Monday 15 July 2024

Artificial Intelligence and Trade Secrets -- Professor Villasenor's Insights

Professor John Villasenor at UCLA has published an interesting and helpful article on AI and trade secrets. He identifies some issues regarding the protection of AI generated trade secrets.  The Brookings Institution has published a summary of his article, here

Wednesday 3 July 2024

European Commission’s proposed top-down approach would massively reallocate SEP royalties to China

This is the second in of a pair of postings on aggregate SEP royalties and shares of those received among licensors. The first article, published yesterday, is an update on how much in royalties are paid and how these have declined, with a reminder on why these are so much lower than maximum rates headlined. This second article is on how these payments would be reallocated with the European Commission’s proposed rate setting using the top-down approach based on simplistic patent counts. I’m doing this to show the hypothetical extent of such reallocations and who the winners and losers would be. Please let it not be construed that I advocate such a disruptive and harmful intervention. I do not.

Adverse disruption

The European Commission’s scheme to regulate royalties by setting aggregate royalties and apportioning them with the top-down approach based on standard-essential patent counts would very disturbingly and harmfully effect SEP licensing. My analysis shows that apportioning current levels of aggregate royalties based on declared-essential patent counts would massively reallocate royalties received by US and European licensors to Chinese companies.

US and European companies account for 94% (i.e. $8.1 billion) of $8.6 billion in total licensing revenues reported in public annual filings by the five largest SEP licensors. Total royalties are estimated to be $16 billion including unreported amounts paid and cross-licensing value, as indicated in the Exhibit. Royalties to US and EU licensors would reduce by at least $3.8 billion when reapportioned on the basis of declared-essential patent counts with Chinese firms — including Huawei, BBK Electronics (with brands Oppo, Vivo OnePlus, Realme), Xiaomi and Lenovo — being the dominant recipients with over 40% of the total 5G patent count among Top-20 declarers alone. Royalty reallocations to Chinese companies would likely be even much larger because some of the additional $4.3 billion of value in “Other licensors’ cash royalties” and $3.1 billion in “Cross-licensing value”. This represents nearly half of total royalty value. I have not been able to quantify individual amounts of that being obtained by particular companies or nations. However, some of that value is being captured by US and European companies, such as Philips with a relatively low 0.12% share of 5G SEPs, that would also have their royalties reduced with reallocation by patent count.

Exhibit: Global royalty payment reallocation based on top-down apportionment by 5G patent count share is mostly China's gain
All dollar figures in billions per annum. Patent count shares are of declared patent families with a 5G grant in US, Europe, China, Korea or Japan declared to ETSI as of 18 October 2023, as published in "5GEssential Patents: Key Findings for facilitating 5G licensing negotiations globally – Q3 2023; Lite version”, PA Consulting.

Most SEP licensing revenues are generated by four large public companies — Ericsson, InterDigital, Nokia and Qualcomm that disclose these every year, and by Huawei that has disclosed these for 2022. My previous analysis on those since 2013, followed by that in other published research and in light of overall licensing market and foreign exchange rate trends since then, leads me to believe that total cash royalties paid would be no more than $12.9 billion in 2023. That is one third more than approximately $8.6 billion in total these five currently generate annually. Grossing figures up also to include non-cash royalties notionally considered in cross-licensing SEPs owned by implementers including Huawei, Samsung and others, I estimate total SEP licensing value of $16 billion per annum.

As also explained in my previous article, my grossing up is to a total SEP licensing value figure less than that indicated in the European Commission’s Impact Assessment. According to the European Commission, “The largest share of royalty payments for SEP licenses comes from the mobile telecommunications industry, which generates an estimated patent royalty yield of EUR 14 – 18 billion per year with additional EUR 4 billion of non-monetary benefits from cross-licensing (Impact Assessment Report, p.9). However, the sources cited for these figures are based on licensing around 2015 when total royalties for the four largest licensees were 40% higher and the Euro was rather stronger against the dollar than it is now. My current figures are consistent, approximately, with those old figures after prorating for these changes.

Implications

Exactly who all the winners and losers would be, and to what extent — among various different licensors, as well as between licensors and licensees in general— with a top-down approach in SEP royalty rate setting depends on the aggregate levels set and which patent counts are used. Nevertheless, the reallocation of royalties to Chinese companies is so pronounced that these firms would be the overall winners — with any conceivable aggregate rates and patent counts. For example, my previous research indicates that essentiality checking only somewhat mitigates rather than eliminates the effects of over declaration on patent counts.

My above analysis is of global impact on licensing revenues. If this readjustment is applied only to royalties derived from European product sales, readjusted royalties would be in line with Europe’s approximately estimated 15% volume share and 25% value share of global smartphone sales.

Top-down approach royalty apportionment would massively reallocate to Chinese patent owners much of the SEP royalty income US and European licensors currently receive. Patent portfolio strength comes from technical contribution to standards, value in implementation and validity rather than simplistic patent counts — with or without essentiality checking. It would be shooting itself in the foot and a great shame for Europe to make such a harmful change. The money is vital to fund ongoing US and European R&D investments in 5G and 6G technologies.

Keith Mallinson, founder of WiseHarbor, has more than 25 years of experience in the telecommunications industry as a research analyst, consultant and testifying expert witness.

Tuesday 2 July 2024

Declining SEP royalties payments yield rates significantly below licensors’ headline figures

This is the first in a pair of articles on standard-essential patent (SEP) royalties. It updates my previous publications on this topic over the last decade to show how royalty payments have trended including how rates compare against licensors’ touted maximums. The second article shows how these royalties would be massively reallocated to Chinese companies with the top-down approach in rate-fixing regulation proposed by the European Commission.

Aggregate royalties paid to major SEP licensors Ericsson, InterDigital, Nokia and Qualcomm have declined by 28% in nominal terms since peaking in 2015 to 2023, as indicated in Exhibit 1. The aggregate “royalty yield” (i.e. total royalties paid divided by handset sales revenues) for these licensors has dropped even more steeply by 38% since 2015.

The value of total royalty payments has eroded an additional 22% in real terms after inflation over those last eight years.

Percentage royalty yields have been diminished by royalty base caps and the switch to monetary amount per unit royalty rates in some cases. While ad valorem percentage rates charged hedge for inflationary increases in phone prices, caps and fixed amounts per unit are not indexed.

The recent plunge in total royalties from 2022 to 2023 is largely due to falling smartphones sales. However, increasing quarterly smartphone sales figures in 2024 suggest there will also be a bounce in royalties this year.

Exhibit 1: Cellular SEP royalties including percentage yields have generally decreased



Royalty yield is based on all indicated royalty revenues, but on only handset sales revenues.
Apple agreed in April 2019 to make a $4.7 billion one-off payment to settle its dispute with Qualcomm, following non-payment of royalties for two years. Under a long-term agreement with Huawei in July 2020, Qualcomm received $1.8 billion covering previously unpaid licence fees.
Figures have changed slightly from versions of this chart published in previous years as I have now switched to using smartphone revenue figures from Statista. The revealed trends and my conclusions are unaffected.

Cellular SEP licensors obtain significantly lower royalties than the maximum percentage rates and monetary rates per unit publicly headlined on their web sites. That’s only to be expected because licensees insist that royalties are capped on higher-priced smartphones. Some inevitable major discounts are explicit in program rate cards. Other reductions arise from various different relationships between licensing parties — such as cross-licensing to access each other’s technologies in some cases. Average royalties received are also diminished where licensing is delayed or never agreed.

Qualcomm remains the clear leader in SEP licensing. I also show in this article that the royalty rates it obtains are much closer to its rate card figures than other licensors achieve versus their rate card figures.

Fit for purpose in rebuttal

When I published my seminal article on mobile handset aggregate royalties in 2015, my objective was to disprove — with an approximate yet conservatively high estimate — the absurd assertion from Intel and others that aggregate royalties paid to license a $400 smartphone could be as high as $120 (i.e. 30%). I coined the term royalty yield (i.e. royalties paid divided by product prices or revenues) to depict effective rates paid as distinct from licensors’ notional maximum rates before caps, other discounts and cross-licensing reductions. I concluded that the aggregate yield was no more than around 5% of cellular handset prices or revenues. Others validated my methodology and came up with similar (i.e. in my ball park versus Intel’s), but even lower figures.

My cellular handset-focused methodology was for fit for purpose because it conservatively somewhat overestimated rates paid. There are various approximations in this kind of royalty yield analysis:

·        it includes licensing fees for non-phone devices (numerator) but excludes device sales revenues for these (denominator).

·        it includes licensing fees for base station network equipment (numerator) but excludes the sales revenues for this equipment (denominator).

·        It includes some non-cellular SEP royalties such as for video codecs (numerator).

·        It includes some non-SEP royalties (numerator) — for implementation technology patents and even brand licensing royalties, such as to Nokia from handset manufacturer HMD.


SEP royalties generated from devices other than cellular handsets still account for less than 10% of total royalties. These are mostly from relatively small numbers of connected tablets, laptop PCs and cars in comparison to Statista’s figure of 1.3 billion smartphones in 2023. Avanci is a success story with its patent pooling platform for cars leading by far in Internet of Things (IoT) licensing revenue growth over the last few years. Car licensing generates royalties at an annual running rate of around half a billion dollars on automotive sales of 75 million units, with around 60% of those connected — few with 5G just yet — and approximately 80% licensed at an assumed royalty rate of $15.

Tracking overall royalty trends

While I have been able to chart the individually fluctuating and overall declining royalties generated by those named major licensors since 2013, it has not been possible to accurately track the trends for royalties garnered by most other licensors since then.  However, I was able to estimate — very approximately with conservatively large figures for the purpose of my rebuttal — the remaining share of royalties paid to other licensors back then. For example, I allocated up to $4 billion for 4G patent pool royalties at their rate card rates (by that I meant somewhere between zero and $4 billion). I expected those embryonic licensing programs to fail, but I wanted to conservatively give them the benefit of any doubts.

Those named large licensors have continued to account for the majority of royalties generated. The aggregate of their individual royalty yields each year has remained as broadly representative of the entire licensing marketplace as it was back then. Individual licensors’ royalties are more affected by their specifics: for example, InterDigital significantly licenses video codec SEPs.

Grossing up


In addition to the $8.6 billion royalties currently reported in annual audited financial statements by the five major licensors I have named, I estimate total cash royalties paid now are around half as much again at $12.9 billion (i.e. $4.3 billion more). I also include an additional non-cash value of $3.1 billion in cross-licensing. That all totals $16 billion in licensing value. For reasons I explain in my 2015 article and in another article in 2024, I never include cross-licensing value in my royalty yield figures which are cash based. However, I introduce a value figure for this here because it should and would likely be included in any royalty prospective reallocations using the top-down approach — that I did not anticipate in 2015 — but as I will focus on in my next article.

My grossing up is less than that proposed in the European Commission’s Impact Assessment. According to the European Commission, “The largest share of royalty payments for SEP licenses comes from the mobile telecommunications industry, which generates an estimated patent royalty yield of EUR 14 – 18 billion per year with additional EUR 4 billion of non-monetary benefits from cross-licensing (Impact Assessment Report, p.9). However, the sources cited for these figures (who had extensively cited my seminal article) are based on licensing around 2015 when the Euro was rather stronger against the dollar than it is now and before total royalties for the four largest licensees reduced 28% to 2023. My current figures here are consistent, approximately, with those old figures after prorating them for these declines.

Percentages and monetary amounts per unit, maximum, headline and program rates

We can view royalty rates including yields as percentages or as monetary amounts, but the lens chosen can make a huge difference to how costly charges are perceived to be.

When 2G, 3G and 4G licensing rates were first offered or publicly disclosed, they were almost invariably stated as ad valorem percentages of handset selling prices. This was even though some licensing deals were for lump sum payments rather than running royalties and some licenses ended up including caps and floors to the monetary amounts paid. Percentage royalty yields, therefore, could be much lower than the headline percentage rates.

However, in more recent years and since the disclosure of royalties sought by licensors for 5G, in some cases royalty rates are stated to be monetary amounts per unit. Nokia’s rate is Euro 3.00 per handset, and Ericsson’s handset rate is $5.00, or as low as $2.50 in “exceptional circumstances.”  With smartphones increasingly including functionality, manufacturing cost and value that is purportedly not dependent on cellular SEPs, it is commonly argued that this is a more appropriate than ad valorem percentage royalty rates. While OEMs such as Apple with relatively high iPhone selling prices averaging at around $1,000 are likely to agree, OEMs such as Transsion targeting developing nations with many of its smartphones selling for less than $100 tend to oppose paying the same amount per phone as OEMs selling smartphones at much higher prices. InterDigital and Qualcomm still headline rates as percentages, but they both apply royalty base caps that effectively convert those percentages into monetary amounts per unit for higher-priced devices such as premium smartphones.

There is significant ambiguity and scope for confusion as commentators switch from depicting royalties as percentage rates to monetary amounts per unit, and back again. The example of InterDigital’s rate card, as published on its web site, illustrates this. This indicates a 4G handset royalty rate of 0.5%. However, with an royalty base cap of $200 the royalty cannot exceed $1.00 per unit. If the handset selling price is $1,000 then the percentage royalty paid (i.e. the yield) will only be $1.00/$1,000 = 0.1%. That qualifier is very transparent; it can make a huge difference to how its rates are perceived. While 0.5% is the maximum, headline rate, that 0.1% rate is also clearly without any deception, disguise or confidentially customized discounting.

However, royalty rate terminology is poorly defined and unstandardised. For example, are “program rates” the same thing as maximum rates, or can that 0.1% yield figure I calculated also legitimately be called a program rate?

Exhibit 2 shows how the yearly average royalty yields of the each major SEP licensors have fluctuated. Nokia’s yields sharply increased for a few years after it sold its handset business to Microsoft, concurrently captured a 10-year licensing agreement with the firm and with relief from needing to cross-license any handset sales with other licensors. A decline in InterDigital’s yields has been reversed as new leadership there has signed up additional licensees in the last couple of years.

Exhibit 2: SEP royalty percentage rate yield trends for major licensors in the 4G licensing era

 

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

Average

Ericsson

0.57%

0.44%

0.47%

0.32%

0.25%

0.22%

0.25%

0.29%

0.21%

0.25%

0.25%

0.31%

InterDigital

0.09%

0.12%

0.12%

0.18%

0.14%

0.07%

0.08%

0.09%

0.09%

0.11%

0.13%

0.11%

Nokia

0.28%

0.28%

0.33%

0.32%

0.48%

0.43%

0.41%

0.42%

0.40%

0.40%

0.29%

0.37%

Qualcomm

2.77%

2.42%

2.21%

2.09%

2.00%

1.94%

1.76%

1.47%

1.41%

1.52%

1.29%

1.85%

Total

3.72%

3.26%

3.14%

2.90%

2.86%

2.67%

2.50%

2.27%

2.11%

2.27%

1.96%

2.64%


Includes all licensing income including royalties for all standards divided by total handset sales revenues.

Many factors affect the difference between the maximum royalty rates and royalty yields (stated either as percentages or as monetary amounts per unit). These include: royalty base caps, cross-licensing, volume discounts (e.g. in up-front lump sum royalty agreements), and unlicensed handset sales. I’ve also explained these mechanisms in detail in my other publications. As indicated above, this article focuses on yield levels, trends for these since 2013 and the ratios of these versus licensors’ publicly disclosed maximum rates.

Qualcomm continues to obtain royalties that are much closer to stated maximum rates than for other licensors. Exhibit 3 shows how percentage yields compare to major licensors’ touted maximum rates over that 11-year period in which 4G licensing has predominated. Other firms that disclosed 4G licensing terms by 2010 were acquired, sold off by their parents or did not pursue licensing programs.

Exhibit 3: SEP percentage royalty rate yields for major licensors in comparison to their maximum rates in the 4G licensing era

 

Royalty yield*

Maximum rate

Ratio

Rate disclosed for and by when

 

Ericsson

0.31%

1.50%

4.9

Single mode 4G, 2010

 

InterDigital

0.11%

0.50%

4.4

4G, 2020

 

Nokia^

0.37%

1.50%

4.0

Single mode 4G, 2010

 

Qualcomm+

1.85%

3.25%

1.8

Single mode 4G, 2010

 

Huawei

0.13%

1.50%

11.2

Single mode 4G, 2010

 

Total

2.78%

 

* Includes all royalties for all standards divided by total handset sales revenues 2013 to 2023. Huawei for 2022 only.
^ Disregards headline rates of 2% for Alcatel Lucent and 0.8% for Nokia Siemens Networks, as also merged into Nokia.
+ Changed in 2017 with disclosure of 3.25% 5G multimode rate.

Where licensors are also heavily exposed as implementers, yields are significantly depressed due to cross-licensing with charges being significantly netted-off. Ericsson and Nokia substantially reduced their need for cross-licensing by exiting the handset business in 2012 and 2014, respectively. SEP owners Huawei and Samsung remain OEMs significantly exposed as licensees.

Exhibit 4 shows how monetary amount per unit yields compare to maximum rates disclosed over the last few years in which 5G has been introduced. According to Statista figures, 5G smartphone sales increased to 49% of all units sold in 2023.

Exhibit 4: SEP dollar per unit royalty yields for major licensors in comparison to their maximum rates since 5G was introduced

 

Royalty yield*

Rate per unit

Ratio

Rate disclosed for and by when

 

Ericsson^

$0.75

$5.00

6.7

Multimode 5G NR Rel 15,  2017

 

InterDigital~

$0.32

$1.20

3.7

5G, 2020

 

Nokia+

$1.13

$3.21

2.9

5G SEP portfolio, 2018

 

Qualcomm~

$4.27

$13.00

3.0

Multimode 5G, 2017

 

Huawei

$0.40

$2.50

6.2

5G, 2022

 

Total

$6.87

 

* Includes all royalties for all standards divided by total handset sales revenues 2020 to 2023. Huawei for 2022 only.
^ Ericsson indicates its $5.00 rate might be reduced to as low as $2.50 in “exceptional circumstances.”
~ My dollar equivalent assumes a $200 selling price with that royalty base cap for InterDigital and a $400 selling price or cap for Qualcomm.
+ Assuming an exchange rate of Euro 1.00 = $1.07.
Upside potential

Royalty yields will always remain much lower than the maximum royalty rates SEP licensors seek to charge before explicit caps and various other discounts, with implementers such as Huawei cross-licensing their handsets for use of others’ SEPs, and as some implementers hold out from paying royalties.  However, the different ratios of these two figures among licensors can provide an indicator for discussions about potential royalty yield growth or preservation and of how effective licensors are in pursuing that.

 

Keith Mallinson, founder of WiseHarbor, has more than 25 years of experience in the telecommunications industry as a research analyst, consultant and testifying expert witness.