Is the Top-Down FRAND Approach Officially Trending in Europe?

Is the Top-Down FRAND Approach Officially Trending in Europe?


The High Court of Justice in London recently issued its third judgment on a fair, reasonable and non-discriminatory (FRAND) rate for a license to cellular standard-essential patents (SEPs). In the two prior UK FRAND decisions—Unwired Planet Int’l Ltd. v. Huawei Techs. (UK) Co. Ltd., 2017 EWHC 711, and Interdigital Tech. Corp. v. Lenovo Group Ltd., 2023 EWHC 539—the Court determined a FRAND benchmark rate after analyzing comparable licenses and considered alternative calculations based on the top-down approach only as a potential cross-check on the benchmark rate. In its latest decision—Optis, 2023 EWHC 1095 (Ch)—the Court applied a somewhat unique top-down approach to the exclusion of the comparable licenses analysis.

In Depth

Optis owns a portfolio of SEPs that cover various cellular standards and asserted some of the patents in litigation. The Court was presented with several alternative approaches to calculating the FRAND rate for Optis’s portfolio, including proposed comparable licenses. Optis relied on licenses it previously granted to other implementers on the exact portfolio (or at least a subset of it) asserted in the litigation. The defendant, by contrast, relied on licenses it previously took to other licensors’ SEP portfolios on the same standards.

Dismissing Optis’s previous licenses to the portfolio at issue as non-comparable, the Court instead sought to answer the “FRAND Question” by first determining the “stack value” (i.e., the aggregate royalty) and then apportioning to Optis its slice of that value. Optis thus departs from prior case law in the United Kingdom, which favored the comparable licenses analysis for determining benchmark FRAND rates and considered the top-down approach only as a cross-check on those benchmark rates.[1]

The top-down approach is a two-step method for calculating the FRAND rate owed to a particular SEP licensor according—at least in its simplest “patent counting” form—to the following formula:

where RL is the royalty owed to SEP licensor L; RAgg is the aggregate royalty for all SEPs on a given standard; NSEP,L is the number of SEPs (or SEP families) owned by SEP licensor L; and NSEP,Tot is the total number of SEPs (or SEP families) on the standard. There are numerous ways to determine the aggregate royalty, and various adjustments can be made to account for factors such as patent quality, the relative contribution of a licensor’s IP to the standard, the geographic scope of patent coverage and patent expiration. However, the defining characteristic of the top-down approach is first determining the aggregate royalty and then apportioning it among the various SEP owners.

Every judicial application of top-down to date has its own quirks regarding how the various inputs to the top-down formula are calculated. Optis is no different.

Compared to other top-down cases, the Court took a simple approach to calculating NSEP,L and NSEP,Tot (the ratio of which the Court referred to as Optis’s “stack share”). It used patent families (rather than individual patents) as the relevant unit of accounting and relied on simple family count data from a third-party source. In particular, the Court preferred the data from Innography (relied on by the defendant) because Innography counted all declared SEPs. By contrast, Optis’s source (PA Consulting) attempted the “judgemental exercise” of filtering declared SEPs for essentiality, which the Court viewed as too “ambitious” and a potential source of error. The Court also rejected the parties’ attempts to adjust the stack share for quality, finding no basis to determine that Optis’s patents were any better or worse than the average cellular SEP.

Two additional details on the Court’s stack share analysis are worth noting. First, the Court seemingly presumed that only patents formally declared to the relevant standards-setting organization should count in the denominator. This would exclude patents that are, as a technical matter, essential to the standard—and thus could be asserted in litigation to derive royalties from or enjoin standard implementers—but whose owners (for one reason or another) do not formally declare them as such.[2] Second, the Court admonished against mixing sources for the numerator and denominator. In particular, the Court criticized Optis’s reliance on data from PA Consulting to count Optis-owned SEP families but factual findings from Unwired Planet to determine the total SEP family count as an inappropriate “apples and oranges” approach.

Perhaps the most interesting aspect of the Court’s top-down analysis is its method for calculating the aggregate royalty. Previously, courts have divined the aggregate royalty from indicators such as the profit margin on the smallest saleable patent practicing unit (SSPPU) (Innovatio, 2013 WL 5593609 (N.D. Ill. 2013)), licensors’ pre-standardization public statements about the appropriate stack value (TCL, 2018 WL 4488286 (C.D. Cal. 2018)) and Unwired Planet (2017 EWHC 711 (2017)), and the rate charged by a large patent pool on the same standard (Microsoft, 2013 WL 2111217 (W.D. Wash. 2013)).

In Optis, by contrast, the Court calculated the stack value based on comparable licenses from the defendant-licensee.[3] At a high level, the Court started by determining the royalties the defendant paid for those licenses, scaled the royalties up to determine the implied stack value based on the proportion of the SEP universe in each portfolio and then applied several subjective adjustments to reduce potential sources of error. The Court declined to consider 19 licenses Optis previously granted to all or a subset of its portfolio, finding these licenses to be non-comparable and a poor indicator of the aggregate royalty.

The Court’s approach is somewhat surprising. What others paid for the Optis portfolio would seem, at first blush, to be the most direct evidence of the market price for that portfolio. But two observations fueled the Court’s skepticism about the Optis licenses. First, the previous Optis licensees were smaller, less sophisticated players with less incentive and power to negotiate compared to the defendant. The Court concluded that these smaller licensees had different incentives and that Optis likely entered the prior licenses to develop a “forensically advantageous” precedent that it could wield against larger players, such as the defendant. Second, the Court was committed to calculating the FRAND rate via a top-down approach. To ensure an accurate result, it sought the most accurate assessment of the inputs (including the stack value) to the top-down formula. The Court noted, however, that the Optis licenses (at most) indicate the price of only the Optis portfolio—a small fraction of the relevant SEP universe. By contrast, the defendant’s proposed comparable licenses indicate what it paid for portfolios collectively covering a broader, more diverse swath of the SEP universe and, thus, the Court viewed them as more reliable indicators of the overall stack value.

Optis is the latest indicator of a geographical shift in top-down jurisprudence. In TCL, 943 F.3d 1360 (Fed. Cir. 2019), the US Court of Appeals for the Federal Circuit relegated FRAND rate determinations to the jury, all but ending the development of the top-down approach in US district courts. But top-down is far from dead. As evidenced by the recently-proposed EU regulations on SEP licensing and now Optis, European authorities continue to explore the top-down approach as a tool to resolve FRAND royalty disputes.


[1] In fact, in Interdigital, the Court ultimately rejected the result of a top-down cross-check proposed by the licensor because it was out of line with the benchmark rates suggested by what the Court determined was the correct comparable licenses analysis. (See Interdigital, 2023 EWHC 539 at ¶ 881.)

[2] US courts have adopted diverse approaches to this issue. In TCL, Judge James Haberman of the Northern District of Illinois started from a count of declared patents and then adjusted downward to account for over-declaration and invalidity. In Innovatio, Judge James Selna of the Central District of California adopted an expert’s estimated count of patents based on a keyword search to identify patents “potentially essential to the” standard, which would presumably not be limited to only patents formally declared to be essential.

[3] The Court considered but ultimately rejected other proposed measures of the FRAND rate, including a 15% ad valorem rate based on analysis of known “headline” rates, which the Court found to be unreliable; the profit margin for the SSPPU (the cellular chipset), which the Court found insufficiently representative of the stack value to the defendant; the benchmark royalties determined in Unwired Planet, which was based on facts not before the Court in the present case; the profit margin of a basic-level 4G cellular phone with minimal features unrelated to cellular service, which the Court believed was relevant only to ad valorem rates; and Optis’s proposed comparable licenses.