The price is zero. The extraction is real.

When economists teach monopoly, they usually start with a higher price. That's still correct, but it's no longer enough.

In many AI and platform markets, the user-facing price is zero or heavily subsidised. Yet the extraction is real. The firm still captures value — by turning user behaviour into data, data into prediction, and prediction into durable market power.

That matters directly to who controls AI and who captures its value. If the core bargain is no longer money for a service but behavioural exhaust for a model, then consumer surplus in monopoly has to be read through ownership of data, control of distribution, and the contractual terms that make exit costly.

The Textbook Model of Lost Surplus

The standard monopoly diagram still matters because it isolates the basic mechanism of extraction. Consumer surplus is the gap between what buyers would have paid and what they pay. Under competition, price is pushed toward marginal cost, output expands, and more mutually beneficial trades occur.

A monopolist changes that result by restricting output. In the textbook single-price case, the firm produces where marginal revenue equals marginal cost, then sets the price consumers will tolerate for that smaller quantity. The effect is clear. Some value that would have remained with buyers is shifted to the producer, and some value disappears because transactions that would have occurred in a competitive market no longer occur.

What exactly is lost

The diagram separates the welfare effect into distinct pieces:

• Transferred surplus: the higher monopoly price moves part of the buyer's surplus to the firm on the units still sold.
• Deadweight loss: the reduction in output eliminates trades that would have created value for both sides.
• Lower total welfare: the producer gains some of what consumers lose, but the foregone transactions create a net social loss.

This distinction is more than classroom geometry. It shows that monopoly is not only a distributional issue. It is also a production and coordination problem, because the market serves fewer people than it could at a cost-justified price.

That baseline also explains why antitrust law historically focused on price, output, and barriers to entry. Those are the observable signs that a firm has acquired the power to convert market control into rents.

The point carries into technology policy. Debates about AI governance often concentrate on safety, standards, or geopolitical coordination. Those matter, but market structure matters too. Policy analysis on AI cooperation is useful here because governance choices help determine who gets to set the rules of access, accumulation, and control.

The textbook model therefore remains the right starting point, but only as a starting point. It identifies the logic of surplus capture. In industrial markets, that usually appeared as a higher posted price. In AI and platform markets, the same logic often works through weaker forms of consent, contractual lock-in, and the appropriation of user activity that improves the firm's models and strengthens its position.

Where the Textbook Model Breaks Down

The standard monopoly diagram misses the central fact of digital power. In AI and platform markets, firms often capture surplus without raising a posted price at all.

A search engine, chatbot, recommendation layer, or AI assistant may be free at the point of use. That can make the transaction look consumer-friendly, even generous. But a zero cash price does not eliminate extraction. It changes the channel through which extraction occurs.

The hidden price

The textbook model assumes the monopolist gets paid in money. Digital firms often get paid in something else: contractual control, behavioural data, default positioning, reduced interoperability, and the right to observe how users act over time. Those mechanisms matter because they let the firm appropriate value while keeping the visible price low.

That changes how surplus should be measured. If users "pay" with data, attention, feedback, switching costs, and foreclosed alternatives, the old price-and-quantity picture captures only part of the transfer. The monetary charge can stay near zero while the platform still expands its rents through terms of access and ownership of the information users generate.

The political economy point is straightforward. User activity does not just consume the service. It also helps produce the asset that entrenches the firm.

Why AI sharpens the problem

AI systems intensify this dynamic because use improves the product. Prompts, corrections, rankings, workflow patterns, and acceptance or rejection signals all have economic value. The firm can aggregate that value across millions of interactions, convert it into model refinement and behavioural prediction, then use the resulting advantage to strengthen its control over distribution and standards.

That is why the narrow question — "did the consumer pay more?" — often misses the mechanism of capture. A better question is whether the gains created by participation remain with users or are converted into proprietary assets that raise barriers to exit and barriers to entry.

The market therefore appears more efficient than it is. Output may be high and access may be cheap, but the firm is still extracting surplus through architecture rather than price. In practical terms, monopoly power shows up not only in what users buy, but in what they must surrender to participate.

Your Data as the Captured Surplus

The central claim is blunt. In AI and platform markets, consumer surplus is often not merely reduced. It is converted into an asset the firm owns.

Every prompt, click, correction, upload, preference signal, and usage pattern creates informational value. A single interaction has little stand-alone significance. Aggregated across millions of users, those interactions become a production input that improves targeting, ranking, model refinement, interface design, and behavioural prediction.

The user experiences the exchange as consumption. The firm records it as both consumption and capital formation.

From use value to proprietary asset

In the textbook account, consumer surplus is the gap between what a buyer would have paid and what they pay. That framing still matters, but it misses a different transfer that has become economically decisive. In platform markets, the user may keep the immediate utility of the service while the firm captures the longer-term value created by the interaction itself.

That distinction matters because data is cumulative. Once user activity is turned into training signals, workflow knowledge, and prediction capacity, the value does not sit with the individual who generated it. It sits inside the provider's models, recommendation systems, and product architecture. What looked like a cheap transaction can therefore produce a large one-way transfer of economic value.

The result is easy to miss because the cash price remains low.

Why "free" is often a misdescription

A free service is not free in economic terms if participation supplies an input the provider can privatise.

User activity functions as a form of unpaid digital labour. The term should be used carefully. It does not mean users are employees or that every interaction should be treated like formal work. It means their actions produce something with exchange value, and that value is retained, scaled, and monetised by the firm rather than returned to the people who generated it.

Three things happen in the same transaction:

• The user gets immediate use value.
• The firm gets data that improves future performance.
• The market gets more concentrated, because those improvements raise switching costs and entry barriers.

That third effect is the least visible and often the most important. A better model trained on user interaction is not just a better product. It is a stronger moat.

I explored that dynamic in more detail in this analysis of how user data will fuel AI. The underlying issue is larger than privacy. It concerns who owns the value created through participation and who can compound it over time.

When users improve the system through routine use, the platform is not simply meeting demand. It is absorbing the productive value embedded in demand.

Why lock-in converts data capture into monopoly power

Data extraction matters most when users cannot carry the value they helped create to another provider. If histories, prompts, preferences, organisational workflows, and learned system behaviour remain trapped inside one firm's infrastructure, then the surplus generated by use has already been appropriated.

The monopoly story evolves beyond a complaint about terms of service. The firm does not need to raise headline prices to capture more value. It can keep access cheap, accumulate behavioural and training advantages, and then use those advantages to make exit costly for users and entry costly for rivals.

In that setting, consumer surplus is not only what disappears through higher prices and lower output. It is also what becomes proprietary data, model improvements, and switching constraints that strengthen the monopoly itself.

Algorithmic Rents and Price Discrimination

Once a firm has enough behavioural data, it can do more than improve the product. It can price more aggressively.

The textbook monopoly charges one price to everyone. Data-rich firms can do something much closer to segmentation by willingness to pay. They don't need a perfect map of each user's preferences to shift bargaining power decisively. They only need enough signals to sort users into categories, infer urgency, and detect switching constraints.

Data turns information asymmetry into rent extraction

Monopoly in AI markets starts to look less like old industrial pricing and more like continuous rent extraction. The firm learns who is price-sensitive, who is locked in, who needs speed, and who cannot afford interruption.

The underlying structure is not neutral. This Cowles Foundation analysis of monopoly pricing and segmentation shows that the segmentations that maximise consumer surplus also minimise producer surplus — a direct opposition between the user's welfare and the monopolist's interest. In standard settings, consumer surplus and total surplus can align. That breaks when the monopolist uses data to price-discriminate and capture what would otherwise remain with the user.

The economics of privacy makes the limiting case explicit. In Loertscher and Marx's model of digital monopoly, the monopolist's profit always rises as privacy falls — and with no privacy at all, the match between user and product becomes perfect, while the monopolist extracts its entire value.

The rent doesn't need to look like a price hike

This is the part many executives miss. Extraction can happen without an obvious list-price increase.

It can appear as:

• Bundled tiers that make the lower option functionally unusable.
• Personalised offers that reveal different prices to different users or organisations.
• Contractual dependence where exit becomes costly because accumulated usage data cannot travel.
• Feature gating where the firm reserves interoperability, speed, or administrative control for higher-paying segments.

Each mechanism converts information advantage into economic rent. The user experiences this as inconvenience, dependence, or opaque pricing. The firm experiences it as margin.

I've written before about the bargaining implications for content owners in this analysis of pay-per-crawl and AI traffic. The same logic applies more broadly. Once the intermediary controls discovery and learns enough about counterparties, it can intermediate the value and then keep a disproportionate share of it.

The most powerful monopolist is not the one with the highest public price. It is the one that knows enough about you to set the terms invisibly.

Why this matters for consumer surplus in monopoly

In the old graph, some consumer surplus remained because the monopoly used a single posted price. In data-rich markets, that residual cushion shrinks. The firm doesn't just restrict output. It studies demand at a granular level and organises the market so that more of the area under the demand curve ends up on its side of the ledger.

Implications for Antitrust and Regulation

If consumer surplus in monopoly is now captured through data, prediction, and lock-in, then a price-only view of competition policy is too narrow.

That doesn't mean prices no longer matter. It means they no longer tell the whole story. A service can be cheap, subsidised, or free and still impose substantial economic harm if it transfers the gains from user participation to the operator, blocks exit, and strengthens a feedback loop that rivals cannot match.

Competition policy has to move beyond visible prices

The practical implication is that regulators should ask different questions.

• Data consolidation: does a merger combine datasets in a way that deepens behavioural prediction and forecloses competition?
• Interoperability: can users move their history, context, and outputs elsewhere without losing accumulated value?
• Contract terms: do default settings and licences convert customer activity into proprietary model advantage?
• Procurement: is a "free" or discounted AI layer a mechanism for capturing institutional data and deepening vendor dependence?

This isn't abstract. In competitive merchant sectors, AI-driven efficiency gains tend to flow to consumers through lower prices because competition stops firms from keeping the surplus. In monopolistic service sectors with high barriers to entry, firms capture more of that gain as rent instead, as argued in this analysis of AI, competition, and surplus capture. That distinction is one of the clearest policy filters available.

Where rivalry is real, productivity gains diffuse. Where control is concentrated, productivity gains accumulate.

Regulation should target the extraction architecture

I'd focus on four levers.

1. Data portability that works. Formal portability rights mean little if exported data is incomplete, low-context, or unusable in a rival system.
2. Interoperability obligations. In markets with entrenched gatekeepers, interoperability can reduce switching costs and prevent a firm from converting user history into a captive asset.
3. Scrutiny of terms of service. Many of the most important transfers happen through contracts users never negotiate. Regulators should read these as instruments of value capture, not boilerplate.
4. Merger analysis that treats data as a strategic asset. A transaction can reduce competition even if the product remains cheap, because the primary gain is control over training inputs and behavioural prediction.

A useful companion concept here is rent-seeking. This explanation of the economic costs of rent-seeking helps clarify why some firms invest less in creating value than in controlling the channels through which value must pass. In AI, that often means control over data, defaults, procurement pathways, and distribution.

A regulator who looks only for higher prices will miss the core extraction mechanism of many AI monopolies.

What this means for European operators and policymakers

From a European vantage, this matters especially in procurement and public administration. Institutions often accept subsidised software or integrated AI layers because the upfront budget line looks attractive. The deeper cost is strategic: valuable institutional data, usage patterns, and workflow dependencies accumulate inside systems the institution doesn't control.

That is why proactive regulation matters before the market fully settles. I've addressed that broader urgency in this piece on the case for proactive AI regulation. Once a platform has combined scale, data, and default distribution, recovering contestability becomes much harder.

The old monopoly diagram still teaches something important. It teaches that surplus can be transferred or destroyed. The digital update is that surplus can also be absorbed into infrastructure. The user still creates value. The monopolist captures it before it ever appears as consumer welfare.

Sources

• How a profit-maximizing monopoly chooses output and price — OpenStax, Principles of Economics
• Consumer surplus, monopoly profit, and deadweight loss in the standard model — Misra, SSRN
• Digital monopolies: privacy protection or price regulation? — Loertscher & Marx, International Journal of Industrial Organization
• Monopoly pricing, segmentation, and the conflict between consumer and producer surplus — Cowles Foundation, Yale
• AI, competition, and where economic surplus flows — Drift Signal
• The G7 and the future of AI governance — Global Governance Media
• Economic costs of rent-seeking — Unitism

Fabio Lauria

CEO & Founder, ELECTE

Every week, we explore AI without the hype — using data, analysis and an independent perspective.