The judgment of Marcus Smith J in Optis v Apple is worth close consideration for anybody negotiating or litigating over what licence terms for standard essential patents are FRAND (Fair, Reasonable and Non-Discriminatory).
First, it discloses a relatively simple method for determining a FRAND value for a licence under a portfolio of 4G standard essential patents, which was devised by the Judge based on the limited portion of the record on which he felt that he could rely. Based on the figures in the judgment, the theoretical FRAND annual lump sum rate that Apple would have to pay for the whole 4G Stack is US$ 1.35 billion. Adjusting this figure to take into account (a) the portfolio's fraction of the Stack, and (b) the fraction of handsets sold by the Implementer compared to Apple, will yield the figure. For example, if the patentee owned, say, 2.5% of the patent families in the Stack, and the implementer's handset sales were, say, one fifth of Apple's, the annual licence fee would be US$ 6.75 million (US$ 1.35 billion x 2.5% x 1/5th).
Second, it contains a careful analysis of the different licence valuation methods advocated by the parties, which the Judge considered to be one or more of "indefensible", "flawed", "utterly unreliable", and "worse than useless". Parties should think twice about relying on such methods.
Third, the Judge considered that each party's criticism of the conduct of the other was a waste of time and money. He described Apple's case that Optis had abused a dominant position as "hopeless", and Optis's case that Apple had engaged in Hold Out as "legally irrelevant".
In summary, the UK is building a body of publicly available FRAND case law, comprising long and well-reasoned judgments following hard fought and well-resourced litigation. The present judgment comes just weeks after the decision of Mellor J in InterDigital v Lenovo, discussed here. The judgments (Unwired Planet, Interdigital, and Optis) are a resource for parties interested in how to (and how not to) engage in negotiations and litigation. This ought to be make subsequent cases in the UK much less expensive and much less complex.
The Headline Maths
The relevant standards were 2G GSM, 3G UMTS and 4G LTE (Standards). 5G was not considered. As of April 2018, a total of 26,600 families had been declared as essential to the Standards (Stack). Optis's portfolio comprised 135 of these patent families. The Judge calculated Optis's share of the total Stack 135/22,000 - 0.61%. The Judge chose the smaller denominator to reflect that "the absolute Stack size was increasing over time".
The Judge considered that, if Apple had required a licence of all 135 families, an annual lump sum payment of US$ 8.235m would be FRAND. However, for reasons that are redacted in the judgment (but possibly related to an earlier settlement), Apple only needed a licence for some of them. Accordingly, the appropriate annual rate is US$ 5.13 million.
The term of the licence is from 1 January 2023 until the expiry of every patent in the portfolio. The Judge considered that payment of five times the annual fee was "an appropriate price for Apple to pay for that length of licence" - US$ 25.65 million. He also considered that a general release for six years (from the beginning of 2017, when Optis first approached Apple, until the end of 2022) was required - US$ 30.78 million. Interest at 5% per annum, compounded every 6 months, on each payment that should have been made was also appropriate. Although the Judge did not make this calculation, an estimate is approximately US$ 5.84 million.
So, in summary, Apple will be required to pay to Optis a lump sum of approximately US$ 62 million for a licence covering all Apple products. The licence terms also require settlement of all foreign litigation under the portfolio.
The Judge's method
Before explaining the Judge's method, it is appropriate to explain why he felt he needed to adopt it.
The trial comprised 20 hearing days, nine experts, and seven witnesses of fact. Optis was represented by three King's Counsel, four junior barristers and two firms of solicitors. Apple was represented by four KC's, four juniors, and one firm of solicitors. Notwithstanding the length of the trial and the number of witnesses, documents, and lawyers involved, the Judge rejected so much of the parties' evidence and submissions that he had to give careful consideration to whether he had enough material in the record to decide the FRAND question.
He recognised the approach in other cases involved an extrapolation from a single or small number of truly comparable licences:
"A comparable that was fairly negotiated, between a willing buyer and a willing seller, and was genuinely similar to the present case, would have significant evidential value."
However, such an approach (which was used in Interdigital), was simply not available in the present case.
The licences that Apple put forward were licences between it and other SEP owners. They were not really comparable, because they were for different portions of the Stack. Optis contended that they were also not helpful because they were obtained by Hold Out, so the rates were lower than FRAND. The Judge did not accept this submission, including because the counterparties to the Apple licences were “big beasts” and well able to look after themselves. Nevertheless, he undertook to factor into his assessment, that Apple may have been able to deploy its "undoubted commercial clout and market power" to negotiate their terms.
On the other hand, while the licences Optis put forward were all for the same portfolio (i.e. the very thing that the Judge was tasked with valuing), he felt that he could not rely on them, including because Optis had been trying to game the system:
"[I]t is difficult to avoid the conclusion that Optis was not dealing with these small counterparties for the revenue streams they would bring in, but because the licences they agreed would produce comparables that would assist Optis in this litigation."
To compound problems, the Judge considered that the attempts by the respective experts to "unpack" the licences, for example to convert a lump sum payment to an ad valorem rate, or vice versa, merely introduced more uncertainties, making an assessment based on the unpacked licences unreliable.
Indeed, the Judge was very critical of the expert evidence. He had made case management directions requiring each party to (a) put their "cards on the table" in a position statement supported by evidence, and then (b) respond to the other side's case by challenging (or agreeing with) their maths. During pre-trial reading, the Judge realised that the parties had not done this. He wrote to them, asking for a summary table characterising each "comparable" licence, so that he could consider each party's case on a "like-for-like" basis. This didn't work.
"Both experts “unpacked” in accordance with their respective client’s instructions, and failed to produce agreed common workings of their unpacking. I have referred to this unhelpful approach by both parties above and, as I have described, my efforts to rectify the problem came to nothing."
The Judge ultimately found that the experts had not provided sufficient independent judgement when preparing their reports. Accordingly, he derived no benefit from either valuation expert, and rejected their evidence as unhelpful.
He then did the best that he could with what evidence was left:
"The best approach … is to seek to price the value of the entire Stack to Apple, and then to apportion that price pro rata amongst the co-owners of the Stack in proportion with their holding … I am not making any assessment of the value of the individual patents. I am pricing the Stack and what Implementers (and, specifically, Apple) should pay for it."
To value the Stack, he used the licences that Apple had entered into with other SEP owners. In each case, Apple had paid a lump sum for the licence. The judge normalised the financial terms of each licence - to what they would have been for 1% of the whole Stack. For example, if a licence were for 3% of the patent families in the Stack, the values would be divided by 3. For licences of less than 1% of the Stack, the Judge assumed that they were for 1%, to account for the fact often "more is paid for less". He used as many of the Apple licences as possible in this process, because:
"Use of multiple data sources means that outliers or unrepresentative cases can be averaged out, and a safer, more reliable, overall figure obtained."
After discounting some values that he considered to be clearly unreliable outliers, he then took an appropriate value to calculate an average annual lump sum payment for 1% of the Stack, and then scaled that to Optis's share of the whole.
Here it is worth observing that the Non-Discrimination requirement of FRAND is often viewed from the perspective of the implementer. For example, Birss J observed in Unwired Planet that it would not be FRAND for a small new entrant to the market to have to pay a higher royalty rate than an established large entity. Mellor J agreed with this in Interdigital v Lenovo.
In this case, Marcus Smith J added that the question of non-discrimination also arises as between SEP Owners. Since the key differentiator between SEP Owners is the proportion of the Stack that they own, to avoid discrimination an implementer ought to pay the different SEP owners on the same basis according to that proportion.
Further, since it was impossible to assess the essentiality and validity of every patent in a portfolio, let alone of every patent in the Stack, it would be impossible to assess whether or not the "quality" of a particular portfolio was above or below average. Any attempt to assess the quality of a portfolio versus the quality of the Stack would just introduce subjectivity and result in greater uncertainty. So, it was better to regard the Stack as essentially homogeneous.
Apple's "indefensible" approach
The Judgment contains a detailed description of Apple's approach to negotiations, which it maintained in the litigation.
Apple's approach focussed on the "smallest saleable patent practicing unit" (SSPPU). For the Standards, the SSPPU is a baseband chipset. Apple's evidence was that such chips sell at approximately US$ 20 per unit, and that average profits were about US$ 5 per unit. So, a licence to all of the SEPs comprising the Standard would need to be a portion of US$ 5 per unit - otherwise chip manufacturers, who needed licence of all of the declared SEPs, could not make any profit. The amount payable to any particular SEP owner would then depend on the size of their portfolio.
Apple also contended that SEP owners should only be rewarded for the value of their technical contribution to the Standards, not just for participating in them. Accordingly, the price for the portfolio needed to be adjusted following an assessment of its essentiality, validity, and technical merit, on a patent-by-patent basis.
The Judge rejected Apple's approach as indefensible.
First, he considered it absurd to presuppose that a chipset manufacturer would forego profit instead of passing the cost of the licence on to the handset manufacturer. If the chipset came with an implied licence of all SEPs in the Stack (alternatively, if the patent rights were exhausted), then the price premium would the same as the amount the handset manufacturer would pay for a licence.
Second, it was not practically possible to review the contribution of each SEP in a portfolio. Here, the Judge noted that the point of technical trials was not to get a sense of portfolio quality, but to establish the court's jurisdiction.
Third, the Standards themselves had value. The value of the Standards to the consumer, and therefore to handset manufacturers, was access to cellular connectivity – allowing handsets embodying the Standards made by different implementers, operating in different territories, via different network providers, to connect one to the other. The Judge considered that Implementers ought to pay SEP owners for their respective contributions to this access.
In the Judge's words:
"The US$5 has an entirely specious validity and precision. Apple’s SSPPU approach is to be rejected for this reason alone; but, much more fundamentally, focusing on the SSPPU in no way assists in deriving a price for the technology licence here in issue".
He went on to explain, perhaps more controversially:
"Where an SEP Holder owns a part of the Stack that enables use of the Standard, then a part of the total Stack price should be payable to the SEP Owner – irrespective of the validity, essentiality or importance of the patents owned, because the ETSI process for the creation of a Standard works by reference to declaration. What is being bought and sold is not a bundle of intellectual property rights, but an ability to use the Standard."
Optis's three "utterly unreliable" approaches
Optis proposed three different methods for arriving at a FRAND rate for its portfolio:
- Scaling up from the Judgment of Birss in Unwired Planet, since Optis had acquired Unwired Planet's portfolio and then built upon it;
- Using the Optis comparables as a reference point; or
- Adopting a top-down approach - with the aggregate royalty for the Stack being 15% of the handset sale price, and taking into account the high quality of Optis's portfolio.
The Judge had no hesitation in rejecting the first as "utterly unreliable". The parties, issues, and evidence before Birss J were very different.
The Judge also had no hesitation in rejecting the second as "entirely unreliable". The Optis comparables were not genuine, and extrapolating them resulted in "grossly excessive" rates. Further, the process of unpacking them introduced subjective assessments, with the result that they were "worse than useless at deriving a rate for Apple to pay."
Optis's third approach was also deeply flawed. The Judge considered that tying the cost of a licence to the average sale price of particular handsets was contrary to law. Such an approach improperly tried to capture extra value associated with the technologies, innovations and components of parties other than the SEP owner. Second the royalty was far too high. And third, there was no practical way that the Judge could assess the quality of the portfolio (or Stack). In any event, two Optis documents - over which it had initially claimed privilege – tended to show that the portfolio had been constructed strategically. That is, it comprised some “litigation grade” patents, intended to make good any claim in the courts (whether on validity or essentiality); some that read onto the Standard, to enrich the apparent quality of the pool; and some “makeweights”.
Hold up, hold out, and abuse of dominant position
Apple accused Optis of 'Hold Up' and abusing a dominant position.
Optis accused Apple of 'Hold Out'. First, in defence of Apple's allegations of abuse, and, second, in support of its case that the prices that Apple had paid for other portfolios were not FRAND.
The Judge considered that these allegations went nowhere. The short answer to them was that:
"… given the regime that applies in these courts – the allegations are in fact entirely irrelevant and both Optis and Apple are to be criticised for raising them and wasting valuable time and money in litigating them".
The Judge considered that Hold Up and Hold Out were "manifestations of rights that are vested respectively in SEP Owners and Implementers". SEP Owners are entitled to seek an injunction, and Implementers are entitled to demand that the SEP Owner establish infringement. Indeed:
"Licence terms have to be negotiated, and these negotiations will be complex and hard fought. A certain degree of Hold Up and Hold Out is to be expected as inevitable."
Apple's claim of abuse of dominant position was "hopeless". If Apple had been forced to pay an abusive price for the Optis portfolio, or if it had been excluded from the market, or inhibited in its commercial activities, then theoretically such a claim might lie. However, none of these things had happened, or could happen, including because of Optis's undertaking to ETSI (to offer to licence its portfolio on FRAND terms).
On the other hand, Optis's arguments that Apple had engaged in Hold Out were not only "misconceived and wrong", but they brought an uncomfortable focus onto Optis's own behaviour during negotiations. The Judge considered that Optis's attempts to tie licence rates to the selling price of Apple's products were not FRAND. In any event, its offers defied "logic and economics". For example, extrapolating its demands to the entire Stack would suggest a total royalty of over US$ 365/unit.
The Judge also considered that to the extent that Optis’s offers could be understood, they were "all over the place". He concluded that Optis:
"… failed to conduct the negotiations competently (contrary to their own interests) with the result that Apple simply did not know where Optis were coming from."
Some commentators have published summary assessments of who is the winner and who is the loser in these FRAND cases, asserting the outcomes are out of step with market expectations and so on, but without exploring important context explained in depth in the judgments. Whilst Optis achieved an outcome better than Apple's best offer, it would nevertheless be difficult to characterise it as a "winner" in this case. Having regard to the Judge's criticisms of Optis's case, it is feasible that it will end up with a net liability to contribute to Apple's costs of the FRAND trial. It will also have to bear significant unrecoverable costs associated with technical trials and appeals, and of litigation in other jurisdictions.
The only publicly available version of Marcus Smith J's judgment at the moment is heavily redacted, including to protect third party information. The level of redaction makes drilling into the Judge's workings difficult. However, the Judge expressed the view that a lot of the redactions were probably unjustified. Indeed, he explained that:
"A certain level of information exchange between SEP Owners and Implementers is necessary in order to derive FRAND rates; and controlling information as to stack size and rates so that a market price cannot be discerned runs the risk, in and of itself, of infringing competition law."
So, we can expect another version of the judgment, which much less redaction, to be available soon.