The UK Court of Appeal has issued a long-awaited judgment in a standard essential patent (SEP) case filed by three subsidiaries of PanOptis Holdings, LLC (collectively, “Optis”) against Apple. On May 1, the appellate court overturned a May 2023 decision from High Court Justice Marcus Smith that had set a global fair, reasonable, and nondiscriminatory (FRAND) rate of $56.43M, ruling that he had had erred by rejecting the parties’ expert evidence and rate-setting arguments in favor of a rate-setting approach not proposed by either party. The Court of Appeal instead ruled that the proper FRAND rate was $502M plus interest.
The litigation concerned 4G-LTE-enabled iPhones first introduced in 2012. Apple and Optis began licensing negotiations in 2017 but did not reach an agreement, after which Optis sued Apple in the Eastern District of Texas and in the UK High Court’s Patents Court in February 2019. The Texas action went to trial in August 2020, at the end of which a jury returned a verdict awarding $506.2M in damages. However, that jury heard no mention of FRAND principles whatsoever, after PanOptis requested a separate bench trial for issues related to its FRAND licensing obligations—leading District Judge Rodney Gilstrap to overturn that verdict as unacceptably ambiguous and grant a new trial on damages, at Apple’s request. A second jury returned a lower figure, $300M, in an August 2021 damages retrial. Prior to the retrial, Judge Gilstrap declined to stay the case pending the outcome of the UK litigation.
The UK action, meanwhile, saw four technical trials, Trials A-D (i.e., addressing issues relating to the patents—infringement, validity, and essentiality), that took place from October 2020 through January 2022. Two trials were then held on FRAND issues (Trials E and F), the former held in June and July 2022 to determine a FRAND rate. In Trial E, the parties differed on the proper amount but agreed on certain aspects of the approach to determining it, including the use of comparable licenses and the need to unpack those licenses that provided a lump sum into per-unit royalties, though their experts differed on the how to handle past sales vs. future use. Following Trial F, which was held in July 2021, the court held that same September that “Optis was entitled to an injunction unless and until Apple undertook to take a licence on the terms determined by the English courts to be FRAND”, an undertaking Apple gave the following month without prejudice to its right to appeal.
A key dispute during Trial E was the proper way to determine per-unit rates. At the time, the only FRAND trial judgment was the UK High Court’s decision in Unwired Planet v. Huawei, which adopted an ad valorem approach where per-unit royalties are expressed as a percentage of a unit’s selling price. Optis sought to unpack an ad valorem rate from its comparable licenses. However, Apple argued that this was wrong because its devices were sold at a much higher price for reasons unrelated to the technology here at issue—countering that the rate should be expressed as a fixed sum per unit (a “dollar per unit” or “DPU” approach). In addition to these “bottom-up” approaches, both parties also used a top-down approach—which seeks to determine the patent owner’s share of the total royalty stack—as a cross-check.
In his May 2023 judgment, Justice Smith held that the license fee should amount to a yearly lump sum of $5.13M applied for a period of 11 years—including five years from the date it was to be entered, plus six years of past sales—adding up to a total of $56.43M. Justice Smith reached this amount after rejecting the unpacking approach proposed by both parties, deciding instead to “manipulate the data in terms of lump sums and also, rather than identifying the best comparables, to use an averaging approach”.
By doing so, the court rejected the parties’ “exclusionary” comparables cases wherein both “pick[ed] out the best comparable licences from the pool of available evidence”, declining to accept the parties’ accountancy experts after finding that underpinning their cases on this issue were three “subjectivities”—meaning, different rate types (lump sums, ad valorem, and DPU), the need to adjust for which licenses included cross-licenses, and “future royalties/past releases”—that presented “insurmountable” difficulties. The judge additionally faulted the experts for what he characterized as straying outside their expertise and failing to exercise independent judgment in reaching their respective conclusions.
In recounting the various adjustments made by Justice Smith as part of this “averaging” analysis, Lord Justice Colin Birss (writing on appeal) took particular issue with his decision to treat the total value of the release for past sales as the same as the value for the future, a “50%-50%” split chosen by the judge because “it best disincentivises hold out”. The appellate court remarked that such a choice is inappropriate at this stage of the analysis, at which point the judge should only “evaluate and draw conclusions about what happened in the past”.
On appeal, Optis asserted that the license rate determined by the court was too low, and that the High Court fundamentally erred by discarding the expert evidence and adopting its own approach. Optis countered that the proper basis for a license rate should be a “conventional comparables approach” (as characterized by the court) based either on a comparable license between it and Google or between Apple and Qualcomm, which using a “simple” unpacking should be reduced to a per-unit fee that would then be scaled for Apple. (While the actual relevant sales numbers are confidential, the appellate court noted that the resulting lump sum would be “plainly a far larger sum” than the yearly license fee determined by the High Court—Optis claiming at trial that the correct amount could be as high as $7.4B.) In contrast, Apple argued that Justice Smith’s approach had been correct and “in fact was generous to Optis”. To the extent that any changes to that approach were necessary, Apple only sought the exclusion of the Optis-Google license.
Judgment by Lord Justice Birss
Lord Justice Birss, writing for the Court of Appeal to “address the money”, began with the exclusion of the evidence from the parties’ accountancy experts, starting by agreeing with Optis that Justice Smith had been wrong to reject the experts’ evidence. In particular, with respect to whether the experts had exceeded bounds of their expertise, Lord Justice Birss found that Justice Smith had been “distracted” by references to the experts as “valuation experts”, when in fact they had not been expressing what form of royalty rate would be FRAND—only seeking to extract information from the cited licenses to aid the court in making that determination. Lord Justice Birss also indicated that Justice Smith had conflated the issues of comparability itself, which “is primarily concerned with the situation of the parties and the subject matter being licensed”; with the reliability of the evidence used to unpack the underlying evidence.
Next, Lord Justice Birss addressed Justice Smith’s approach, as laid out in a table wherein “19 licences are placed on a common scale normalised by patent stack share”, the scale being the “implied total lump sum stack price”; taking a simple average of this figure; making adjustments to account for past releases; and then applying Optis’s stack share to arrive at an annual sum applicable to Optis. Lord Justice Birss found the use of a common scale to be logical but faulted Justice Smith for having removed a 20th license (Grand Mesa) and adjusting (in part) for his determination that another license (Qualcomm) was impacted by “hold up”, as Justice Smith “was in fact adopting the exclusionary approach previously deprecated”.
However, the “real issue”, per Lord Justice Birss, was with Justice Smith’s use of a simple average of the entries in the table, finding that this approach “had no precedent or basis in the evidence before him nor can it be justified in principle”. The court also found the weighted average suggested by Optis as an alternative was no more justifiable—determining that “[b]oth forms of averaging share the same weakness in giving weight to all licences in the table as comparables, albeit that weight is different”. Yet neither approach accounted for the fact that even with the aforementioned exclusions/adjustments made by Justice Smith, “the range of values is very wide”: “The obvious conclusion is that either this approach does not work or that these licences cannot all be useful comparables. The approach puts them too far apart from each other”. Lord Justice Birss then took this point further: “In truth there is a simple explanation for the spread, which is not solved by taking any kind of average. Either these licences are not all equally good comparables for various reasons or, putting it another way, not all these rates are FRAND”.
Lord Justice Birss next turned to Optis’s argument that the lower court was wrong to reject its claim that “the Apple licences were tainted with hold out and not FRAND” due to certain practices by the defendant, such as pushing for a patent-by-patent licensing approach. Here, the court held that Justice Smith erred by drawing a distinction between “legitimate” and “illegitimate” hold-out, and by placing all of Apple’s licenses on the “legitimate” side of the line. Considering the licensing conduct at issue in light of the spread of values in the table including the Apple licenses, Lord Justice Birss concluded as follows:
When it can do so, Apple’s significant negotiating strength leads some parties to agree lower rates than would be agreed between a willing licensor/willing licensee. There is a degree of hold out involved. Using the judge’s Table 13 and ignoring Qualcomm and Blackberry, by simple inspection it is apparent that although there are exceptions (e.g. Huawei and LGE), by and large it is the organisations with larger stack shares like Ericsson, InterDigital and Nokia who have agreed licences with Apple which imply the total stack is more valuable than the stack implied by the licensees with smaller stacks.
As a result, Lord Justice Birss held that the lower court was wrong to give weight to the Apple licenses, finding that hold-out was “an explanation for the spread of values of these licences whenever they are put on a common scale by stack share”.
After rejecting Apple’s request that the case be sent back to the High Court for a retrial, Lord Justice Birss held that the Court of Appeal could “reach a just conclusion on the FRAND rate starting from the position of identifying the potential comparables in issue and analysing them by putting them on a common scale of some kind by reference to the stack share under licence in each case”. This was the core of the court’s approach to the extent not challenged by the parties, he explained, but this time he would use “the DPU data produced by the experts, which [Justice Smith] wrongly rejected”.
Here, while Lord Justice Birss found that Justice Smith had in part been trying to account for both licenses from Optis and licenses to Apple, his “flawed methodology led him to be misled into the impression that they imply similar FRAND prices. They do not”. As such, Lord Justice Birss proceeded to address the proper DPU (again, dollars per unit) based on a relevant royalty stack share of 0.38% for the relevant Optis patents, as unpacked from the Google license offered by Optis. After excluding the Blackberry, Qualcomm, and Grand Mesa licenses as well as three others (ETRI 2017, Orange 2017, and KPN 2019), this left four of the Apple licenses that the court described as those from which the “highest four implied Optis rates” are derived: licenses with Ericsson; InterDigital, Inc.; Nokia; and Sisvel International S.A.
Analyzing data on those licenses, Lord Justice Birss determined that a DPU based on the four remaining Apple licenses would be too low, while one based on the unmodified Google rate would be too high. As a result, Lord Justice Birss landed on a rate between the two, holding that “$0.15 per Apple unit for Optis is FRAND”—explaining that he “reach[ed] that conclusion based on using Google and also Ericsson/InterDigital/Nokia/Sisvel as the best comparables, recognising that they are not similarly situated and that there would appear to be degrees of hold up and hold out involved”. Lord Justice Birss then converted that rate to a lump sum using DPU data from Optis offering two potential bases—going with one that used a total sales volume of 3.347M units, a number “derived by taking all relevant actual sales from 2021 back into the past”, with sales for 2021 to 2027 taking the 2020 figure and multiplying by seven, and then applying a 10% discount for future sales. This resulted in a total lump sum (without interest) of $502M.
Judgment by Lord Justice Arnold
Lord Justice Richard Arnold issued an opinion in which he agreed with the royalty rate judgment from Lord Justice Birss and then dealt with several remaining issues, the first of which were certain non-royalty terms of the license not related to foreign proceedings. One point of contention on appeal for Optis was the manner in which the court set the terms of the license, which Lord Justice Arnold described as “extraordinary and undesirable”: After a months-long process in which the parties hashed out their respective positions and agreed to many terms, they produced a “composite draft license” that laid out the remaining terms in dispute. However, Justice Smith replaced this draft license with one of his own making, leading Optis to complain of a “misunderstanding of the court’s role in declaring FRAND licence terms”, that the terms did not constitute a “commercial license”, were procedurally unfair, “wrongly gave Apple benefits”, and erred with regard to the interest stop date.
By the time of the appeal, the only remaining important legal question related to the interest stop date. Justice Smith had set a stop date of January 1, 2023, but Optis argued on appeal that “Apple should be required to pay interest until the date when Optis receives payment (or, to the extent that the money was paid into court by Apple on 11 March 2024, until that date)”. Lord Justice Arnold agreed with Optis.
Finally, Lord Justice Arnold addressed the parties’ dispute over how to account for potential double recovery between the license rate from this case and the $300M damage award from the Eastern District of Texas. Here, after dismissing various objections from Optis, including procedural fairness, Lord Justice Arnold rejected the notion that the UK court’s judgment could be amended to account for the Texas judgment due to the former court’s valuation methodology making that impractical. Instead, he held that the final amount of the Texas judgment should serve as a floor, and that to the extent the global FRAND rate determined by the UK court exceeded that amount, Apple would pay the balance to Optis.
Judgment by Lord Justice Newey
Lord Justice Guy Newey filed a one-sentence judgment that concurred with the judgments of Lord Justices Birss and Arnold.
For more on other notable FRAND developments so far this year, including other UK appellate decisions, see RPX’s report on the first quarter.
