Artificial Intelligence and Pricing Power: Optimization Or Coordination

As someone who works professionally in the artificial intelligence space, I spend less time debating whether AI should be used and more time observing how it actually behaves when deployed at scale. AI does not invent new business incentives. It accelerates old ones. Few incentives are older—or more powerful—than the desire to sell any product at the highest price a consumer is willing to pay.

That tension sits at the center of the current U.S. policy moment around artificial intelligence.

On December 11, 2025, the White House issued “Ensuring a National Policy Framework for Artificial Intelligence,” an executive order designed to reduce regulatory friction for AI companies and establish a unified federal approach rather than a fragmented, state-by-state regime. The order is explicit in its objective: preserve U.S. leadership in AI by minimizing regulatory burdens that could slow innovation, distort outputs, or force developers to alter truthful model behavior.

At the same time, ongoing litigation, most notably the Justice Department’s antitrust action involving RealPage, highlights a parallel reality: AI systems can materially change how markets behave, especially when they sit at the intersection of pricing, shared data, and competitive decision-making.

These two developments are not in conflict. They are two sides of the same economic coin.

AI does not decide what is fair. It optimizes toward the goals it is given.

In marketing, pricing, and revenue strategy, the “holy grail” has always been precision: identifying the maximum price a market will bear, segment by segment, moment by moment. Long before machine learning, companies used surveys, A/B testing, yield management, and behavioral psychology to approximate this outcome. Airlines, hotels, insurance firms, and financial institutions have done this for decades.

AI simply removes friction.

With enough data, an algorithm can identify elasticity thresholds faster than any human team. It can detect patterns across millions of transactions. It can adjust recommendations in near real time. None of this is inherently unlawful. In competitive markets, it is often considered best practice.

The legal risk emerges not from optimization, but from coordination.

The Department of Justice and multiple states allege that RealPage’s AI-driven pricing software crossed that line by enabling competing landlords to coordinate rent prices through shared, non-public data. According to the complaints, the software acted as a central hub, ingesting sensitive information from multiple market participants and generating pricing recommendations that landlords broadly followed—effectively replacing independent pricing decisions with algorithmic alignment.

The DOJ’s position is clear:

Using AI does not exempt companies from antitrust law.

From an AI practitioner’s perspective, this is a critical distinction. Algorithms do not collude on their own. But when competitors feed proprietary data into a shared system and defer decision-making to its outputs, the market outcome can resemble classic price fixing—even if no human executives ever meet in a room.

RealPage disputes this characterization, stating that its software is designed to comply with antitrust laws and that landlords retain discretion over final pricing decisions. The company settled with the DOJ in late 2025 by agreeing to stop certain forms of data sharing, while litigation against individual landlords continues. Meanwhile, states such as California and New York have enacted laws restricting or banning rent-setting algorithms altogether.

The result is a live experiment in how AI-driven pricing intersects with competition law.

The December 2025 executive order does not shield AI companies from antitrust enforcement. It does something more specific—and more structural.

It asserts that:

• Truthful AI outputs should not be altered to satisfy ideological or inconsistent state mandates.

• Interstate AI deployment should not be governed by 50 conflicting regulatory regimes.

• National standards are preferable to fragmented enforcement that creates uncertainty for developers and businesses.

From a systems perspective, this matters because AI models are not easily localized. Pricing algorithms, recommendation engines, and decision systems operate across state lines by design. When states impose conflicting requirements—especially those that compel output manipulation or data disclosures—it increases compliance risk and distorts model behavior.

The executive order attempts to draw a clean boundary: innovation should be protected, while unlawful conduct remains unlawful.

What the RealPage case has surfaced is that AI is extremely good at achieving what businesses have always wanted.

• Reduce uncertainty

• Increase pricing confidence

• Minimize underpricing

• Normalize decision-making across portfolios

In isolation, each of these goals is common. In aggregate, when powered by shared data across competitors, they can change market dynamics in ways regulators are not willing to ignore.

This is not an AI problem. It is a governance and deployment problem.

As AI professionals, we understand that models reflect incentives. If the incentive structure rewards alignment over competition, the output will follow. If it rewards independent decision-making with guardrails around data sharing, the outcome looks very different.

The emerging framework is relatively clear, even if enforcement will remain case-specific:

• AI-driven pricing is not illegal

• Shared, non-public competitive data can be

• Delegating human judgment entirely to a shared algorithm increases risk

• National standards reduce uncertainty, but do not erase antitrust obligations

The executive order seeks to preserve U.S. AI leadership by reducing regulatory drag. The RealPage litigation serves as a reminder that speed and scale do not override foundational market rules.

AI can help businesses find the highest price a consumer will pay. That has always been the goal. The question regulators are now answering—methodically, not emotionally—is whether that price is discovered through competition or coordination.