Thursday, March 23, 2023

Thoughts on InterDigital v. Lenovo

By now, most readers of this blog probably have already read some or all of Mr. Justice Mellor’s exceptionally thorough (225-page, partially redacted) judgment in InterDigital Tech. Corp. v. Lenovo Group Ltd., [2023] EWHC 539 (Pat.)—or, if not the decision itself, some of the commentary that has been published within the past week, including articles by Alex Baldwin on Law360 (here and here),  by Eileen McDermott on IP Watchdog, by Florian Mueller on FOSS Patents, and by Henry Yang on IPKat.  If so, you are aware that the court concluded that a global FRAND license for Lenovo’s use of InterDigital 3G, 4G, and 5G-related SEPs from 2008-23 would be a lump-sum of $138.7 million, equal to a $0.175 per-unit royalty.  This is considerably less than InterDigital was seeking, and substantially more than Lenovo had ever offered to pay (though as others have pointed out, it falls roughly halfway between InterDigital’s lowest offer and Lenovo’s highest).  Mr. Justice Mellor calculated the royalty using comparables, the most important of which was a 2017 license between InterDigital and LG.  Except for this license—which Lenovo proposed as a comparable and InterDigital eventually came around to accepting, in the alternative, as relevant (see para. 387)—the judge concluded that the licenses InterDigital’s experts proffered were not, in fact, comparable (paras. 608-11, 793), insofar as they accounted for only about 2% or so of InterDigital’s licensing revenue, mostly with smaller firms.  He accorded some relevance to some of Lenovo’s other six proposed comparables (see, e.g., paras. 794-814), which accounted for a much greater share of InterDigital’s licensing revenue and involved larger firms including Samsung and Apple. 

Those aspects of the decision that struck me as most important, interesting, or novel include the following:

1.  Although the court expresses this point most forcefully toward the end of the decision, one theme throughout is that InterDigital was an unwilling licensor and that Lenovo was an unwilling licensee (at least some of the time); but that on balance InterDigital, came off worse.  See in particular these conclusions toward the end of the decision:

928. Overall . . . I am driven to the conclusion that by consistently seeking supra-FRAND rates, InterDigital did not act as a willing licensor. . . .


931. . . . Overall, I conclude that Lenovo did drag their heels on occasion and to that extent, did not act as a willing licensee.


932. However, for most of the period of negotiations, my conclusions imply that Lenovo were correct not to agree to any of InterDigital’s offers or positions and justified in seeking information. So, for the most part, Lenovo did conduct themselves as a willing licensee.


933. Once this litigation started, it is more difficult to characterise one side or the other as being unwilling, because each side must develop their case as they see fit. However, the circumstances in this action demonstrate that both sides can be unwilling at the same time.


934. Meade J.’s analysis in Optis F demonstrates that, when Lenovo failed to undertake to HHJ Hacon to take a licence the terms of which were to be determined at this trial, it did not act as a willing licensee. It may also be noted that the objections which Lenovo put up at the FOO hearing in Trial A have evaporated. At the same time, InterDigital were continuing to press for rates which I have found lie comfortably outside the FRAND range, so InterDigital continued to fail to act as a willing licensor.

Note, however, that according to reports, Lenovo has since indicated that it would take a FRAND license, rather than suffer being enjoined from the U.K. market.

2. Relatedly, Mr. Justice Mellor expresses some frustration with the lack of transparency in SEP licensing generally.  (I fully agree; see, e.g., here.)  Consider the following:


200.  . . . There is no doubt in my mind that the SEP universe would be able to converge on and agree FRAND terms very much more quickly if the basics of each SEP licence were made public (by ‘basics’ I mean the number of units covered, the royalty rates or total sum paid/payable and which standards are involved). In other words, the market for mobile telephony SEP licences would work very much more smoothly with transparency of what terms had been agreed in the past.


201. However, it is apparent that, for ETSI, to require transparency was a step too far. Instead, there is the stated expectation that those who are subject to the ETSI undertaking should engage in an impartial and honest licensing negotiation process for FRAND terms and conditions.


202. This has implications for how the SEP licensor should conduct itself as a willing licensor. Consider the following example: a SEP licensor makes an offer to licence an implementer on the basis of payment of a lump sum of $100m for a 10-year licence period. The implementer responds by saying it has insufficient information to be able to assess the offer and requests details, under an NDA, of other PLAs which the SEP licensor has concluded with similarly situated licensees. The SEP licensor does not provide the information. Subsequently, at a FRAND trial, after disclosure and expert evidence, the Court concludes that the FRAND rate for a 10-year licence is indeed $100m. Does that result mean the SEP licensor was a willing licensor? I consider it does not. As Birss J. said in UPHC, FRAND is a process. A willing licensor and willing licensee engage in that process to agree FRAND rate(s). It is not FRAND nor is a licensor acting as a willing licensor if it refuses to provide the information necessary for a willing licensee to evaluate an offer which has been made.

Elsewhere he observes that "this sort of FRAND case is not well suited to adversarial litigation because there is (and was in this case) very little, if any, exploration of the middle ground between the positions taken by the two sides” (para. 53), a point I also find compelling.

3.  As for calculating the rate, the court concludes that the same rate should apply to both past and future sales, and that the parties’ subjective expectations about how to allocate value between past and future shouldn’t matter.  To wit:

423. I propose to adopt the same approach as in UPHC and in TCL v Ericsson, that the same rate should apply to past as future i.e. Mr Meyer’s [Lenovo’s] blended rate approach. This is for two cumulative or alternative reasons.


424. First . . . I formed the impression that InterDigital retained significant room for manoeuvre in the way they apportioned an overall LS [lump-sum] consideration to a past release and therefore as to the sum they ‘recognised’ in their financial reporting as attributable to a past release/past sales. This is reflected in the past and future rates derived by Mr Meyer for the Lenovo 6. One of the consequences of the (relatively smaller) sums which InterDigital ‘recognise’ as attributable to the past is that the (relatively higher) sums attributable to the future result in higher royalty rates for the future. Not only can InterDigital cite these higher future royalty rates as representative to other potential licensees, InterDigital can also use them in any renewal negotiations. Another consequence is the apparently very heavy discounting by InterDigital as to past sales.


425. Thus, my impression is that InterDigital’s allocations of overall LS consideration to past and future was somewhat artificial. These allocations do not feature in the particular PLA [patent licensing agreement] and were not agreed with the licensee. Furthermore, they do not match what I regard as the normal way for others in the market to assess the rate derivable from a LS licence which is to take the total consideration and divide it by the best estimate of the number of units covered to derive a per-unit dollar rate for the purposes of comparison.


426. In case I have formed the wrong impression about these allocations, and they are mandated by accounting principles, there is a second reason (which is an alternative or additional reason) why I consider InterDigital’s allocations should not bind an analysis of what is FRAND and it is this. FRAND is concerned with the relationship between licensor and licensee. Therefore, FRAND rates should focus on the money (and other benefits) which pass between licensee and licensor, with the other benefits being translated into monetary terms as part of the unpacking. FRAND is not concerned with and should not be affected by either one party’s internal justification for the sum paid or received, nor with the way in which one party seeks to deal with those sums in its accounts, whether they are internal or made public, particularly when these internal justifications and financial reporting do not form part of the licence agreement. InterDigital’s consistent approach was to work on the basis of the ‘value’ in their hands of a particular payment. This I find is wrong in principle because it automatically injects InterDigital’s subjective view of that ‘value’ into the analysis.

4. Further as to rate calculation, and as suggested by the above, the judge isn’t impressed with InterDigital’s explanations for the large volume discounts given to the large players (although he states that he is “not deciding that volume discounts of any magnitude are not FRAND” (para. 507, and he recognizes that “discounts which reflect the time value of money . . . are entirely fair and consistent with FRAND” (para. 519)):

495. Having considered all the evidence on the issue of volume discounts I have reached the clear conclusion that the volume discounts said to have been applied to the largest InterDigital licensees (i.e. in the range of 60%-80%) do not have any economic or other justification. Instead, their primary purpose is to attempt to shore up InterDigital’s chosen ‘program rates’. Their primary effect is discrimination against smaller licensees.

He further rejects the argument that the rates should be different, depending on whether the end product is a high-end or low-end model:

502. No doubt chipsets for a given generation of cellular technology will vary in their capabilities and performance, but all will implement the standardised technology along, no doubt, with many optional extras. A more expensive high-end 4GMM chipset might be incorporated into a high-end smartphone with a large touch screen and various other attributes attractive to the consumer. By contrast a lower-end, much more basic phone with the same 4GMM capability may sell for a fraction of the price of the higher end phone. However, in terms of the standardised 4GMM technology, both phones use the same technology. Against this backdrop, I find it difficult to understand why the royalty paid for each of those phones should differ significantly or, for that matter, at all.

I need to think about this some more.  I would be inclined to think that a user who derives more value from a product overall would be more willing to pay more the functionality the SEPs add to the product, but maybe it makes more sense to think of the "nondiscriminatory" aspect of FRAND as precluding this result, at least for the same type of product.

5.  What I personally find to be the most interesting part of the opinion, largely because I don’t recall seeing discussion of this elsewhere in the case law or literature (though I may have missed it), is the court’s discussion of why the statute of limitations really doesn’t matter in the present context.  (I’m inclined to think the court is right, and that doctrinally this result could be justified on the theory that an implementer who wishes to rely on the FRAND commitment in effect waives reliance on the statute of limitations.  I need to think about this some more, however, and I do find it to be something of a bold move on the court’s part.)  Here are the key excerpts:   

528. Having given this issue considerable thought, I have reached the conclusion that limitation periods do not have a role in the relationship between willing licensor and willing licensee and, indeed, that they are inconsistent with that relationship. As I have explained above, a willing licensee will, notionally or otherwise, set aside funds to pay for its licence. If, for some reason, those willing parties are not able to reach a deal for some time (assuming the negotiations last for longer than 6 years), I do not believe that a willing licensee would refuse to pay whatever licence fees were eventually determined to be applicable in respect of units produced and sold more than 6 years prior to the determination. A licensee who did that would no longer qualify as ‘willing’.


529. In my view, a willing licensee would not seek to benefit from delay in agreeing FRAND terms or payment of FRAND royalties. Thus, I have concluded that a willing licensee will pay in respect of all past units. Specifically, I do not consider that a willing licensee would seek to avoid making payments of FRAND royalties by taking advantage of one or more national limitation periods. The willing licensee would say: ‘I have set these monies aside to pay to the SEP licensor(s) and I will pay them over just as soon as the appropriate rates have been agreed or set’. If the position was otherwise, that would automatically insert into the process (and FRAND is a process) an on-going perverse incentive to delay the agreement or setting of FRAND terms for as long as possible i.e. the longer the delay, the less the licensee has to pay. This cannot be FRAND. . . .


532. If a Defendant to this type of action does not want to invoke the Claimant’s undertaking to ETSI, then, assuming one or more SEPs have been found to be valid and essential, the Defendant is highly likely to be subject to an injunction to restrain further infringement in the future and to have to pay damages in respect of past infringements. Assuming also that the Defendant has pleaded limitation as a defence, damages will only be recoverable for infringements committed within 6 years of the date of the claim form. Limitation is applicable in those circumstances because the Defendant has turned its back on entering into the relationship of willing licensor and willing licensee.


533. By contrast, if the Defendant does wish to invoke the Claimant’s undertaking to ETSI, then, howsoever the action is characterised, the relationship invoked (that of willing licensor and willing licensee) is central. It is that relationship, in my view, which takes this type of action outside the normal realms of actions in tort or contract where limitation applies. It is that relationship which is inconsistent with one party in these circumstances being able to rely on limitation defences.

For now, however, the court remains undecided on whether interest should be awarded on the past royalties, and will hear submissions on this point (para. 552).

6.  Another important part of the decision is the court’s conclusion, consistent with previous case law, is that, under the ETSI IPR, the licensor’s FRAND commitment is irrevocable--and thus even an unwilling licensee must, like Don Giovanni, be provided the opportunity to repent à l’ultimo momento (my analogy, not the court’s):

536. . . . [T]he reasons why a SEP owner and an implementer have failed to agree on FRAND terms may be numerous and varied. The blame may attach wholly to one side or the other, but in most cases, the blame is likely to be shared. However, whatever the situation as regards blame, at least under the procedure adopted in these UK FRAND cases, the parties will reach a point where the Court decides that the implementer infringes a valid SEP.


537. At that point (at the very latest), the implementer (it almost invariably is the implementer) must be put to his election: either cease infringement of the SEP found to be valid, infringed and essential to the standard in question (and pay damages for the prior infringements) or take a licence.


538. At that point, a FRAND licence must be available for acceptance by a willing licensee – that must follow from the irrevocable undertaking given by the SEP owner. It seems to me that however unwilling an implementer may have been in the past, at that moment, with a suitable change of heart, the implementer retains the ability to change its position to that of willing licensee. At the same time, it seems to me to be axiomatic that the willingness must be unconditional, and the grant of a FRAND injunction will ensure that is the case at least so far as the immediate present and future are concerned.

This strikes me as a major difference from the German approach, under which an unwilling licensee is likely to find itself enjoined, and the courts have little if any interest in calculating a FRAND royalty themselves.  It would seem to follow, in my view, that the German approach is likely to induce implementers to give in to supra-FRAND offers—which may mean, as some of the commentators have already pointed out, that licensors are now going to have further incentives to flock to Germany and the UPC and away from England.  

7. The final calculation, based on the judge’s weighting of the relevant comparables, “different rates [for] different periods of time" (para. 803), and adjustments for developing and emerging markets, results in a single blended rate for all the relevant SEPS:

807. Accordingly, the most important adjustment is to reflect the split between Developed and Emerging Markets. Overall, I decline to make any separate adjustment to reflect patent coverage. I will apply a single adjustment ratio of 0.728 to reflect all the differences between LG and Lenovo, which brings Mr Meyer’s LG 2017 rate of $0.24 down to $0.175.


808. For the period 2019-2023, the contrasting sets of figures for LG as against Lenovo show a similar pattern:


i) LG’s sales mix over that period (3G: 0.3%, 4G: 94.1%, 5G: 5.6%) was very close to Lenovo’s (3G: 1.2%; 4G: 96.0%; 5G: 2.7%). Mr Meyer derived an adjustment ratio which was close to 1, 0.986.


ii) LG’s split between Developed and Emerging markets was closer to the mirror image of Lenovo’s – approximately 2/3:1/3, whereas Lenovo was almost 1/3: 2/3. On the data, Mr Meyer derived an adjustment ratio of 0.798.


809. Some caution is required over this period because LG’s licence term ended at the end of 2020, so the LG data only cover 2 out of the 5 years. However, LG is still clearly the best comparable not least because the sales mixes remain almost identical, whereas the sales mixes under Huawei 2020 and Xiaomi 2021 are very different. Furthermore, the way that the experts unpacked LG 2017 effectively eliminated any influence of LG exiting the market in July 2021. For the period 2019-2023, I apply the same rate as before, $0.175.


810. In respect of the earlier period of 2007-2011, LG 2017 only provides sales data for 2011. ZTE 2019 is the only LS PLA which provides sales data going back to 2007. In the absence of what I regard as reliable data, I will apply the same rate as for 2012-2018.


811. Although I have found the three periods useful for the purposes of comparison and analysis, and potentially they could have given rise to different rates, I have decided to apply the same rate across all three periods. I am conscious that I have, in the end, relied on a single comparable, but, for the reasons I have explained, I do not regard any of the other Lenovo 7 as assisting. Each one was more different to Lenovo’s situation than LG 2017, in some cases, significantly so.


812. Finally, I remind myself of the task in hand. It is to determine what a willing licensor and a willing licensee would agree by way of FRAND terms, in this context a lump sum, to cover the period from 2007 to the end of 2023. In this context, InterDigital’s start date of 1st January 2018 is irrelevant.


813. With my decisions on the points of principle in mind, I consider the willing licensor and willing licensee would agree a single per unit rate which would reflect all the considerations I have discussed above. I conclude that rate is $0.175 per cellular unit.


814. The calculation model provided to me by the experts included sales figures for Lenovo going back to 2007. The $0.175 rate yields a lump sum payment of $138.7m.

8. Mr. Justice Mellor found unpersuasive InterDigital’s top-down analysis, which the firm had offered as a “check,” but which would have resulted in an award substantially higher than $0.175 per unit (paras. 815-85).  Without going into all the details, I’ll just this summary in para. 881:  “I    remain unpersuaded by any part of InterDigital’s top-down analysis. The principal reason is because of InterDigital’s overall contention that the top-down analysis supported the rates in their 5G Extended Offer. Since the comparables analysis does not provide any support for those rates, and I have found those rates to be inflated and discriminatory, the results of the comparables analysis represent a solid reason for dismissing InterDigital’s top-down cross-check as pleaded.

9. This is not a major part of the judgment, but in para. 168(i) the court expresses agreement with Mr. Justice Birss (as he then was) in Unwired Planet, that “I do not see how one can eliminate or distinguish the value of an invention being incorporated into a standard from the invention itself.”  Norman Siebrasse and I have expressed agreement with this point too (see here), though it runs contrary to some statements made by the Federal Circuit in Ericsson v. D-Link.