In new research, Tingting Song examines how FRAND principles typically used to discipline excessive or discriminatory terms in SEP licensing can be applied to data brokers in data licensing.
The legal process of acquiring, maintaining, and renewing the rights to use another entity’s property, also known as the licensing cycle, is essential for promoting innovation. In technology markets, devices often need to work together to ensure consumers can seamlessly use products from multiple companies. In order to achieve this uniformity, standard-setting organizations (SSOs) create common technical standards, such as 4G and 5G cellular standards. A tech patent becomes standard-essential when implementing the standard requires the use of the patented technology, which is shared by the patent owners via standard essential patent (SEP) licensing. However, once a technology becomes standard, there is no way to realistically avoid using the SEP. The SEP holders may then gain bargaining power from this lock-in, charging excessively high fees or blocking competing companies’ access. The principles of FRAND—fair, reasonable, and non-discriminatory—were developed to constrain such bargaining power and are the core tenants of keeping SEP licensing fair and competitive.
In practice, FRAND principles are typically applied through commitments made by SEP holders to SSOs. An SEP holder commits to license patents essential to the standard on fair, reasonable, and non-discriminatory terms. The SEP holder and implementer then negotiate the specific license against the background of the FRAND commitment. If they disagree over whether the proposed royalty or other terms are FRAND, either side may ask a court or arbitral tribunal to resolve the dispute.
In recent research, I propose that FRAND, which has mainly been used in SEP licensing, can play an important role in disciplining excessive or discriminatory licensing terms imposed by data brokers that trade data as a product. A FRAND framework adapted to data licensing—what I refer to here as Data-FRAND—can draw substantially on FRAND principles developed in SEP licensing.
Why do data licensing and data brokers matter?
Data licensing refers to a data controller granting a downstream user, such as a business, agency, or individual, permission to use data. It can play an important role in unlocking the value of data, particularly consumer data. In the digital economy, large volumes of consumer data, including identifying information, transaction histories, behavioral data, preferences, and inferred information, are collected and controlled by digital platforms and other data holders. Where such data remain within the control of a limited number of firms, its potential downstream uses may remain underdeveloped. Licensing can enable other firms to use the data to improve existing products or services, develop new ones, and generate additional value from the same underlying resource.
This is not merely a theoretical possibility. Commercial data access and licensing are already emerging in a range of settings. In finance, for example, data aggregators may provide financial institutions and fintech firms with paid access to data-related infrastructure and information services. Data intermediaries may likewise license data to downstream users. However, we’ve also seen a range of data licensing abuse, such as a 2024 case from China. The Shanghai Administration for Market Regulation found that financial technology company Ningbo Sumscope had unlawfully refused to provide relevant data to downstream financial service providers and ordered it to cease that conduct. These examples illustrate a broader development: data are increasingly being made available beyond the firm that originally collects or controls them, allowing data to function as an input into downstream products and services.
Data brokers are particularly important in this process. They aggregate consumer and firm data from various sources, facilitate their management, and license them to data licensees. A data broker may therefore connect two groups: data providers and data licensees. Consumers and firms provide data, and may receive services, insights, or monetary rewards in return. Meanwhile, data licensees, including digital platforms, advertisers, data analytics firms, firms in particular industries, and specialized data brokers, pay for access to data or data products. The main focus here is on third-party data brokers whose business model centers on licensing data to downstream users. In the model considered here, data brokers may also operate under trust-based duties toward data providers, requiring them to responsibly handle the data. Data brokers are attracting increasing regulatory and policy attention in jurisdictions including the United States, European Union, and United Kingdom. For data brokers, data licensing is not merely incidental to another business activity. Instead, it is central to how they commercialize data. This makes the terms on which they license data particularly important.
Those terms can themselves become a source of dispute. Consider a simple example. A data broker licenses data to a third-party data user, a lending-service provider. The lending-service provider needs financial transaction data, credit data, and asset data. It uses such data to assess risk and provide loans. The provider pays the broker a licensing fee. The provider, however, argues that the broker charges excessive fees, discriminates between similarly situated licensees, and fails to negotiate in good faith.
Why FRAND?
The lending-service provider could pursue three different routes to force the data broker to agree to fair terms. First, the provider could bring an excessive pricing claim under competition law if the broker is dominant in the market. However, excessive pricing is hard to prove. Such claims are also highly case-specific and case-by-case enforcement doesn’t necessarily provide a general framework for recurring licensing disputes involving other data users.
Second, the provider could rely on the data access approach. This approach, reflected in instruments such as the EU Data Act and Digital Markets Act, focuses primarily on whether and under what conditions a user can obtain access to data. Data access rules are useful, but they do not resolve the problem here. The broker is already in the business of licensing data and has approved for the provider to access this data. That the provider can access the data does not resolve the terms of licensing, such as price, quality, scope, timing, and discrimination.
The third route would be to proactively apply FRAND principles that are most commonly used in SEP licensing. FRAND asks precisely whether licensing terms are fair, reasonable, and non-discriminatory. These principles aim to restrict unfair practices in cases where one party has immense bargaining power, giving the data broker a set of principles to follow when negotiating terms for licensing to downstream users. Therefore, unlike an excessive pricing claim, FRAND offers a more structured framework for recurring licensing disputes, although its application may still require case-specific analysis.
Can FRAND be transposed?
Can this logic be transposed to data licensing? The answer is yes, but only in a structural sense. SEP licensing and certain forms of data licensing can exhibit a similar structural problem: downstream users may become dependent on an input after building their products or systems around it, creating a need to constrain opportunism without undermining investment incentives. Both SEP licensing and data licensing can involve valuable, non-rival inputs. For a particular downstream use, access to a specific dataset or data environment may also become practically essential. Once downstream users build products or systems around that data, switching to an alternative may become costly.
This dependence can increase the broker’s bargaining power, and the resulting risk is not limited to price. Unfavorable renewal terms, quality degradation, or restrictions on scope may also matter. This is where FRAND becomes useful: the objective is not simply to lower data licensing fees, but to constrain exploitation of downstream dependence while preserving incentives for data aggregation, governance, and intermediation.
However, differences remain between SEP licensing and data licensing. Key differences include the more dynamic and context-specific nature of data value and the different sources of lock-in. The dynamic and context-specific nature of data value affects both reasonableness and non-discrimination. The same dataset may have very different value for different downstream products or services. Valuation and comparison must therefore take into account factors such as data type, purpose, quality, and consent limits. The sources of lock-in also differ. In data licensing, lock-in may arise from integration into downstream systems and switching costs, rather than from formal standardization alone. These differences mean that SEP-FRAND can be transposed, but not without modification. Data-FRAND requires data-specific recalibration.
How should FRAND be transposed?
FRAND can be divided into substantive and procedural dimensions. The substantive dimension concerns the data licensing terms themselves, including fairness, reasonableness, and non-discrimination. The procedural dimension concerns how the terms are formed. The main procedural tool is good-faith negotiation. When adapting these tools, the distinctive characteristics of data licensing—including the treatment of data as a traded product and the trust-based responsibilities that data brokers may assume toward data providers—must be taken into account.
Reasonableness asks how the distinctive characteristics of data licensing should affect the assessment of licensing terms. Non-discrimination asks which licensees should be compared and which benchmarks, such as price, costs, or downstream profits, should be used to assess differences in treatment. Good-faith negotiation requires a transparent and workable data licensing process for reaching and revising licensing terms.
To conclude, FRAND is useful because data-licensing disputes often concern licensing terms, not access alone. SEP-FRAND provides a useful structural reference because certain data-licensing relationships may generate similar forms of dependence and bargaining power. Any transposition, however, must be recalibrated to reflect the distinctive characteristics of data licensing.
Author’s Disclosures: The author reports no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
Subscribe here for ProMarket’s weekly newsletter, Special Interest, to stay up to date on ProMarket’s coverage of the political economy and other content from the Stigler Center.
