Business — Empirical View Of SEP Royalty Rates

An Empirical View Of SEP Royalty Rates:

The Brief History Of Trying To Separate Fact
From FUD (Fear, Uncertainty, And Doubt)
By Eric Stasik*

T
he discussion of SEP royalty rates, and in particu- the classic economic
lar the costs of licensing cellular SEPs, has always theory of Cournot ■ Eric Stasik,
been plagued by hyperbole. In 1998, an obscure Complements, the Director,
lobbying group called the International Telecommu- authors also created Avvika AB,
nications Standards User Group (ITSUG) produced a a new villain, the
Stockholm, Sweden
report alleging “when GSM mobile handsets first ap- patent holder who
peared on the market cumulative royalties amounted would “hold up” the E-mail: eric.stasik@
avvika.com
to as much as 35 percent to 40 percent of ex-works industry by using
selling price.”1 It was known at the time by many work- wildcat litigation and
ing in the industry2 (and subsequently proved by empir- the threat of injunctions over standard essential pat-
ical evidence) that this was never close to being true. ents (SEPs).4
It was easy to see that this was a misleading narrative In 2010, a survey of Royalty Rates for LTE (4G) SEPs
designed to create fear, uncertainty, and doubt (FUD) was published in les Nouvelles.5 In it, a table of an-
for the purpose of motivating regulatory intervention– nounced royalty rates from a number of leading SEP
what the cool kids today would call part of a campaign holders was presented. Adding up these announced
of “disinformation.” royalty rates (which represented only a few of the pat-
In 2006, one year before the launch of Apple’s revo- ent holders who at that date had made declarations to
lutionary iPhone, Mark Lemley and Carl Shapiro pub- ETSI of patents essential to 4G) indicated a total roy-
lished an influential article (funded by Apple Comput- alty stack of 14.8 percent of the net sales price of an
er, Cisco Systems, Intel, Micron, Microsoft, and SAS) end-user device. The article included a caution that the
called “Patent Holdup and Royalty Stacking.”3 Ob-
serving that “literally thousands of patents have been 4. Heiden and Petit explained patent holdup in more sanguine
identified as essential to the proposed new standards and slightly different terms in 2017: ”Across the world, a prob-
for 3G cellular telephone system,” the authors com- lem known as “patent holdup” has become a central issue of
mented that they had “seen estimates as high as 30 discussion in academic and policy circles. Patent holdup is said
to occur when a patent owner makes licensing or cross-licensing
percent of the total price of each phone...based on demands that are more onerous than those anticipated by tech-
summing royalty demands before any cross-licensing nology implementers when they decided to enter the industry.
negotiations began.” Warning of an even more dysto- Patent holdup is often considered more severe in relation to a
pian future than the “royalty stacking” predicted by category of patents that are declared essential to the implementa-
tion of an industrial standard, known as standard essential pat-
ents ((SEPs)”). SEPs are limitedly open to design around. Hence,
when technology implementers have made early sunk invest-
*Eric Stasik is an independent SEP licensing consultant located ments in a standardized technology, they are locked in with no
in Stockholm, Sweden. No compensation for this survey was re- other choice but to take a licence to practice the standard. With
ceived by the author. The views and opinions expressed in this this, patent owners are said to be able to extract ex post royal-
survey are those of the author alone and may or may not repre- ties in excess of what they could have anticipated ex ante had
sent the views and opinions of any client past or present. eric. the patented technology that they declared essential not been
[email protected]. selected for inclusion in the standard.”
1. “The GSM Standards, IPR and Licensing (An Example of the Heiden, Bowman and Petit, Nicolas, “Patent ‘Trespass’ and
Restrictive Effects of Standardization),” International Telecom- the Royalty Gap: Exploring the Nature and Impact of Patent
munications Standards User Group, December 1998. Holdout” (June 6, 2017). 34 Santa Clara High Tech. L.J. 179
2. The author began working with patents and licensing in the (2018), Santa Clara Computer and High Technology Law Jour-
handset division of Ericsson in 1992. In 1998, he was manager nal, Vol. 34, No. 2, 2018, Available at SSRN: https://ssrn.com/
of IPR Activities for the GSM Systems Division of Ericsson Radio abstract=2981577.
Systems AB. 5. Stasik, Eric, “Royalty Rates and Licensing Strategies for Es-
3. Available at https://law.utexas.edu/conferences/ip/Lem- sential Patents on LTE (4G) Telecommunication Standards,” les
leyPaper.pdf. Nouvelles, September 2010, pp. 115-116.

les Nouvelles 7 March 2025

Business — Empirical View Of SEP Royalty Rates

“announced royalty rate may be significantly different percent” cited in the original paper by ITSUG.10
than the actual royalty rate resulting from a bilateral ne- In 2018, Alexander Galetovic, Stephen Haber, and
gotiation.” Evidence of this was provided, i.e., reference Lew Zaretzki, following the pioneering approach by
to a press release where Nokia “confirmed that until Mallinson, published a similar empirical study revealing
2007 it had paid less than 3 percent aggregate license that SEP royalties amounted to approximately 3.3 per-
fees on WCDMA [3G] handset sales under all its patent cent of the average selling wholesale price of a mobile
license agreements.”6 phone in 2016.11 It began to appear as though these
In 2014, Ann Armstrong, Joseph Mueller, and Tim empirical studies had finally put the 30 percent boogey-
man to rest, but in the world of SEPs, feelings don’t care
Syrett published an estimate of the smartphone roy-
about your facts.
alty stack relying, in part, on the announced royalty
rates that appeared in the survey.7 The authors con- In 2019, David L. Cohen commented on the effec-
cluded (ignoring the caution) that “setting aside off- tiveness of these disinformation campaigns:
sets such as ‘payments’ made in the form of cross-li- “In a great irony, the judge that recently issued a de-
censes and patent exhaustion arising from licensed cision lambasting Qualcomm for engaging in royalty
sales by component suppliers, we estimate potential stacking among other anti-trust sins, uncovered certain
facts from Apple’s testimony under oath that under-
patent royalties in excess of $120 on a hypothetical
mine the premises behind the royalty stacking crisis.
$400 smartphone.” This was remarkably exactly the Specifically, she found that the aggregate SEP royalties
same 30 percent estimate that Lemley and Shapiro Apple pays on their 4G phones is less than $15 dol-
had seen. What the authors did was simply add up lars. This can easily be deduced from a comment on
the gross royalty figures presenting a worst case, but page 85 by Apple’s COO that [Apple] pays $7.50 in SEP
completely unrealistic, scenario. royalties to Qualcomm and his statement on page 174
Facts, however, have a stubborn habit of challeng- that Qualcomm charges Apple more than everybody
ing narratives. In 2015, Keith Mallinson from Wise- else put together. Simple mathematics then requires
Harbor was the first to publish a paper containing an that Apple’s total SEP royalty load to be less than $15.
empirical estimate of the total smartphone royalty On an $800 phone, that means the aggregate royalty
stack.8 Mallinson collected the reported licensing in- “stack” that Apple is paying, according to its own testi-
come from a large number of SEP holders and patent mony, is less than 1.88 percent (or less than $15). This
pools and compared this to known information about is a far cry from the $121 to $124 in the theoretical pa-
per [the 2014 paper by Armstrong, Mueller, and Syrett
smartphone sales and average sales prices. His con-
mentioned above] funded by Apple and Intel.”12
clusion was that actual cumulative mobile-SEP royal-
ty payments were “no more than around 5 percent In 2020, the author and David L. Cohen published
another survey in les Nouvelles for 5G royalty rates.13
of mobile handset revenues.” Mallinson, who always
Unlike the announcements for 4G, the holders of 5G
provides citations for his evidence, separately provid-
ed a much more detailed and lengthier analysis.9 Mall- 10. The fuller passage from ITSUG reads “When GSM mobile
inson’s analysis completely undermined the 30 per- handsets first appeared on the market place cumulative royalties
cent narrative originating with ITSUG and provided amounted to as much as 35% to 40% of ex-works selling price.
Although this figure is now considerable reduced (typically 10 to
proof that the industry–through commercial negotia-
15%), it is still substantially higher than the telecommunications
tion and without any regulatory intervention–arrived industry norm of about 5%.”
at an average royalty stack remarkably in line with 11. Galetovic, Haber, and Zaretzki, “An estimate of the aver-
“the telecommunications industry norm of about 5 age cumulative royalty yield in the world mobile phone industry:
Theory, measurement, and results,” Telecommunications Policy,
Vol. 42, Issue 3, April 2018, pages 263-276.
6. Nokia, April 2007 press release: http://press.nokia.
12. https://www.kidonip.com/resources/blog/the-sep-roy-
com/2007/04/12/nokia-has-paid-less-than-3-per-cent-gross-
alty-stacking-myth-apple-calls-its-own-bluff/. The page numbers
royalty-rate-for-wcdma-handsets/.
David refers to come from the U.S. District Court, N.D. Cali-
7. Armstrong, Ann and Mueller, Joseph and Syrett, Tim, “The fornia, Case No. 17-CV-00220-LHK, Findings of Fact and Con-
Smartphone Royalty Stack: Surveying Royalty Demands for the clusions of Law, May 21, 2019. https://www.ftc.gov/system/
Components Within Modern Smartphones” (May 29, 2014). files/documents/cases/qualcomm_findings_of_fact_and_con-
Available at SSRN: https://ssrn.com/abstract=2443848 or clusions_of_law.pdf.
http://dx.doi.org/10.2139/ssrn.2443848.
13. Stasik, Eric and Cohen, David L., “Royalty Rates and Li-
8. http://www.ip.finance/2015/08/cumulative-mobile-sep- censing Strategies for Essential Patents on 5g Telecommunica-
royalty-payments.html. tion Standards: What to Expect” (July 22, 2020). les Nouvelles
9. https://www.wiseharbor.com/pdfs/Mallinson%20on%20 –Journal of the Licensing Executives Society, Volume LV No.
cumulative%20mobile%20SEP%20royalties%20for%20IP%20Fi- 3, September 2020, Available at SSRN: https://ssrn.com/ab-
nance%202015Aug19.pdf. stract=3658472.

les Nouvelles 8 March 2025

Business — Empirical View Of SEP Royalty Rates

SEPs announced their headline royalty rates as a mix the largest SEP holders provides a different and no less
of ad valorem percentages and fixed price (or range of revealing empirical dataset of SEP royalties. To do this,
fixed prices) for end-user devices. This made adding up public companies, who are subject to stringent financial
a theoretical royalty stack for 5G somewhat more chal- reporting rules so the data obtained from their annual
lenging, but the general conclusion arrived at was the reports may be deemed to be accurate, complete, and
royalty rates for 5G were likely to be about the same, reliable were selected. In Table 2, the annual reported
or slightly higher than for 4G. For convenience, I re- licensing income from the annual reports of each com-
produce the announced royalty rates from that survey pany listed was collected. Ericsson reports its income
below in Table 1. in Swedish crowns and Nokia in Euro, so these figures
The reader will note that in its announcement for have been converted into USD using the exchange rates
5G, InterDigital also provided its headline royalty rates contained within each year’s annual reports. License
for 4G. In addition, subsequent to the survey, Huawei fees are most often paid in USD, so it is relevant to
announced in 2021 their headline rates for 5G, stating, make this conversion.
“for every multi-mode 5G smartphone, Huawei will pro- There are a lot of caveats in Table 2. For simplici-
vide a reasonable percentage royalty rate of the handset ty, there is an assumption that reported royalty income
selling price, and a per unit royalty cap at U.S. $2.5.”14 comes from license fees paid by licensees for smart-
It should be noted that the royalty rates shown in Table phones alone. It is likley that this is not 100 percent
1 are not entirely theoretical numbers. In the 2010 sur- accurate, especially in more recent years where license
vey, it was stated that “having made a public announce- income is coming from IoT (in particular the automo-
ment, a potential licensee might reasonably expect this tive sector) and from other commercial sectors. It is,
to be the opening offer in a negotiation. That is all that however, certain that for these four SEP holders, a large
should be assumed from these announcements.” There percentage of their reported licensing income does
is another not entirely theoretical aspect to these head- come from smartphone licenses. For Nokia, a not-in-
line rates mentioned in the 2010 survey, which will be significant portion of their reported licensing revenues
commented on later. in Table 2 comes from the 2015 agreement with Mi-
A Granular Analysis of Empirical Royalty Rates crosoft. Lew Zaretzki confirmed that Nokia received its
While Mallinson and Galetovic et al. estimated the final tranches of annual lump sum license fee payments
total royalty stack based on the reported (and estimat- from Microsoft in the period covered by Table 2. No al-
ed) licensing income across a wide range of SEP hold- lowance has been made for these. It is also true that not
ers, a more granular approach looking at just a few of all smartphones are licensed. Litigations involving some

Table 1: Summary Of Announced Licensing Terms And Conditions

For 4G And 5G Standards (reproduced from the article in footnote 7)
Announced Handset Royalty Rates
SEP Holder 4G 4G Multimode 5G Only 5G Multimode
Ericsson
1.5% 5G multimode $2.50 to $5.00
subject to subject to reciprocity.
reciprocity
Nokia 1.5% 2.0% Capped at €3 per device
subject to subject to (assumed subject to reciprocity
reciprocity reciprocity but not mentioned)

Nokia-Siemens 0.8% n/a

InterDigital 0.5% 0.6%

ASP Floor $50 ASP Floor $60
ASP Cap $200 ASP Cap $200

Qualcomm 3.25% 5.00% 2.275% 3.25%

Capped at $400

14. https://www-file.huawei.com/-/media/CORP2020/pdf/download/Huawei_IPR_White_paper_2020_en.pdf.

les Nouvelles 9 March 2025

Business — Empirical View Of SEP Royalty Rates

of these SEP holders has revealed that a number of major cross-licenses embedded into most of the revenue fig-
smartphone vendors were not licensed during the period ures. Cross-licenses have a non-cash value that can be
covered in Table 2.15 Striving for simplicity, no correction significant. The royalty yields in Table 2 for all licensors
has been made for this. It is also likely that aggregate (except InterDigital, who has no need for a cross-li-
license income comes from a mix of 2G, 3G, 4G, and cense) excluding cross-licenses would be higher.16 This
5G standards, as well as IEEE and ITU standards, as well caveat has the effect of reducing the average empirical
as from non-SEPs. No effort has been made to separate royalty yield attributed to smartphones.
these. All of these caveats have the effect of inflating the Taking all of this into consideration, the empirical
average royalty yield attributed to smartphones. royalty yields in Table 2 expressed as a royalty per
No effort has been made to estimate the value of smartphone are likely to be conservatively high.17 Even

Table 2: Annual Reported Licensing Income (from annual reports) For Four
Of The Largest 3GPP SEP Holders (converted to USD) And The Resulting
Empirical Total Royalty Stack For Those Four SEP Holders
2023 2022 2021 2020 2019 2018
Global Smartphone Shipments
1140.0 1205.5 1354.8 1281.2 1372.6 1404.9
(millions units)
Ericsson
Annual Reported Licensing Income
11 101 10 399 8 134 9 975 9 631 7 954
(million SEK)
Average Exchange Rate (SEK/USD) 10.01 10.38 8.56 9.14 9.41 8.68
Licensing Revenues (million USD) 1 109 1 002 950 1 091 1 023 916
Average Royalty per unit $0.97 $0.83 $0.70 $0.85 $0.75 $0.65
Nokia
Annual Reported Licensing Income
1 085 1 595 1 502 1 402 1 487 1 501
(million EUR)
Average Exchange Rate (USD/EUR) 1.082 1.13 1.13 1.23 1.12 1.15
Licensing Revenues (million USD) 1 174 1 806 1 701 1 720 1 670 1 719
Average Royalty per unit (USD) $1.03 $1.50 $1.26 $1.34 $1.22 $1.22
InterDigital
Annual Reported Licensing Income
549.6 457.8 425.4 359.0 318.9 307.4
(million USD)
Average Royalty per unit (USD) $0.48 $0.38 $0.31 $0.28 $0.23 $0.22
Qualcomm
Annual Reported Licensing Income
8 625 10 295 9 902 10 404 12 675 8 153
(million USD)
Average Royalty per unit (USD) $7.57 $8.54 $7.31 $8.12 $9.23 $5.80

Total Royalty Stack for Ericsson,

Nokia, InterDigital, and Qualcomm as 3.03% 3.42% 2.92% 3.25% 3.69% 2.32%
percent of $250 ASP

15. It was revealed in the 2023 UK litigation between InterDigital and Lenovo that InterDigital first reached out to license Lenovo
in 2008. InterDigital v. Lenovo, [2023] EWHC 539 (Pat), Paragraph 154.
16. When InterDigital released its Q4 and full-year results for 2024, they reported smartphone royalties separately at $597.5m up
from $467.3m in 2023. Using the same annual unit volumes as shown in Table 1, the average royalty yield per unit would have been
$0.41 per device in 2023.
17. Qualcomm’s fiscal year ends in September. Ericsson, Nokia, and InterDigital report on a calender basis. Given the monotonic
character of the numbers over time, no effort is made to correct for this temporal offset, so the year-to-year comparison between SEP
holders is not 100% apples-to-apples.

les Nouvelles 10 March 2025

 Business — Empirical View Of SEP Royalty Rates

if the empirical yields in Table 2 are conservatively high, number or a formula; FRAND is a commercial pro-
these empirical yields are much lower than one might cess—the result of a negotiation between SEP holders
expect, given the headline rates in Table 1. Ericsson’s and implementers of SEPs. As such, FRAND is tailored
floor of $2.50 per 5G multimode device (from Table 1), by the reality of commercial negotiations. When it
for example, is nearly 2 ½ times higher than Ericsson’s comes to commercial bargaining power, large imple-
average royalty yield in 2023 of $0.97 per device. menters–like a large consumer of any product–are en-
Factors That Depress Royalty Yields dowed with substantial negotiation leverage such as the
The empirical data in Table 2 (subject to all of the ability to obtain quantity discounts. This should come
caveats mentioned) reflect actual royalty yields, but do as no surprise. A builder who, for example, purchases
not explain the very large gap between those royalty 100 million bricks for a project is generally able to nego-
rates and the announced headline rates in Table 1. tiate a lower price per brick than the homeowner who
Part of the explanation lies with the caution provided only needs to buy a few dozen bricks. There is nothing
in the 2010 survey (mentioned previously) that the per se untoward or discriminatory about quantity dis-
“announced royalty rate may be significantly different counts as these reflect legitimate commercial bargain-
than the actual royalty rate resulting from a bilater- ing power. Comparing this analogy to SEPs, however,
al negotiation.” The difference for some SEP holders the builder in this scenario has already taken the 100
is so large, however, that this alone cannot explain million bricks from the brickyard and has used them to
it. What is fair to say is that the empirical royalty construct its building. The brickyard has no leverage in
stacks derived by Mallinson, Galetovic et al., (and to withholding delivery of its bricks, and this enhances the
a much less impressive extent, the individual royal- negotiation leverage of the builder. This is the situation
ty yields derived in Table 2,) reflect three realities of with SEPs that are contained in open standards, which
SEP licensing, which are: 1) the ability for licensees
can be (and often are) implemented into products be-
to cross-license, 2) commercial bargaining power, and
3) the impact of hold-out. Each of these is addressed fore license negotiations take place.
separately, beginning with cross-licenses: Hard Bargaining
Cross-Licensing It is fair and reasonable that implementers want to pay
Every one of the announced royalty rates in Table 1 as little as possible for all the components that go into
was made on the basis of “reciprocity.”18 Cross-licensing their bill of material. This is just how business works.
agreements - where each party in a bilateral negotiation Royalties for SEPs (and other patents that need to be li-
agrees to license their SEPs to each other on FRAND censed) are part of the bill of material. Within the realm
terms and conditions—have always been a major factor of legitimate commercial bargaining power is the concept
in SEP licensing. The value of a cross-license depends of hard bargaining. Implementers will take advantage of
on two things: 1) the strength of the SEP portfolio being the circumstances surrounding the negotiation to drive
licensed; and 2) the market share of the party need- a hard bargain and obtain a lower price. The ability to
ing the license. Where you have two implementers in pay large lump sums as advance payment is an example
a bilateral negotiation with similar market share and of one such circumstance. The timing of an agreement
similarly strong SEP portfolios, little or no balancing is another example. The best negotiators combine sev-
payment changes hands. The reason Nokia was able eral commercial advantages into one.
to claim (in footnote 8) that “it had paid less than 3 In an article published in IAM in December 2020 on
percent aggregate license fees on WCDMA [3G] hand-
the subject of Ericsson’s litigation against Samsung, it
set sales under all its patent license agreements” was
largely the result of Nokia’s ability to negotiate cross-li- was opined that Ericsson’s agreement with Apple was
cense agreements with other SEP holders who were, likely the reason that Samsung resisted renewing its li-
like Nokia, also implementers. cense agreement with Ericsson.20 The circumstances of
the 2015 agreement between Ericsson and Apple sug-
The Effect of Commerical Bargaining Power gested that Apple was able to take advantage of the tim-
It also has to be recognized that FRAND19 is not a ing of the agreement—in combination with an ability
to put a large advance payment on the table—to obtain
18. The licensing commitment under §6.1 of the ETSI IPR more favorable terms and conditions than might other-
Policy concludes with the statement: “The above undertaking wise be obtained under other circumstances with a dif-
may be made subject to the condition that those who seek li- ferent negotiating partner. Observing (with hindsight)
cences agree to reciprocate.” that Ericsson’s agreement with Apple occurred on 21
19. Here “FRAND” is used as an abbreviation for royalty rate. December, 2015–and noting a significant increase in
This is both imprecise and lazy—although it does make the nar- Ericsson’s licensing revenues between 2014 and 2015–
rative easier to read. The more complete phrase from §6.1 of
the ETSI IPR Policy is “fair, reasonable and non-discriminatory
(FRAND) terms and conditions” of which a royalty rate may be 20. https://www.iam-media.com/article/ericsson-samsung-
one item. stasik-new-strategy.

les Nouvelles 11 March 2025

Business — Empirical View Of SEP Royalty Rates

it was not difficult to envisage that Apple agreed to pay no bargaining power, however, it is not unreasonable
a very large up-front lump sum to Ericsson in time for to expect little difference between the announced and
some of those revenues to be included in Ericsson’s actual royalty rates.”22
2015 result.21 With Ericsson’s financial year ending on Holdout
Dec. 31, 2015–10 days after the agreement with Ap-
Empirical patent yields are also depressed by the
ple was completed—it would have been very difficult
practice of “holdout.” There is little or no available
for Ericsson’s executive team to turn down a large cash
empirical information on the financial impact of hold-
payment at a moment in time when Ericsson’s finan-
out, but evidence of holdout has been expertly docu-
cial result might have needed a boost. Revenues from
mented by Kirti Gupta and Urška Petrovčič, who iden-
patent licensing contribute significantly to EBITDA.
tified systemic holdout behavior in a comprehensive
With its strong cash position, Apple was able (maybe
analysis of court cases.23 In addition, every injunction
uniquely able) to make a 9- or 10-figure up-front cash
issued by a European court over the infringement of
payment within 10 days of signing the agreement and
SEPs since the CJEU’s 2015 Huawei-ZTE decision may
they used this ability to obtain more favorable terms
be considered as evidence of holdout by a so-called
and conditions than a similarly situated, but less liq-
“unwilling licensee.”
uid, implementer would be able to negotiate. While
there are those who might argue that these types of Other than simply refusing to pay anything at all for
discounts are discriminatory, not everyone shares this the use of the patented technology they incorporate
view. Receiving cash upfront reapportions risk and into products, there are many ways in which imple-
offers other benefits that have different values for dif- menters engaging in holdout behavior can gain leverage
ferent licensors and licensees at different times. This in negotiations. Implementers can, for example, hold
is one example of how commercial bargaining lever- out by avoiding to take a license for a decade, running
age can be used in FRAND negotiations. There are up a sizeable bill for past use, which they can then use
countless others, including, for example, guaranteed as commercial bargaining leverage when finally mak-
minimum payments, the transfer of patents, and other ing a deal. Faced with the prospect of seeking recovery
business arrangements that represent an exchange of for past use under the limitations of patent law (due to
value between parties not accounted for as licensing statutory limits on the recovery of damages that can
income. The ability of some large implementers to vary from country to country, different applications
drive a hard bargain explains, in part, the gap between of interest, the general uncertainty of litigation, etc.),
Ericsson’s headline rates and Ericsson’s empirical roy- many licensors will reluctantly accept discounts for the
alty yield in Table 2. Before moving further to the im- settlement of past use as a wholly practical matter—re-
pact of holdout, one more observation about commer- warding holdouts with what might appear to be ordi-
cial negotiating power needs to be made: not every nary commercial discounts.
implementer has it. In the 2010 survey, after caution- It must be said that there is a fine line between the
ing the reader that announced royalty rates might legitimate exercise of hard bargaining and the abusive
merely inform potential licensees about the opening practice of holdout—the two are often used in combi-
offer in a negotiation, an additional caution was pre- nation. Courts, when given a chance to examine the
sented: “For a company with no essential patents and totality of negotiation history, should be able to tell the
difference. Not all instances of holdout end up in litiga-
tion, however. The systemic practice of holdout results
21. “As part of the seven-year agreement, Apple will make in lower royalties, as manifested in the empirical data.
an initial payment to Ericsson and, thereafter, will pay on-going
royalties. The specific terms of the contract are confidential. In-
cluding positive effects from the settlement, and including the 22. It was revealed in the litigation between InterDigital and
ongoing IPR business with all other licensees, Ericsson estimates Lenovo that InterDigital made offers to Lenovo that embodied
full year 2015 IPR revenues will amount to SEK 13-14 b.” - the program rates InterDigital had published on its website since
Dec. 21, 2015. https://www.ericsson.com/en/press-releas- early 2020. InterDigital v. Lenovo, [2023] EWHC 539 (Pat),
es/2015/12/ericsson-and-apple-sign-global-patent-license- Paragraphs 20-21. InterDigital attempted to present license
agreement-settle-litigation. agreements at these headline rates with smaller implementers as
Ericsson reported licensing revenues of 9 917SEK ($1.439b) evidence of the “FRANDness” of these offers. Mr. Justice Meade
in 2014. These increased to 14 396SEK ($1.716b) in 2015. The rejected this argument as it was pled, but some think it was
total amount of the advance payment would have likely been poorly pled, and that a more fulsome explanation of commercial
higher than the difference between these numbers due to the bargaining power as provided herein might have resulted in a
way these revenues would be realized, but the substantial uptick different outcome for InterDigital.
in reported licensing revenues between 2014 and 2015—along 23. Kirti Gupta and Urška Petrovčič, “Evidence of Sys-
with the announcement of a settlement with Apple—would temic ‘Patent Holdout,’” Berkeley Technology Law Journal,
have been as favorably received by Ericsson’s executive team as Vol. 38:371, pp. 575-608. https://btlj.org/wp-content/
it would have been received by Ericsson’s investors. uploads/2023/11/0010-38-Haas-Gupta.pdf.

les Nouvelles 12 March 2025

 Business — Empirical View Of SEP Royalty Rates

Finally, there is also the fact that some SEP holders Qualcomm appears to collect anything close to
do not actively license their patents, preferring to keep its announced headline rates. This is attributed
them for defensive purposes. All of these things men- to Qualcomm’s industry-unique licensing model,
tioned above explain the difference between the boo- which enables Qualcomm and Qualcomm alone
geyman of 30 percent and the actual royalty yields seen to combat holdout effectively.
in the empirical data. The final conclusion is that the numbers in Table 2
It is difficult to draw hard conclusions from the empir- reflect the reality that FRAND is a commercial process
ical data, but a few things can be said with confidence: whose outcome depends on the circumstances sur-
1) The empirical estimates made by Mallinson (and rounding the negotiation—including legitimate hard
later by Galetovic et al.) are much closer to the bargaining and the withering effects of abusive hold-
actual royalty yields than any of the theoretical es- out. This empirical data should call into question the
timates of the smartphone royalty stack. European Commission’s proposal that conciliators from
2) The theoretical estimates are and have always re- the EUIPO will be able to make non-binding aggregate
flected the situation at the beginning of negotiations. royalty rate determinations, a number that implement-
3) Of the four SEP holders shown in Table 2, only ers will use as the starting point for negotiation. ■

les Nouvelles 13 March 2025