Reaffirming FRAND Fundamentals: A Detailed Examination of Optis v Apple

In a landmark decision handed down on 1 May 2025, the Court of Appeal overturned the High Court’s determination of FRAND terms for a global patent licence between Optis and Apple. The decision reaffirms core principles in UK FRAND case law and provides important guidance for global SEP licensing. Below are the key takeaways, followed by a more detailed analysis. You can also find a link to our stop press article here and podcast on the decision here: 

What you need to know

1. FRAND Determination Overturned - The Court of Appeal set aside the High Court’s FRAND determination and reasserted the primacy of a comparables-based approach to royalty valuation. 
2. FRAND Rate Set at $0.15 per Unit - The Court determined a FRAND rate of $0.15 per unit for Optis’s SEP portfolio. This figure sat between the higher unpacked rate from the Optis-Google licence and the lower unpacked rates from four key Apple licences (Ericsson, InterDigital, Nokia, and Sisvel), aligning more closely with the top end of the Apple-derived rates. In selecting the $0.15 DPU rate, the Court accepted that mathematical precision was neither possible nor required. Instead, it applied a “broad axe” to arrive at a commercially fair and evidentially grounded outcome - a pragmatic judicial style likely to influence future FRAND rulings.
3. Top-Down Analysis Used as a Cross-Check - While the core valuation relied on comparables, the Court used a top-down approach as a cross-check to ensure the selected rate was commercially reasonable. This marks a shift from recent UK cases, where top-down analysis had been largely sidelined.
4. Global Lump Sum Set at $502 Million Plus Interest - Applying the $0.15 DPU to actual and forecasted device sales from 2013 to 2027 (the period of the licence) yielded a lump sum of $502 million. With interest, the total award is expected to exceed $700 million.
5. Rejection of Lump Sum Averaging - The Court criticised the trial judge’s methodology of averaging lump sum values from a broad set of licences. It found that such an approach lacked economic justification, particularly given the divergent rates in Apple’s own licences which demonstrated that not all licences could plausibly be FRAND.
6. Clarification of “Hold-Out” in FRAND Context - The Court rejected the trial judge’s distinction between “legitimate” and “illegitimate” hold-out. It confirmed that any form of hold-out or hold-up distorts negotiations and should be excluded from a FRAND determination. Apple’s insistence on patent-by-patent licensing was evidence of hold-out, undermining the reliability of its comparables.
7. Treatment of US Proceedings - The Court addressed the overlap with US litigation, holding that Optis could retain its damages award from the Eastern District of Texas. It rejected attempts to force Optis to vacate the US judgment and ruled that the damages should be treated as a floor when calculating the total FRAND payment under the Court determined licence - a resolution that respects international comity.
8. Implications for Future FRAND Disputes - Overall, this decision reinforces the UK’s role as a leading forum for resolving global SEP licensing disputes, while offering important guidance on how FRAND assessments should be structured and how foreign proceedings should be factored into the outcome.

Detailed Summary

The case concerned Apple’s use of standard-essential patents (SEPs) in Optis’s telecommunications portfolio, and followed a lengthy multi-jurisdictional dispute, including parallel proceedings in the Eastern District of Texas (EDTX). The appeal was brought by Optis against Mr Justice Marcus Smith’s decision, which had set licence terms following Trial E in the UK’s FRAND framework.

The Court of Appeal, comprising Lord Justices Birss, Arnold and Newey, allowed Optis’s appeal on the principal grounds and substituted its own FRAND determination, although several appeal grounds were abandoned or not pursued. Optis had been granted permission to appeal on 25 grounds, covering both valuation methodology and the form of the licence imposed at first instance. Apple supported the first instance decision through a Respondent’s Notice advancing six grounds of its own. The Court addressed the central issues of royalty valuation, procedural fairness, and the treatment of overlapping foreign proceedings. Birss LJ delivered the leading judgment on FRAND methodology, while Arnold LJ gave the principal judgment on the non-royalty terms of the licence. All members of the Court were in agreement. The key aspects of the decision are summarised below.

Trial judge’s rejection of expert evidence

The Court of Appeal held that the trial judge’s wholesale rejection of expert evidence from the parties’ accountancy experts (Mr Bezant and Ms Gutteridge) was both procedurally unfair and legally flawed. 

Central to the Court’s criticism was the fact that the judge found the experts lacked independence, operated beyond their expertise, or followed narrow client instructions, yet these concerns were never put to them in cross-examination. Making such adverse findings without giving the witnesses a chance to respond amounted to clear procedural unfairness.

The Court clarified that both experts were acting appropriately within their accountancy expertise, tasked with extracting financially useful information from complex licence agreements through unpacking. This role did not require them to take a position on whether ad valorem or dollar-per-unit (DPU) rates were preferable – this is an evaluative judgment for the court, not for expert evidence.

A key conceptual error in the first instance judgment was the conflation of comparability (whether a licence is contextually a good match) with reliability (the quality of data that can be derived from it). The Court emphasised that comparability is for the Court, whereas reliability falls within the experts’ domain. The unpacking process was not an attempt to assert comparability but to improve the reliability of data used in comparing licences.

Rejection of the trial judge’s lump sum methodology

The Court conducted a detailed assessment of the trial judge’s lump sum methodology for determining the FRAND rate. It noted that both parties had abandoned key parts of their original cases on appeal, simplifying the valuation task. In particular, Optis dropped reliance on all of its licences except its licence with Google, accepted its 0.38% patent stack share, and agreed to use Innography data for SEP counts. This made the trial judge’s methodology the focus of the appeal.

The judge’s approach, summarised in Table 13 of the original judgment, listed the 19 licences being used as comparables (18 Apple licences and the Optis-Google licence) in descending order according to the total stack price they implied. The judge then took a simple average of the rows of the table to obtain the average implied stack price, after having adjusted the row for Qualcomm which was identified as being a cause of “significant distortion”. Optis’s stack share was then applied to the result to derive an annual royalty. While the Court accepted that placing licences on a common scale is reasonable in principle, it identified several fundamental flaws with the methodology which had been adopted.

First, averaging the entries in the table from highly divergent licences with no clear economic basis was unprecedented, unsupported by the evidence, and inconsistent with expert methodologies. Even a weighted average, as Optis proposed, did not address the fundamental flaw, namely that the licences were too different to be treated as equally valid comparables. The wide spread of values in Table 13, despite exclusions, demonstrated that not all licences could plausibly be FRAND. The Court concluded that averaging such divergent figures obscured rather than clarified FRAND value.

Second, the treatment of the Google licence revealed a serious inconsistency. The judge had included Google in the lump sum table without adjusting for vastly different sales volumes (Google sold less units than many of the other implementers), thereby underweighting its importance. Using an unpacked value for the Google licence which properly takes into account sales volumes suggested a far higher total royalty burden. The method used by the trial judge thus masked critical differences in licence structure and commercial context.

Third, the Court rejected Apple’s argument that excluding Google or focusing solely on Apple’s own licences could salvage the approach. Apple’s sales volumes varied significantly across those licences, and the judge had deliberately chosen to ignore volume-based considerations. His method could not be rescued by assuming volume constancy or dismissing the Google licence as a nuisance value, especially as the trial judge had expressly rejected that view below.

Finally, the Court found that the judge’s method exemplified the warning in Cimetidine: using less reliable comparables to dilute the weight of stronger ones. It held that the proper approach was a comparables-based method focused on identifying and relying on the best available comparables, consistent with established FRAND jurisprudence.

Hold out and the Apple comparables

The Court of Appeal held that the trial judge had erred in his treatment of Apple’s prior licences (“the Apple Comparables”) by misapplying the concept of “hold out” in FRAND analysis. Optis argued that Apple’s strong negotiating position and insistence on patent-by-patent licensing suppressed royalty rates, rendering the Apple Comparables unreliable.

The trial judge had distinguished between “legitimate” and “illegitimate” hold out, treating Apple’s conduct as legitimate negotiation. The Court of Appeal rejected this framework as flawed. Citing the Supreme Court’s Unwired Planet decision, it confirmed that both hold out and hold up are forms of “mischief” the FRAND system is designed to prevent. While not unlawful, such behaviours distort licensing outcomes and may result in terms that are not FRAND.

The proper test, the Court held, is the willing licensor/willing licensee standard. This is an idealised benchmark that excludes both hold out and hold up. The judge erred by treating Apple’s conduct as legitimate and overlooking how its market power depressed rates. In particular, Apple’s insistence on a patent-by-patent approach, described by the judge as “indefensible,” was clear evidence of hold out. Analysis of the Apple licences showed that larger licensors (e.g. Ericsson, InterDigital, Nokia) secured higher rates, while smaller licensors accepted lower ones, supporting Optis’s case that the comparables were affected by hold-out.

The Court therefore concluded that trial judge was wrong to place weight on the values derived from the Apple licences as a whole.

Was a retrial necessary?

The Court of Appeal considered whether a retrial was necessary following its findings against the trial judge’s methodology. Optis did not seek a retrial and asked the Court to determine FRAND terms directly. Apple argued a retrial was warranted if the Court adopted Optis’s new case based on the Google licence, claiming this would have affected its trial strategy.

The Court rejected that argument, noting the Google licence was already a key comparable at trial and Optis’s shift from an ad valorem to a DPU-based approach did not unfairly prejudice Apple. Both parties now agreed on major inputs, including Optis’s 0.38% stack share and the use of Innography data.

Citing Simetra v Ikon, the Court reiterated that a retrial is a last resort and only appropriate if it is the only just course, which was not the case here. It concluded it could determine FRAND terms itself using a broad-axe DPU-based analysis, and no retrial was required.

The FRAND rate set by the court

The Court of Appeal conducted a detailed assessment of the expert evidence to determine an appropriate FRAND royalty. 

Starting with the Google licence, Optis’s expert Mr Bezant calculated a “simple” unpacked DPU, then scaled it to reflect Optis’s 0.38% share of the SEP stack, which was used for the final FRAND analysis (reduced from 0.61% to account for the effect of the recent Ericsson licence between Apple and Ericsson). This scaling produced a significantly higher per-unit rate than any derived from the Apple licences. The Court then examined unpacked DPU figures for the Ericsson, InterDigital, Nokia, and Sisvel licences (the four Apple licences with the highest implied royalty rates), adjusting each by the respective licensor’s stack share to place them on a common scale. The result showed that even these stronger Apple comparables implied rates materially lower than the Google-derived rate.

Both experts, Mr Bezant and Ms Gutteridge, had also unpacked these licences using a “free release” method, with broadly consistent results. The free release rates were slightly higher than the simple unpacked rates, but not materially so. In both sets of data, the same pattern emerged: the highest implied rates came from the same four Apple licences, but none matched the level implied by the Google agreement.

The Court drew several conclusions. First, any FRAND rate for Optis had to be at least as high as those implied by the top-tier Apple comparables. The trial judge’s conclusion, which implied a DPU of $0.01–$0.02, was “clearly too low”. Second, the Google licence showed that relying solely on the Apple licences would produce a rate that was also too low. Third, the rate implied by the Google licence alone was also too high when cross-checked against Apple’s and Google’s average selling prices (ASPs). Applying a 15% total royalty burden to Google’s average selling price (ASP) of $470 yielded a maximum stack DPU of $70.50. Optis’s share of that stack would be $0.27 at 0.38% or $0.43 at 0.61%, both of which are lower than the DPU directly unpacked by Mr Bezant from the Google licence. Conversely, using Mr Bezant’s unpacked rate implied an overall stack burden well above 15%, which the trial judge had already found too high.

The Court ultimately concluded that a DPU of $0.15 per unit was FRAND, selecting a rate that lay between the higher figure implied by the Google licence and the lower rates derived from the best of the Apple comparables. Although these licensors were not identically situated, the Court considered them sufficiently reliable, while acknowledging that varying degrees of hold up and hold out were likely involved.

As a cross-check, the Court noted that grossing up the $0.15 rate to represent a 100% royalty stack implied a total royalty burden of approximately 6.3% of Apple’s historical ASP ($625), 3.9% of a more current ASP ($1000), and 8.4% of a Google ASP ($470)—all commercially reasonable figures.

The Court also confirmed that the $0.15 rate sits between the higher rate implied by the Google licence and the lower rate from Ericsson, and is much closer to the Ericsson rate, reinforcing its suitability. The free release unpacking figures for Ericsson from both experts did not materially affect this conclusion.

To convert the per-unit rate into a lump sum, the Court used Optis’s spreadsheet model, which applied the $0.15 DPU to total unit sales derived from actual past and projected future sales from 2021 to 2027. A 10% discount was applied to projected volumes. The resulting lump sum of $502 million (excluding interest) was taken to reflect a FRAND licence covering the full period from 2013 to 2027, consistent with the original licence term adopted by the trial judge. This sum encompasses both past infringement and forward-looking use and was ordered as the lump sum payment under the final FRAND licence.

Procedural unfairness of paragraph 6(2) of the trial judge’s final Order

Optis challenged paragraph 6(2) of the trial judge’s final order - which restricted Optis’s ability to act in respect of the US proceedings without permission from the Patents Court - as procedurally unfair, arguing that it resembled an interim anti-suit injunction (ASI) that was imposed without being sought by either party or properly argued. Apple disagreed with that characterisation, but the Court of Appeal agreed with Optis that the provision was imposed unilaterally by the judge without proper procedural safeguards. While Optis had a formal opportunity to object during consequential submissions, it was clear the judge had already made up his mind. However, the Court found that any procedural unfairness had since been cured on appeal, as Optis had now had a full opportunity to challenge the provision.

Treatment of the EDTX Judgment and Foreign Proceedings

Optis appealed the trial judge’s inclusion of certain terms in both the FRAND licence and the final order that interfered with the ongoing US litigation in the Eastern District of Texas (EDTX), particularly the final US judgment awarding damages for infringement of five US patents. It argued that these terms were not consistent with FRAND principles and were contrary to international comity. 

The central issue was how to address the potential inconsistency between the US damages award and the FRAND royalty determined by the English court (a conflict that only crystallised after Apple changed its position on 15 September 2023, committing to take the UK Court determined licence). Optis made clear it did not seek double recovery, but contended that a willing licensor and licensee would have found a way to accommodate the US result within the global licensing framework.

Optis proposed three possible solutions:

1. Adjust the English FRAND royalty to reflect the actual US award by replacing the English valuation for the five US patents with the US judgment sum;
2. Deduct from the FRAND royalty an amount representing the global value of the five US patent families; or
3. Treat the US judgment as a ‘floor’, allowing Optis to retain the US award and receive the balance of the FRAND sum if it exceeded that figure.

The Court agreed that the first two options were impractical due to the valuation methodologies used by both sides. However, it found the third option (treating the US judgment as a floor) to be fair, pragmatic, and the least-worst solution. In particular, the Court held that Apple’s litigation conduct had created the inconsistency, and a willing licensee in Apple’s position would not expect the licensor to abandon the US award, especially without reimbursement of litigation costs.

The Court also rejected Apple’s arguments that Optis needed expert or factual evidence on licensing practice, holding that the FRAND question in this context could be answered without such evidence. It likewise disagreed with the judge’s characterisation of Optis as having changed its case or being at fault for the situation, finding instead that Optis had consistently reserved its position and that Apple’s September 2023 shift in stance was the real cause of the conflict.

Further, the Court held that comity supported this approach. The US judgment was properly obtained in a competent forum under US law, and the English courts should not interfere with it absent compelling reasons, none of which were present.

Given these conclusions, the Court held that paragraph 6(2) of the order, even if not a formal ASI, was unjustified. It was unnecessary, unclear in scope, and risked creating an inappropriate regime of ongoing court supervision. The same objections applied to related provisions in the licence itself. The Court expressed doubts as to whether such terms were even within the court’s jurisdiction when determining FRAND licence terms, though it did not decide that point.

Final thoughts

The Court of Appeal’s judgment in Optis v Apple marks a significant reaffirmation of the principles governing the determination of global FRAND terms under English law. It re-establishes the centrality of a comparables-based methodology, firmly rejects arbitrary averaging of divergent licence data, and restores the evidential value of expert unpacking. This is likely to increase predictability and clarity for parties seeking FRAND valuations from the UK Courts.

While recent UK decisions had downplayed the relevance of top-down approaches, the Court here applied it with renewed significance. Although it was careful to state that the determined FRAND rate was based on comparables, the top-down check appears to have played a material role in positioning the rate within the $0.10 to $0.27 DPU range. In effect, it helped justify why $0.15 DPU - a figure implying a 6.3% total royalty stack relative to Apple’s historical ASP - was commercially reasonable.

That said, the Court offered little detail on the evidential basis for accepting that 6.3% figure as a realistic total stack burden, which may invite further scrutiny in future cases. Also of note is the Court’s willingness to adopt broad approximations, applying what Birss LJ described as a “broad axe”, a concept more usually seen in the context of damages case law, to reach a fair outcome grounded in the evidence, rather than seeking mathematical precision where it was not warranted.

Overall, this decision reinforces the UK’s role as a leading forum for resolving global SEP licensing disputes, while offering important guidance on how FRAND assessments should be structured and how foreign proceedings should be factored into the outcome.