus concern raised by the International Center for Law & Economics (ICLE) that this reasoning could chill the independent, industry-wide use of common pricing software, like gas stations using identical spreadsheet formulas. The court distinguished that scenario, however, because the Consolidated Amended Complaint, it said, alleges more than mere independent use of identical software: it alleges the actual exchange of non-public data, consolidated in a hub and shared back to competitors as coordinated price recommendations.

Navigating Cornish-Adebiyi and Gibson

At first glance, the Third Circuit’s decision in Cornish-Adebiyi (reversing dismissal) and the Ninth Circuit’s decision in Gibson (affirming dismissal) appear to create a circuit split on the viability of algorithmic hub-and-spoke price-fixing claims. A closer reading reveals that the decisions address a different set of allegations and, when read together, articulate a coherent framework for analyzing such claims.

Gibson Did Not Allege Data Pooling

A critical distinction is that the plaintiffs in Gibson did not allege that Cendyn pooled, shared, or used one hotel’s confidential information in generating pricing recommendations for another hotel. The Ninth Circuit noted this gap, in a footnote, stating that its “analysis might change if Plaintiffs had alleged that Cendyn shared the confidential information of each competing hotel among the licensees.”

Because the Gibson plaintiffs abandoned their hub-and-spoke (horizontal agreement) claim on appeal and did not allege data pooling or sharing among competitors, the Ninth Circuit treated the individual Cendyn licensing agreements as “ordinary sales contracts” for software services that did not restrain the hotels’ ability to compete. Parallel conduct—independently licensing the same pricing software—plus rising prices, without more, was insufficient to state a Section 1 claim.

Cornish-Adebiyi Alleged Precisely What Gibson Did Not

By contrast, the Cornish-Adebiyi complaint alleges that Rainmaker pooled each casino-hotel’s non-public pricing and occupancy data with its competitors’ non-public data to generate the coordinated pricing recommendations. This is the type of conduct the Gibson court theorized “might change” its analysis.

A Potential Unifying Framework

Read together, Cornish-Adebiyi and Gibson suggest a dividing line: when plaintiffs allege that a pricing algorithm pools or exchanges non-public competitor data to generate coordinated pricing outputs, a Section 1 Sherman Act claim is more likely to survive dismissal, and plaintiffs may proceed to try and prove such allegations. Conversely, the independent, non-pooled use of common pricing software—without a pled horizontal agreement or data-sharing arrangement among competitors—is less likely to state a claim. The key question for plaintiffs and defendants alike is whether the software serves as a mere tool or as a conduit for the exchange of competitively sensitive information among rivals.

Key Takeaways

• Antitrust Litigation Risk Is Material. Companies that use or provide third-party algorithmic or AI-powered pricing tools similar to those at issue in these cases face significant Section 1 exposure, particularly in concentrated markets characterized by repeated competitor interactions and high barriers to entry. Cornish-Adebiyi stands for the proposition that hub-and-spoke claims premised on shared algorithmic pricing infrastructure can survive a motion to dismiss when non-public data pooling is adequately pled.
• Pleading Standards: The Gibson/Cornish-Adebiyi Dividing Line. Plaintiffs can survive dismissal without pleading the proprietary algorithm’s technical mechanics—Cornish-Adebiyi holds that such detail is a matter for discovery. However, plaintiffs must plead more than parallel use of a common software vendor. The presence or absence of pooled non-public competitor data is the critical dividing line between a viable claim (Cornish-Adebiyi) and an insufficient one (Gibson).
• Due Diligence Implications. In M&A transactions and commercial due diligence, counsel should specifically probe whether a target company’s pricing software pools or shares competitor data. The Cornish-Adebiyi/Gibson framework makes the answer to that question critical in assessing the target’s antitrust litigation exposure.
• Compliance Programs Must Address Algorithmic Pricing. Businesses should proactively assess whether their pricing tools and vendor relationships create hub-and-spoke exposure. “Override authority” alone will not shield a company from liability under Cornish-Adebiyi—the focus must be on the architecture of data flows and the vendor’s use of competitor information.

What This Means for Your Business

In light of Cornish-Adebiyi and Gibson, companies that use algorithmic pricing tools—or that provide such tools to competitors in a market—should take the following steps:

1. Conduct an Algorithmic Pricing Audit. Review all algorithmic and AI-powered pricing tools currently in use to confirm that they do not incorporate, pool, or rely upon non-public competitor data in generating pricing recommendations. Under the Cornish-Adebiyi/Gibson framework, the use of pooled non-public competitor data is the critical fact that transforms a lawful software license into potential hub-and-spoke collusion.
2. Evaluate Override Processes. Although Cornish-Adebiyi holds that retained override authority alone is not a defense, companies should still document and exercise independent override authority meaningfully and regularly. The 90% compliance rate alleged in Cornish-Adebiyi was a significant factor in the court’s analysis. Companies should ensure that their pricing personnel exercise genuine independent judgment and that override processes are not merely nominal.
3. Review Vendor Contracts. Scrutinize contracts with pricing software vendors to understand how the vendor sources, aggregates, and uses client data, and any broadly worded rights or licenses to use your data that may exist in your license terms. Negotiate contractual protections including data segregation clauses, non-pooling covenants, and representations that the vendor does not use one client’s confidential data in generating recommendations for a competitor. Review whether existing contracts permit the vendor to pool data and, if so, seek amendments.
4. Assess Data Architecture. Ensure that pricing software relies only on the company’s own proprietary data and publicly available market data—not pooled non-public competitor data. This is the bright line drawn by Gibson and Cornish-Adebiyi: algorithms that draw on a company’s internal data and public information to optimize pricing operate in safer territory than those that aggregate and redistribute competitors’ confidential information.
5. Update Compliance Training. Revise antitrust compliance training programs to address the Cornish-Adebiyi decision and the risks of algorithmic hub-and-spoke arrangements. Training should cover not only traditional price-fixing scenarios but also the antitrust risks of sharing competitively sensitive data with third-party software platforms—even where the “agreement” is algorithmic rather than verbal. Ensure that engineers, data scientists, and product managers—not only commercial personnel—receive this training.
6. Consider Other Solutions Besides “Pricing” Software. Although today’s cases and enforcement actions are focused on pricing and revenue management software, the same logic that underpins these suits could apply to the use of third-party software to assist in other aspects of competitive decision-making, including production decisions, sourcing, hiring, and employee compensation and benefit.
7. Engage Antitrust Counsel Proactively. Consult with antitrust counsel before adopting, renewing, or modifying algorithmic pricing tools or vendor relationships. The legal landscape is evolving rapidly, and the risk analysis under Cornish-Adebiyi and Gibson turns on technical details of data architecture that require careful assessment.

Looking Ahead

Potential Supreme Court Review

The tension between Cornish-Adebiyi and Gibson—even if reconcilable on their facts—may prompt the Supreme Court to grant certiorari and address algorithmic hub-and-spoke liability under Section 1. Although the Supreme Court denied certiorari in Gibson in April 2026 (No. 25-1109, 2026 WL 1052046), Cornish-Adebiyi may generate a new petition, or prompt the DOJ to weigh in as amicus, particularly as similar cases proceed in other jurisdictions.

DOJ Enforcement Priority: “Old Crime, New Code”

The DOJ has signaled unambiguously that algorithmic collusion is an enforcement priority. On May 14, 2026, DOJ Acting Deputy AAG Daniel Glad delivered remarks titled “Old Crime, New Code” at the Antitrust West Coast Conference, declaring that “software cannot launder collusion” and that the DOJ will pursue criminal Sherman Act liability where competitors use algorithms, SaaS platforms, or large language models to coordinate pricing. Key elements of DOJ’s posture include:

• RealPage as the “paradigm case”: The DOJ obtained a consent judgment against RealPage, Inc., whose software allegedly gathered competitively sensitive non-public data from competing landlords and fed it through a shared algorithm to produce coordinated rental pricing recommendations. The consent judgment requires RealPage to limit the granularity of its pricing outputs and submit to a court-appointed monitor. The DOJ has identified RealPage as the paradigm for hub-and-spoke algorithmic enforcement.
• Whistleblower Rewards Program: In January 2026, the DOJ announced its first-ever antitrust whistleblower reward—$1 million—for information leading to bid-rigging charges in online auctions. This creates a “second race” alongside the traditional corporate Leniency Program, incentivizing individual employees to report algorithmic collusion arrangements.
• Updated Compliance Program Evaluation: In November 2024, the DOJ updated its Evaluation of Corporate Compliance Programs in Criminal Antitrust Investigations to include specific risk-assessment questions directed at AI and algorithmic tools—including whether a company’s risk assessment addresses new technologies, whether antitrust risk is assessed when deploying pricing tools, whether compliance personnel are involved in those decisions, and what mitigation steps are in place. As AAG Glad stated: “A company cannot say it has a mature AI governance program if no one is asking whether the tool facilitates coordination with competitors.”
• Broad scope of personal liability: The DOJ has indicated that a wide range of personnel—including not only executives and sales staff but also engineers, data scientists, product managers, and account managers—may face personal criminal liability for algorithmic collusion schemes.

Broader Impacts on Algorithmic Markets

The Cornish-Adebiyi and Gibson decisions, together with the DOJ’s enforcement posture and the RealPage precedent, suggest a framework likely to govern algorithmic pricing cases going forward. As it stands, courts seem more likely accept the independent adoption of common, non-data-pooling pricing software (as in Gibson) but reject arrangements where a shared platform allegedly pools non-public competitor data to generate coordinated pricing outputs (as in Cornish-Adebiyi and the RealPage matter). This area of law remains rapidly evolving and warrants ongoing monitoring by companies in any industry that uses algorithmic pricing tools.

* * *

For more information on the issues discussed in this Legal Update, please contact the authors or your regular contact at the firm.