The Antitrust Landscape for Algorithmic Pricing: Key Decisions and Takeaways

Private litigation and Department of Justice Antitrust Division (“DOJ”) enforcement activity regarding common algorithmic pricing tools have increased in recent years. Specifically, private plaintiffs and the DOJ have brought cases alleging that defendants are violating the Sherman Act by using software systems that analyze supply and demand data and recommend or—in some cases—set prices based on those data.  These cases allege that competitors that use a common algorithmic pricing tool or submit data to common intermediary are able to maintain prices above a competitive level as a result. This client alert highlights key recent decisions and distills practical takeaways from those rulings. [1]

Algorithmic pricing cases are highly fact- and allegation-specific, and the case law is developing. We will continue to monitor trends as additional courts and enforcement authorities weigh in and cases progress. However, certain practical takeaways are emerging from the early decisions:

• Per se Some courts have treated algorithmic pricing that relies on competitors’ sensitive, non-public information as per se illegal. Other courts have applied the rule of reason to this conduct or deferred on ruling which standard applies.
• Data source matters. Courts have generally rejected claims based only on public data or the user’s own data.
• Model training vs. pricing recommendations. Courts and regulators may distinguish between using data to train models and using data to generate pricing recommendations.
• Adoption and uniformity. Courts have credited allegations of uniform pricing and high adoption rates in denying motions to dismiss.
• Independent pricing mechanisms. Courts and regulators may evaluate safeguards to preserve independent pricing, both at the motion to dismiss stage and when crafting remedies.
• Plus factors. Courts look at whether alleged plus factors point more toward coordination or unilateral conduct.

Circuit Court Decisions

Two circuit courts have ruled on algorithmic pricing cases. On July 29, 2026, the Third Circuit, in Cornish-Adebiyi v. Caesars Entertainment, Inc.,[2] reversed a district court decision dismissing antitrust claims against Atlantic City casino-hotels that used a common pricing software. As discussed in our memorandum on the decision, the Third Circuit held that plaintiffs plausibly alleged a horizontal price-fixing agreement among the casino-hotels based on their parallel, continuous delegation of pricing to the same algorithm, which processed their nonpublic data; synchronized price increases despite falling occupancy; and plus factors, including conduct contrary to each hotel’s economic self-interest.

On August 15, 2025, the Ninth Circuit affirmed a lower court’s dismissal of similar claims in Gibson v. Cendyn Group, LLC. [3] The Ninth Circuit held that the hotels’ individual licensing agreements did not restrain trade, because the software did not require hotels to adopt its recommendations and plaintiffs did not allege that the algorithm pooled one hotel’s confidential data into recommendations for its competitors. Plaintiffs had abandoned their hub-and-spoke claim on appeal, so the court did not address it.

Per Se or Rule of Reason

In any antitrust case, an important factor is whether the per se or rule of reason standard applies. Under the per se standard, certain conduct is deemed so inherently anticompetitive that it is automatically illegal once proven, with no inquiry into anticompetitive effects, market power, or business justifications. Some federal courts that have allowed algorithmic pricing cases involving exchange of sensitive, non-public information, to proceed past a motion to dismiss have analyzed the conduct as per se illegal price-fixing.[4] In a May 2026 speech, Acting Deputy Assistant Attorney General for Criminal Enforcement Daniel Glad said that “where competitors have understood that their sensitive non-public data will be used to set prices for competitors and have participated on that understanding—the door to the per se rule, and therefore to criminal enforcement, is open.”[5]

However, at least three other federal district courts have analyzed the conduct under the rule of reason, often emphasizing that algorithmic pricing is a new way of doing business.[6]

• Practical Takeaway:
  • Some federal courts have analyzed algorithmic pricing that relies on competitors’ sensitive, non-public information as per se illegal, which eliminates the need to prove anticompetitive effects and raises the prospect of criminal enforcement by the DOJ. Other courts have applied the rule of reason to this conduct or deferred on ruling which standard applies.

Pricing recommendations vs. data aggregation

Cases that have proceeded past a motion to dismiss have often featured allegations that an algorithm generated specific pricing recommendations. By contrast, where the software functions primarily as an information-sharing or data-aggregation tool, rather than a tool to recommend prices, courts have more often found such allegations fail to state a Section 1 claim.

However, information-sharing alone is not categorically beyond antitrust scrutiny. Most recently, in United States v. RealPage, a court allowed a claim to proceed past a motion to dismiss that alleged only an agreement to share confidential data for use in competitors’ pricing.[7]

• Practical Takeaway:
  • Tools that generate pricing recommendations have received more scrutiny in cases to date.

Characteristics of shared data

Courts have allowed algorithmic pricing claims to proceed where they held that plaintiffs plausibly alleged that competitors exchanged nonpublic, competitively sensitive, proprietary data—including confidential pricing, occupancy, or supply information—that was pooled through a common algorithm or, in some cases, in traditional, offline channels.

By contrast, courts have dismissed claims that alleged use of only (1) publicly available information and (2) an algorithm user’s own confidential data. In Gibson, the district court said that “[c]onsulting public sources to determine how to price a hotel room by viewing your competitor’s rates does not violate the Sherman Act.”[8] The Ninth Circuit upheld dismissal of plaintiffs’ claims where plaintiffs “do not allege that [the algorithm provider] pools, shares, or uses the confidential information provided by a given Hotel Defendant into the pricing recommendations it generates for any other Hotel Defendant.”[9] Similarly, in Platkin v. RealPage, Inc., brought by the then state Attorney General of New Jersey, the court dismissed claims against a defendant landlord that had specified in its license that the software would not incorporate confidential data from competitor landlords into its pricing recommendations, nor would its confidential data be used to recommend prices to any competitor.[10]

• Practical Takeaways:
  • Courts have dismissed claims where allegations involved only public data or a user’s own data.
  • Though not necessarily dispositive, courts have cited contractual guardrails in dismissing claims against a defendant where its license prohibited the sharing of its data with competitors or competitor data with the defendant.

Using nonpublic data to “train the model” vs. to generate pricing recommendations

The DOJ and plaintiff states’ complaint in United States v. RealPage draws a line between “model training”—training the models to learn supply and demand parameters—and “runtime operations.”—generating pricing recommendations. In its consent decrees (settlements) with RealPage and settling defendants, the DOJ permits exchanges of certain information relevant to model training, such as property data from non-active leases that are at least 12 months old. But for runtime operations, the consent decrees prohibit algorithms from using any nonpublic competitor data. Landlords must rely solely on their own data to set prices.

While most algorithmic pricing opinions have not focused on this distinction, the district court in Gibson did, holding that providing data to an algorithm for machine learning (i.e., model training) was not an exchange of confidential information between competitors. The court instead compared this to lawyers improving their skills through each client engagement, based on each client’s confidential information, without actually sharing any client’s confidential information with another.

• Practical Takeaway:
  • Courts and regulators may draw a distinction between using competitor data to train algorithmic models and using competitor data to make pricing recommendations.

Evidence of parallel conduct

Courts that have found that plaintiffs plausibly alleged a Section 1 algorithmic pricing claim have done so by recognizing a hub-and-spoke theory of conspiracy. In the absence of direct evidence (which is rare), courts often look at whether there is indirect evidence to prove the “rim” (that is, the agreement among competitors), including parallel conduct and at least one “plus factor” (discussed in the next section). Courts have found the following allegations of parallel conduct to be sufficient:

• Contemporaneous use of a common algorithm coupled with uniform pricing behavior. In Cornish-Adebiyi, the Third Circuit held that the “contemporaneous use of Cendyn’s software and the synchrony of the casino-hotels’ alleged upward pricing and downward room occupancy” was sufficient to plead conscious parallel conduct.
• Delegation of pricing decisions to the algorithm. In Duffy Yardi Systems, Inc., the court treated defendants’ delegation of pricing decisions to Yardi as key allegations of parallel behavior. [11]
• High adoption rates of algorithm’s recommendations. In In re MultiPlan Heath Insurance Provider Litig., the court noted that payors “adopt MultiPlan-calculated rates with little to no changes”[12]

Some courts have dismissed allegations of parallel conduct where the following details were present, but the Third Circuit (the only circuit court to address parallel conduct) held the opposite:

• Temporal gaps in adoption. In Gibson, the district court noted that approximately 10 years separated the earliest and latest adopters of the algorithm, which the court held “tends to undermine Plaintiffs’ argument that Hotel Defendants’ decisions to use these products evidence an agreement instead of independent conduct.” However, in Cornish-Adebiyi, the Third Circuit explained that the relevant parallel conduct was not the adoption of the software but rather the casino-hotels’ continuous deployment of the pricing agent.
• Retention of independent pricing authority. In Gibson, the district court emphasized that defendants could accept, reject, or override Cendyn’s recommendations. However, in Cornish-Adebiyi, the Third Circuit held that “[p]rices are fixed when they are agreed upon,” regardless of whether conspirators always adhere to them.[13] Notably, however, consent decrees and private settlements include measures to limit users’ ability to automatically accept recommendations without additional user input, showing that courts and regulators may evaluate measures to preserve independent pricing authority in crafting a remedial package even if the measures may not (alone) defeat claims at the motion to dismiss stage.
• Practical Takeaways:
  • Courts may credit allegations of uniform pricing behavior and high adoption rates in denying a motion to dismiss.
  • Courts and regulators may evaluate the presence of mechanisms designed to ensure independent pricing, both at the motion to dismiss stage and in crafting remedies.

Is conduct consistent with coordination or independent decision-making?

A “plus factor” is an economic action or outcome that is inconsistent with unilateral conduct but consistent with coordination.[14] Courts denying motions to dismiss have focused on the following alleged plus factors:

• Actions against economic self-interest. High prices despite declining occupancy; forgoing casino revenue from filled rooms (Cornish-Adebiyi), below-market rates that risked subscriber loss; and sharing competitively sensitive information despite an incentive to keep the information private (MultiPlan).
• Motive to conspire. Financial pressures, such as high debt and depressed cash flow following the 2008 recession (Cornish-Adebiyi). In Dai v. SAS Institute Inc., by contrast, plaintiffs’ allegations of pandemic-related pressure were insufficient because plaintiffs did not explain why this signaled collusion instead of interdependence.[15]

By contrast, courts have found allegations of plus factors lacking where independent self-interest is an alternative explanation for defendants’ common behavior or where plaintiffs did not plausibly allege that plus factors were against self-interest.

• Practical Takeaway:
  • Courts evaluate alleged plus factors to see whether they are more consistent with coordination or unilateral conduct.

Does the conduct suggest an invitation to, and acceptance of, collective action?

Another type of indirect evidence that has been used to allege algorithmic price-fixing is an invitation and acceptance of collective action.

In Duffy, the court held that plaintiffs plausibly alleged that the software provider marketed its software as a way to raise rents above competitive levels, and existing clients publicly touted those results, implicitly inviting other landlords to sign up. New clients allegedly “accepted the advertised invitation to trade their commercially sensitive information for the ability to charge increased rental rates without fear of being undercut.” The court held that the system could work as advertised only if each landlord divulged its confidential data, delegated pricing authority to the algorithm, and adopted those prices with minimal deviation.

• Practical Takeaway:
  • At least one court has permitted claims to proceed on the view that touting the software provider’s supra-competitive pricing benefits could constitute an invitation by the software provider and acceptance by the users.

Algorithmic Pricing Cases in Federal Court with Decisions or Notable Settlements[16]

Algorithm Provider Defendant	Algorithm User Defendants	Description of Algorithm (as alleged by Plaintiffs)	Notable Federal Court Cases
RealPage	Landlords (owners and operators of multifamily and student housing properties)	Revenue management software that ingests subscribing landlords’ pricing and supply data, aggregates it with data from other subscribers in the same market and generates daily rent price recommendations for individual rental units.	In re RealPage, Inc., Rental Software Antitrust Litig. (No. II). Dist. ct. denied motion to dismiss. 709 F. Supp. 3d 478 (M.D. Tenn. 2023). There are preliminary or final class settlements with ~40 landlord defendants.United States v. RealPage, Inc. The DOJ settled with almost all defendants throughout 2025-26; plaintiff states joined some settlements. Dist. ct. denied motion to dismiss. 2026 WL 2943655 (M.D.N.C. Sept. 30, 2026).Kentucky ex rel. Coleman v. RealPage, Inc. Dist. ct. denied motion to dismiss. No. 2:25-cv-00093-DLB-CJS (E.D. Ky. Feb. 2, 2026). Two landlord defendants have settled.Platkin v. RealPage, Inc. (D.N.J.). Dist. ct. largely denied motion to dismiss NJ AG’s claim. No. 25-3057, 2026 WL 2056350 (D.N.J. Mar. 31, 2026).
Yardi Systems	Landlords (owners and operators of multifamily residential units)	Revenue management software that collects subscribing landlords’ pricing, inventory, and market data, incorporates data from other subscribers, and generates recommended rental rates	Duffy v. Yardi Systems, Inc. (W.D. Wash.). Dist. ct. denied motion to dismiss. 758 F. Supp. 3d 1283 (W.D. Wash. 2024). One landlord defendant has reached a preliminary class settlement.
Cendyn (formerly Rainmaker)	Hotel operators (casino-hotels in Atlantic City and on the Las Vegas Strip)	Dynamic pricing software integrated into hotel operators’ property management systems; ingests each hotel’s pricing and occupancy data, incorporates market and competitor data, and generates multiple room rate recommendations per day.	Gibson v. Cendyn Group, LLC (D. Nev.). Dist. ct. granted motion to dismiss. No. 2:23-cv-00140-MMD-DJA, 2024 WL 2060260 (D. Nev. May 8, 2024). 9th Cir. affirmed dismissal. 148 F.4th 1069 (9th Cir. 2025).Cornish-Adebiyi v. Caesars Ent., Inc. (D.N.J.). Dist. ct. granted motion to dismiss. No. 1:23-CV-02536-KMW-EAP, 2024 WL 4356188 (D.N.J. Sept. 30, 2024. 3d Cir. reversed and remanded. 184 F.4th 172 (3d Cir. 2026).
Claritev (formerly MultiPlan)	Health insurers / third-party payors (Aetna, Cigna, UnitedHealth, BCBS members, and 700+ others)	Rate-calculation algorithm that determines out-of-network healthcare payment by referencing provider cost data and median payments for comparable services.	In re MultiPlan Health Ins. Provider Litig. (N.D. Ill.). Dist. ct. denied motion to dismiss. 789 F. Supp. 3d 614 (N.D. Ill. 2025).
IDeaS (a SAS company)	Hotel operators, including Wyndham, Hilton, Four Seasons, Omni and Hyatt	Revenue-management and profit-optimization software that considers pricing, occupancy, demand, booking trends and competitor rates to forecast demand and generate room-rate recommendations.	Dai v. SAS Institute Inc. (N.D. Cal.). Dist. ct. granted motion to dismiss. No. 24-cv-02537-JSW, 2025 WL 2078835 (N.D. Cal. July 18, 2025).

* * *

[1] This client memorandum focuses on federal cases and DOJ enforcement actions—but states have also been active in this area. In addition to cases brought in state court, states have begun to legislate to restrict the use of pricing algorithms. For example, as explained in our client memorandum here, California has outlawed the use or distribution of a “common pricing algorithm as part of a contract, combination in the form of a trust, or conspiracy to restrain trade or commerce” and prohibits an entity from using or distributing a common pricing algorithm if the entity “coerces another person to set or adopt a recommended price or commercial term recommended by the common pricing algorithm for the same or similar products or services”. The California law makes no distinction between public and nonpublic data. Cal. Bus. & Prof. Code § 16729. And as explained in our client memorandum here (passage of law) and here (law enjoined), New York passed a law that says it is a violation of the Donnelly Act for a landlord to knowingly or recklessly set rents or other lease terms “based on recommendations from a software, data analytics service, or algorithmic device performing a coordinating function,” but that law was recently enjoined by the Southern District of New York. N.Y. Gen. Bus. Law § 340-b; RealPage, Inc. v. James, 25-CV-09847 (VEC), 2026 WL 2922956 (S.D.N.Y. Sept. 29, 2026).

Additionally, while personalized or surveillance pricing is outside the scope of this memorandum, we have previously written on it here and here.

[2] 184 F.4th 172 (3d Cir. 2026).

[3] 148 F.4th 1069 (9th Cir. 2025).

[4] Duffy v. Yardi Sys., Inc., 758 F. Supp. 3d 1283, 1297 (W.D. Wash. 2024); In re MultiPlan Health Ins. Provider Litig., 789 F. Supp. 3d 614, 644 (N.D. Ill. 2025).

[5] Daniel W. Glad, Acting Deputy Assistant Attorney General for Criminal Enforcement, “Old Crime, New Code,” Remarks at the Antitrust West Coast Conference (May 14, 2026), https://www.justice.gov/opa/speech/acting-deputy-assistant-attorney-general-criminal-enforcement-daniel-gladd-delivers. In deciding whether to charge criminally, the DOJ weighs several factors, including the nature and seriousness of the conduct, the person’s culpability and intent, the harm caused, deterrence, and whether a civil or administrative remedy would be adequate.

[6] Kentucky ex rel. Coleman v. RealPage, Inc., No. 2:25-cv-00093-DLB-CJS (E.D. Ky. Feb. 2, 2026); In re RealPage, Inc., Rental Software Antitrust Litig. (No. II), 709 F. Supp. 3d 478 (M.D. Tenn. 2023); United States v. RealPage, Inc., 2026 WL 2943655 (M.D.N.C. Sept. 30, 2026) (applying the rule of reason because the complaint states a claim under the full rule of reason, but clarifying that this did not establish the law of the case and that plaintiffs remain free to argue for a different mode of analysis).

[7] 2026 WL 2943655 (M.D.N.C. Sept. 30, 2026). There were two Sherman Act Section 1 claims: the second was that agreements to use the software aligned competing landlords’ pricing. Both claims survived.

[8] 2:23-cv-00140-MMD-DJA, 2023 WL 7025996 (D. Nev. Oct. 24, 2023).

[9] 148 F.4th 1069 (9th Cir. 2025).

[10] No. 25-3057, 2026 WL 2056350 (D.N.J. Mar. 31, 2026). However, the same court later granted leave to file an amended complaint because New Jersey alleged that AvalonBay got around its contractual restrictions by, for example, meeting with other landlords.

[11] 758 F. Supp. 3d 1283 (W.D. Wash. 2024).

[12] 789 F. Supp. 3d 614 (N.D. Ill. 2025).

[13] The discussion of these two factors centers on the district court’s decision in Gibson because plaintiffs abandoned their hub-and-spoke claim prior to the Ninth Circuit’s ruling.

[14] In re Musical Instruments & Equipment Antitrust Litig., 798 F.3d 1186, 1194 (9th Cir. 2015)

[15] No. 24-cv-02537-JSW, 2025 WL 2078835 (N.D. Cal. July 18, 2025).

[16] The chart includes only federal court cases that have decisions and/or notable settlements. As noted above, algorithmic pricing case law will continue to develop. Thus far, in cases with decisions, the alleged “hub” has been a third-party software vendor, and the algorithm user defendants have typically been residential landlords or hotel operators. However, a notable recent complaint alleged that McDonald’s, as franchisor, served as the hub of a hub-and-spoke conspiracy by pooling nonpublic, store-level transaction data from its competing franchisees and company-operated restaurants into a pricing engine that it owned and controlled, and that engine conditioned price increases on whether rival franchisees had already raised prices. Thomas v. McDonald's USA, LLC, No. 1:26-cv-12149, ECF No. 1 (N.D. Ill. Oct. 2, 2026) (class action complaint).