Skip to content
Citations Citations
Blog About Request Access

AI Licensing: Calculate the Break-Even Threshold Before You Negotiate Price

Francois-Xavier Bioul
Francois-Xavier Bioul · CCO at Citations LLC
5 min read

The headline price of an AI license does not show its margin. This article sets out the five variables to model before agreeing to a figure, and the threshold below which a publisher should decline.

In 2026, Wiley reported USD 49.1 million in AI licensing revenue, including USD 19.4 million tied to content licensed from other publishers. In its Q1 results, the group had also noted that higher royalty payouts on the partner share had weighed on margin.

Neither figure shows whether a single contract is profitable. Gross revenue does not answer that question, any more than the buyer's own number should set the market benchmark. Between the headline price and the actual contribution sit content preparation, integration, recurring operations and royalty payouts. The strictly illustrative example below shows how to account for them before negotiating.

Two EUR 100k/year licenses with opposite economics

Consider two AI content licenses at the same price: EUR 100k per year over three years, or EUR 300k in cumulative gross revenue. Their cost structures differ entirely.

License A: structured corpus, clean rights

  • Corpus preparation: EUR 15k

  • Integration: EUR 10k

  • Maintenance: EUR 5k/year

  • Legal and reporting: EUR 5k/year

  • Royalty payout: 30% of gross price

Year one leaves a contribution of EUR 35k, and years two and three each leave EUR 60k. The cumulative three-year contribution is EUR 155k, or 52% of gross revenue.

License B: corpus to reprocess, rights chain to secure

  • Corpus preparation: EUR 60k

  • Integration: EUR 30k

  • Maintenance: EUR 15k/year

  • Legal and reporting: EUR 10k/year

  • Royalty payout: 50% of gross price

Year one produces a loss of EUR 65k. The annual contribution turns positive from year two, at EUR 25k, but the three-year cumulative result remains negative at -EUR 15k, or -5% of gross revenue. It turns positive only in year four if the deal renews on the same terms, and renewal itself is not guaranteed by a one-time signature.

With the same price and duration, the two licenses differ by EUR 170k, driven by EUR 110k in additional operating costs and EUR 60k in additional royalty payouts. These figures are illustrative, and each publisher should replace them with its own deal-specific data.

The threshold is the number that matters

Before considering what the buyer offers, a publisher should establish the minimum price that covers its costs and target contribution. The AI licensing break-even price is the minimum annual price a publisher can accept without losing money on preparation, delivery and royalty costs. Five variables build the model:

VariableDefinition
FUpfront preparation and integration costs
RAnnual recurring costs
NFirm contract duration (years)
rRoyalty payout rate
mTarget contribution margin

The minimum annual price is:

Minimum price = (F + N × R) ÷ [N × (1 − r − m)]

For simple break-even, the target contribution m is set to zero. The break-even annual price is then about EUR 26k for License A and EUR 110k for License B. With an illustrative internal target of 30% contribution, the thresholds rise to EUR 46k and EUR 275k respectively.

At a 50% royalty rate, License B can never reach 52% contribution, even before operating costs, because the deal structure makes that target mathematically impossible. The royalty rate and its base therefore determine how much economic room remains, and should be modeled before the price is set rather than applied afterward.

Test more than one demand scenario

A single threshold can create false certainty. At least three scenarios should be tested:

  • low demand, where the deal ends after one year or does not renew, so upfront costs are absorbed over a much shorter period;

  • a central scenario, where the deal delivers the expected revenue for three years with no material change in scope or cost;

  • high demand, where volumes increase and revenue may grow, along with delivery, support, reporting and some royalty payouts.

In a fixed-price deal, demand mainly affects costs and the likelihood of renewal. In a deal indexed to volume or usage, it also changes expected revenue, which is why the usage number behind that revenue needs its own audit trail. The threshold should be tested against each scenario, beyond the central case.

Three possible decisions

Comparing the offered price with the calculated threshold leads to one of three decisions.

The publisher can accept when the price clears the threshold for the target contribution, rights are clean, preparation costs are controlled and the royalty base is known. In the example, License A clears its 30% contribution threshold comfortably.

It can reduce the scope when the price clears break-even but not the target contribution. This can justify narrowing the number of titles, territories, formats, delivery frequency or service level, moving the deal closer to License A’s profile. The reduction still has to be negotiated with the buyer, since a narrower scope can also lower its willingness to pay.

It can refuse or restructure when the price remains below break-even or a major cost cannot yet be estimated reliably. Refusal is also rational when the royalty base, exclusivity terms, litigation risk or delivery obligations make the central scenario insufficiently credible. Such a deal might become viable if scope, costs or terms change, but it is not viable as currently structured.

What to document before the price

The five variables cannot be filled in without checking:

  • the actual state of the corpus;

  • the rights actually held for the intended uses;

  • non-recurring preparation and rights-clearance costs;

  • the annual cost of maintenance, delivery, support and reporting;

  • the base, rate and beneficiaries of royalty payouts.

The model should then be extended to what it does not yet cover: exclusivity, cannibalization of other channels, cost of capital, overhead, taxation and litigation risk.

A high headline price does not make an AI license profitable. Profitability depends on the price staying above a threshold calculated on the deal’s real scope, and that threshold should be established before the negotiation starts.

Sources

  • Wiley, Form 10-K FY2026: USD 49.1M in AI licensing revenue, including USD 19.4M from content licensed from other publishers: SEC filing

  • Wiley, Q1 FY2026 results: higher partner-share royalties weighing on margin: SEC filing

  • Wiley AI Gateway: enrichment, vectorization, metadata, API/MCP delivery, revenue share: Wiley

  • Copyright Licensing Agency: publisher and author remuneration principles: CLA

Scenarios A and B, their royalty rates, and the 30% contribution target are strictly illustrative and are not sector benchmarks.

Before you price the next deal

For how the buyer-side number gets treated as a market reference before it is verified, see the AI content licensing benchmark.

For what changes at renewal once the first contract expires, see one-time vs recurring AI licensing revenue.

For the records that will justify the price once the deal is running, see AI usage evidence for publishers.

Talk to us about pricing your next AI license