In this post, I report on the “SEP and FRAND Developments” roundtable at the EPIP Conference in Turin on September 10, 2026. The European Policy for Intellectual Property (EPIP) association organizes an annual conference that brings together legal and economic scholars and practitioners. Rafał Sikorski (Adam Mickiewicz University, Poznań) chaired the roundtable, with Wolrad Prinz zu Waldeck und Pyrmont, Enrico Bonadio, Rudi Bekkers, Francesca Ferrari, and Christian Helmers. Discussions ranged from competing courts and global FRAND rates to patent injunctions, essentiality checks, and the limits of AI.

Courts competing for the dispute

Wolrad Prinz zu Waldeck und Pyrmont, a partner at Freshfields, opened with a practitioner’s account of international SEP litigation. Enforcement campaigns by SEP holders now span the United Kingdom, China, Germany, the Unified Patent Court (UPC), Brazil, India, Colombia, and other jurisdictions. Germany remains the center of gravity: alongside their UPC filings, SEP holders almost always also sue before the Munich Regional Court, which is very service-oriented toward patent owners and typically reaches trial within eight or nine months of service.

Implementers have responded by asking UK courts to determine global FRAND rates: a royalty commitment shows they are willing licensees and shields them from injunctions elsewhere. Because UK rate-setting trials take time, some have also sought interim licenses as placeholders until trial. German courts and the UPC have countered with anti-suit and anti-interim-license injunctions: they view an interim license as an attempt to obstruct SEP enforcement and have ordered implementers to withdraw their UK applications on pain of heavy penalties. In InterDigital v Amazon, the UPC’s Mannheim Local Division threatened up to €50 million plus €500,000 per day and declared UK-imposed interim licenses contrary to EU ordre public.

In his assessment, the UK route has consequently become largely ineffective for implementers unless the SEP holder itself submits to English rate setting. In that case, there is no competing enforcement to protect and nothing to enjoin. Nokia did so in February 2026 against Warner Bros. Discovery and Paramount; he expects that choice mainly from operating companies seeking to establish a licensing program, not from non-practicing entities.

Regarding the Munich FRAND guidelines, published by the 7th Civil Chamber of the Munich Regional Court in August 2026, Prinz zu Waldeck und Pyrmont stressed that they are not an official position of the court but the views of one of its two patent chambers, even though patent owners abroad present them as “the German courts’ FRAND approach.” He welcomed one shift: the chamber now accepts that, in most cases, the court must also examine whether the SEP holder’s offer is FRAND, a step Munich courts had not taken in years of SEP litigation. However, he found the proposed approach highly problematic and unbalanced: the reference values it sets out for assessing offers draw partly on evidence the judges have seen in other cases, so a defendant faces benchmarks it cannot examine or contest. The one merit of publishing the guidelines, he suggested, is that litigants now openly know what they are up against.

Enrico Bonadio, Professor of Intellectual Property Law at City St George’s, University of London, saw a broader jurisdictional conflict and was openly pessimistic about it. Anti-suit injunctions, anti-anti-suit injunctions, interim licenses, security payments, and court-specific guidelines are all tools each court uses to shut out the others and make litigating elsewhere more costly. That is no sensible way to share authority among courts with a legitimate stake in the dispute: it rewards whichever party files first in the forum of its choice and plays roughest.

Global licensing and territorial patents

Bonadio then focused on territoriality. Under Article 4bis of the Paris Convention, in force since 1900, patents granted in different countries are independent, and each state remains free to calibrate its own remedies, limitations, and enforcement regime. No worldwide patent exists. However, the standards and product markets that SEPs cover are global, and Bonadio acknowledged that this mismatch creates a genuine economic problem: a single worldwide license is more efficient than licensing and litigating country by country.

Yet when a court sets a global rate, Bonadio argued, it does more than calculate a convenient number. It exports its own conception of FRAND to distant markets whose regulators, courts, and stakeholders had no say in the determination, and it must, in effect, treat the foreign patents in the portfolio as valid, even though only the courts of the granting country can decide that. Bonadio highlighted a tension in the EU’s WTO challenge (DS632) to Chinese courts setting global FRAND rates: European courts do much the same. UK courts have set global rates since Unwired Planet, and German courts and the UPC, while not setting rates themselves, pressure implementers toward global licenses under threat of an injunction.

Rather than abandoning global rate setting, Bonadio proposed a framework for when, and under what constraints, a court may set a global rate: only where it has a strong territorial connection to the dispute, confined to patents and markets closely linked to the forum, with disclosed valuation benchmarks, submissions from affected third-country implementers, and a reasoned decision explaining how those markets shaped the rate. An international body might offer an alternative to national courts setting global rates: Jorge Contreras has proposed a global FRAND rate-setting tribunal, and Lord Justice Arnold has suggested that standards bodies mandate arbitration through their IPR policies. Bonadio found the idea fascinating but difficult to implement in practice.

Prinz zu Waldeck und Pyrmont noted that the industry already thinks territorially: most license agreements distinguish between major and minor markets and apply different rates to each, so territorially limited rate setting would not be alien to commercial practice. If the UK set a rate only for the UK, and an implementer sued in Germany could defend itself with a FRAND offer for the German portfolio, the dispute over who circumvents whom would disappear. Proceedings would run in parallel, as in pharmaceutical litigation, and parties would settle once several outcomes pointed the same way.

Rudi Bekkers, Professor of Standardization and Intellectual Property at Eindhoven University of Technology, agreed and added the corollary. A global rate must be discounted for every country where the licensor holds no relevant patents, and because SEP portfolios are geographically uneven, the discount may be significant. Those who like the fruits of a global rate, he observed, are not always prepared to accept the burden that comes with it.

Injunctions and proportionality

Francesca Ferrari, Professor of Civil Procedure Law at the University of Insubria, approached the debate through the European Commission’s follow-up study on the Intellectual Property Rights Enforcement Directive (IPRED), published in January 2026. The study does not address SEPs specifically, but it asks whether European patent injunctions should move toward a more explicit and structured proportionality assessment, with the U.S. Supreme Court’s eBay framework as a point of comparison.

For SEPs, Ferrari argued, the question is not simply whether Europe needs “more proportionality,” because SEP injunctions already operate within the conduct-based framework of Huawei v ZTE. Is an additional, open-ended proportionality assessment needed on top of that mechanism? Her answer was probably not, at least on the available evidence: proportionality is already part of the European framework, and SEP litigation shows that it can operate through structured rules rather than broad judicial discretion. That is the real transatlantic difference: not that Europe grants injunctions and the United States does not, but that American judges enjoy broad equitable discretion at the remedial stage, whereas the European model relies on statutory rules. Proportionality is part of effective enforcement, not a counterweight to it.

She also cautioned against overinterpreting the study’s two striking figures: permanent injunctions were granted in 94% of the patent decisions examined, and proportionality was explicitly assessed in only 0.6% of cases. The dataset covers only published decisions, so it excludes settlements, withdrawals, and claims that end before judgment. The absence of the word “proportionality” from a judgment does not prove that no such analysis took place. Ferrari also questioned the premise of interim licenses: conceived as a remedy against hold-up by SEP holders, they may neglect the risk of implementer hold-out.

Bonadio read the same figures differently, as evidence of a gap between the law on paper and in practice. Proportionality appears throughout the texts: in the Enforcement Directive, in TRIPS (whose Article 44 says that courts “may” order injunctions, not that they shall), and in the Commission’s 2017 guidance. In practice, he argued, German courts and the UPC still treat a permanent injunction as the near-automatic consequence of infringement, without considering the defendant’s freedom to conduct a business under Article 16 of the EU Charter.

The U.S. injunction debate returns

Christian Helmers, Professor of Economics at Santa Clara University, discussed Collision Communications v. Samsung. The case does not involve standard-essential patents, but it could affect the broader debate over patent remedies. Collision is a non-practicing entity, but its original business model was not purely monetization: it maintained an R&D operation and sought “design wins,” developing its patented technology jointly with manufacturers so they would build it into their products. It negotiated with Samsung for several years and lost out when Samsung developed the technology in-house. Collision sued in Texas in 2023 and, in October 2025, won a $445.5 million jury award for past infringement.

Collision then sought a permanent injunction, which Judge Rodney Gilstrap denied in May 2026. Crucially, Judge Gilstrap found for Collision on the first two eBay factors, irreparable harm and inadequacy of monetary compensation, and refused to treat non-practicing entities as “a lower second class of patent ownership.” However, he found the balance of hardships and the public interest in Samsung’s favor. Helmers cautioned that the irreparable-harm finding rested on Collision’s earlier joint-development model and may not extend to a plain-vanilla monetizer.

On appeal, Collision challenges the application of eBay through an originalist argument: the “traditional principles of equity” eBay invokes were those applied by the English Court of Chancery when Congress passed the Judiciary Act of 1789, and Chancery ordinarily enjoined continuing infringement. The argument may sound outlandish, Helmers noted, but the Supreme Court used the same historical reasoning in Trump v. CASA (2025), the birthright-citizenship case, to curb universal injunctions. Whether the Federal Circuit will engage remains uncertain.

Helmers linked the case to recent U.S. government interventions favoring injunctions: in February 2026, the Department of Justice and the USPTO filed a statement of interest in the case, arguing that injunctions should generally be available because courts cannot reliably determine the appropriate royalty. On this view, injunctions enable price discovery. That position revives a question most economists had considered settled, and that motivated the entire literature on hold-up: do injunctions help parties discover the technology’s price, or do they give patent owners leverage to capture value created by standardization and switching costs?

Can AI determine essentiality?

Bekkers concluded by stepping out of the courtroom: ideally, parties should negotiate a license before these conflicts arise, and one obstacle is essentiality. Nearly one million patents have been declared potentially essential to standards, yet only a small fraction is believed to be genuinely essential, and both the SEP holder and the implementer need to know which. A public system of essentiality checks was central to the European Commission’s proposed SEP Regulation, which was never adopted. Could AI do the job instead?

Bekkers identified three obstacles. First, essentiality is not about textual similarity. Under ETSI’s IPR policy, a patent is essential only if it is technically impossible to comply with the standard without infringing it; if the standard can be implemented in another way, the patent is not essential, so the assessment must consider the entire universe of alternative implementations. Second, no sufficiently large, authoritative dataset of positive and negative essentiality determinations exists to train and validate such a system. Third, and for Bekkers decisive, both parties to a negotiation must trust the results. AI systems are neither accountable nor transparent, and parties would end up shopping among dozens of commercial services with different results, each picking the one that supports its position. AI could help, Bekkers concluded, but without trusted data, transparent methods, and acceptance by both sides, it is unlikely to resolve the essentiality problem on its own.

A fragmented system

The problems discussed were distinct, and the panel offered no unified answer. What they share is a structure of trade-offs: global rates reduce transaction costs but may obscure territorial differences; injunctions support enforcement but alter bargaining power; and AI may improve essentiality assessment while creating another contested layer of evidence. The challenge is not only to identify the correct FRAND methodology but also to decide which institution may apply it, over which patents and markets, and subject to which safeguards.

I hope you enjoyed this brief. If you did, please consider subscribing to The Patentist SEP Edition.1 It is free, and you can unsubscribe at any time.

1

To subscribe to The Patentist SEP Edition, subscribe to The Patentist, then open “Manage subscription” and turn on the “The Patentist SEP Edition” section.

As discussed in the first brief of The Patentist SEP Edition, determining a FRAND rate is rarely about finding a single, magical number. It is about the method used to arrive at that rate. Because this ecosystem thrives on predictability and fairness, any effort to shed light on how FRAND determinations are made is a step in the right direction.

The recent study published by the EPO Observatory is thus a welcome addition to the field. Co-authored with Brela Consulting, Methodologies for FRAND determination surveys 65 court decisions across seven jurisdictions to map how global case law addresses the thorny issue of FRAND valuations.

Transparency in this space is sorely needed, and the EPO report offers a valuable synthesis of global judicial trends. However, as we dig into the findings, it is crucial to understand what the report actually tells us—and, perhaps more importantly, what it doesn’t.

Complementary lenses: EPO vs. WIPO

The EPO study and WIPO’s recent FRAND economics report serve different purposes. WIPO’s report is normative and analytical, approaching FRAND valuation primarily from an economic theory standpoint. It explains, among other things, why structured valuation matters, why comparables can mislead when they reflect bargaining distortions, and why bottom-up and top-down methods have distinct economic rationales.

The EPO study takes a different approach. It is descriptive and practice-oriented, asking how courts have, in fact, proceeded when forced to translate FRAND into numbers, ranges, or judgments about offers. Read together, the two reports are highly complementary.

Strong in operational details

The EPO study is particularly strong in its operational details. It does not merely tell us that comparable licenses are common; it shows how difficult the comparables exercise is.

Across the surveyed cases, courts were presented with anywhere from two to 54 proposed comparables, yet accepted far fewer as genuinely comparable. The hard work lies in selection, unpacking, and discretionary adjustment. If a license is a lump sum or a cross-license, what is the effective rate once it is unpacked? If the license reflects different bargaining conditions, how should it be normalized?

Furthermore, the report highlights that forum and procedural posture shape methodology. This nuance is critical to avoiding the trap of conflating legal constraints with economic best practices. FRAND assessments in the sample are heavily concentrated in the EU (especially Germany), often in the context of injunction analysis under Huawei v. ZTE and Article 102 TFEU. By contrast, US cases often arise in patent-damages litigation, while UK cases appear as standalone portfolio-rate determinations. Therefore, any apparent judicial “preference” for one method is inseparable from the institutional task those courts are performing.

Judicial prevalence is not economic superiority

The report’s most quoted finding is that comparable licenses have emerged as the main method relied on by courts, while the top-down approach is used less frequently and often as a “cross-check.”

While this statement reflects the surveyed case law, it does not imply that comparables are economically superior. The fact that UK and German courts often treat top-down as a cross-check should not be mistaken for proof that top-down lacks an independent economic rationale. When applied well, the top-down framework imposes vital discipline on portfolio claims, directly addresses royalty stacking, and tests whether a proposed portfolio rate is plausible once scaled to the standard as a whole. As the EPO study itself acknowledges, Chinese and US courts have at times treated top-down as a principal method, noting insufficient evidence that one method is superior to the other. Neither method is self-executing: methodological pluralism is often wiser than methodological triumphalism!

The danger of unscrutinized comparables

This discussion brings us to a subtle but critical implication from the report’s findings. If courts default to the comparables method without rigorous scrutiny, there is a profound risk of perpetuating rates that are fundamentally non-FRAND simply because precedent treats them as FRAND.

A comparable is only as good as the fairness of the underlying agreement. If past licenses were signed under the threat of an injunction, or between parties with vastly asymmetric market power, using them as a baseline embeds historical imbalances into future FRAND determinations. Therefore, the section of the judicial debate that scrutinizes whether these underlying comparable decisions were fair in the first place is paramount.

A shared vulnerability to data

FRAND discourse often treats the top-down approach as the only method vulnerable to data weaknesses (such as patent counting). The EPO report offers a useful corrective. It shows that data limitations do not solely plague the top-down approach; they affect the comparables method just as much. As the study shows, patent counts, essentiality screens, validity, family deduplication, and portfolio-strength metrics are equally necessary when courts must unpack cross-licenses or adjust rates across disparate portfolios. The report candidly acknowledges that patent counting is rough, the gap between declared and truly essential patents can be large, and no universal measure of SEP value has been accepted by courts—evidentiary challenges that ultimately undermine both methods alike.

What’s missing and what’s next

We should read the EPO report as a beginning rather than an ending. Public availability of data remains uneven and exceptionally thin. Arbitration outcomes and valuation methods are generally confidential, leaving the pool of observable SEP licenses small and incomplete. We need more information from private, undisputed contracts if we want comparables analysis to become robust.

In short, the EPO is an excellent, careful survey of judicial practice. But if we want to know which method is economically justified, we must keep the WIPO report close at hand. The real challenge for the next phase of this debate is bringing these two conversations together. The field now needs better public data on licenses, reliable essentiality-checked portfolio metrics, and standard, objective frameworks to value technical contribution.

I hope you enjoyed this brief. If you did, please consider subscribing to The Patentist SEP Edition.1 It is free, and you can unsubscribe at any time.

1

To subscribe to The Patentist SEP Edition, subscribe to The Patentist, then open “Manage subscription” and turn on the “The Patentist SEP Edition” section.

In this first brief of The Patentist SEP Edition, I discuss WIPO’s recent report on FRAND valuation. One reason the report is interesting is that it brings together perspectives from both sides of the SEP ecosystem: those closer to standard developers and those closer to implementers. That is also what makes the exercise valuable. SEP debates are often polarized, and the report seeks to provide a balanced, economics-based view of the issue. I should disclose that I served as a peer reviewer for the report, but the views expressed here are my own.

Standards solve one economic problem and create another. They allow firms to coordinate around common technologies, making products interoperable, markets larger, and complementary innovation easier. But once a technology is embedded in a widely adopted standard, pricing becomes difficult. Implementers may no longer have realistic alternatives, while patent holders are constrained by their commitment to license standard-essential patents, or SEPs, on FRAND terms: fair, reasonable and non-discriminatory.

WIPO’s report, FRAND Economics: Valuation Methods in Licensing Standard Essential Patents, is valuable because it treats this problem for what it is: not only a legal issue, but an economic valuation problem.

The report’s core insight is that FRAND is not a single number waiting to be discovered. It is a disciplined attempt to identify a plausible range of licensing terms that preserves incentives on both sides of the standardization ecosystem.

That balance matters. SEP holders invest in technologies that may become part of standards. Implementers invest in products, services, distribution, and complementary innovation that make those standards valuable in the market. A standard creates value jointly. The economic challenge is to divide that value without allowing either side to exploit the other after the market has coordinated around the standard. I like to think of the relationship as symbiotic. Each side hopes to enjoy a larger slice of the cake, but the cake itself exists only because both sides helped bake it.

The report reviews three main approaches to valuation. The first is comparables: looking at real-world license agreements as market evidence. The main strength of this approach is its grounding in actual transactions. But SEP licenses are not ordinary market prices. They may reflect litigation pressure, cross-licenses, confidentiality, injunction risk, geographic scope, or bargaining power. Comparable licenses are therefore useful, but they need to be unpacked carefully.

The second is the bottom-up approach. This asks what incremental value the patented technology contributed over the next-best alternative before the standard created lock-in. Economically, this is close to the ideal benchmark: reward the technology, not merely the fact that it was selected into the standard. But the method is hard to apply because the relevant counterfactual is often difficult to observe.

The third is the top-down framework. This starts from a reasonable aggregate royalty for the standard and allocates a share to a particular portfolio. Its strength is discipline: it guards against royalty stacking. Its weakness is measurement. If allocation relies too heavily on declared SEP counts, it inherits all the familiar problems of patent statistics: not all declared SEPs are essential, valid, or equally valuable. It is also sensitive to aggregate benchmarks and allocation rules.

The report’s main contribution is not to settle the FRAND debate. It does something more modest and more useful: it clarifies the problem’s economic structure. FRAND valuation is about bargaining under lock-in, complementary innovation, imperfect information, and noisy evidence.

That is why this report is a good starting point for The Patentist SEP Edition. SEP disputes are often presented as legal battles or industry lobbying. They are also, fundamentally, applied economics problems.

I hope you enjoyed this first brief. If you did, please consider subscribing to The Patentist SEP Edition.1 It is free, and you can unsubscribe at any time.

1

To subscribe to The Patentist SEP Edition, subscribe to The Patentist, then open “Manage subscription” and turn on the “The Patentist SEP Edition” section.