The IP of Things

BACKGROUND

This is an article that I have been intending to write for quite some time already. And now I finally got the time for it. I hope that it is a worthwhile read. If for no other reason, then at least because you will almost certainly be surprised about the conclusions at the end. At least it is based on extensive licensing experience. Really hundreds and hundreds of licensing deals in highly relevant industries, both cellular/consumer electronics and computing industries. As well as even quite many IoT licensing deals more recently.

Essentially, my intention is to lay out the IP (i.e. patent) landscape for the emerging field of IoT (Internet of Things). But I will do it in the manner that works, i.e. from the general to the specific. From macro to micro. And not the other way around. Not from micro to macro. Many in the patent profession are tempted to take the opposite approach, i.e. from the specific to the general. In other words, by starting out from looking at recent patent filings, patent portfolio sizes, nitty gritty patent data or even individual patents.

The problem, of course, with such approach is that, in addition to being immensely time consuming, it will also not work to yield valid overall observations. At best, it wastes precious time and resources. And, at worst, it creates a distorted, and frankly often a truly flawed, overall picture. Many such analyses are available online. But none that I have seen, e.g., apply the ever important licensing lens. Patent landscapes cannot be understood based on patents alone and, frankly, the licensing practices and histories tend have an overriding importance. One simple example: a strong patent portfolio already broadly licensed to the industry may be essentially devoid of any remaining value. And it does not even really matter how good, or bad, such patents are. Just does not matter. Quite a paradox.

IOT REALITY CHECK

The phenomenon currently called IoT has had different names in the past. One of them should be very familiar to those having done their time in the cellular industry: machine-to-machine. Or M2M for short. M2M was something that was supposed to happen at scale for .... about 20 years. Only it never did. So what was the problem? The problem was that, while the client side "machine" was up to the task, the server side "machine" was not. Essentially, the servers receiving and processing the data were hosted on-prem and quite limited in capacity and notoriously unrealiable.

I actually had a M2M generation security camera at our house. The client side "machine" of it, the camera, took nice pictures with its still image sensor whenever it sensed movement with its motion sensor. It really was the stereotypical client side IoT device already during the M2M era. Only that those pictures were then sent to a proprietary on-prem server "machine" hosted by a small importer of security cameras. I suppose in their private basement somewhere. So, the pics got stored and processed there and were sent forward. In theory at least. Pics from an outdoors security camera must be (also) stored elsewhere, as a bad guy may also steal the camera. Makes sense, right? Well, of course the server was not only slow, but failed a lot. You cannot really have a security camera with a downtime of like 25% or more. That′s not security grade.

So, I got it replaced by more of an IoT generation one. Essentially the same device on the client side, but the server side is an IoT era one. The pics the new camera takes are beamed to the cloud with essentially endless storage capacity, abundant processing capacity and truly robust reliability. Some virtual server fails somewhere and its workloads are offloaded elsewhere in nanoseconds. Not to mention that all the data stored in the failed server are replicated elsewhere, many times over. And, what′s best, none of that is even visible to me. Other than as the results. Downtime went down to practically 0%. So, M2M generation solution never really worked in that case, while IoT generation solution works almost perfectly. The change was on the server side. Starts to make sense, right?

CONNECTIVITY FOR IOT

IoT essentially comprises of client side devices with different sensors (speed, acceleration, motion, proximity, location, distance, weight, humidity, altitude, depth, compass, temperature, pressure, still image, video image, infrared, audio, noise level, blood sugar, heart rate, number of steps, etc.) that are connected to a server side backend somewhere.

Notice that the connectivity required for the IoT market to take off has been here for more than a decade. Most client side IoT devices send simple ASCII format number and letter data, say temperature and humidity perhaps with associated metadata (such as time stamps). These can easily be beamed off to the cloud with basic 2.5G GPRS or 2.75G EDGE connectivity. 3G, in turn, generally suffices as the cellular connectivity for sending more data, such as still and video images. 4G cellular connectivity will be almost an overkill for most IoT use scenarios, but of course helps with cost and capacity. Those matter also. Future technologies with very low latency will mainly be needed for things like self driving cars and the like in the future. But are really not needed for basic IoT functionalities.

Then, shorter range connectivity technologies like BlueTooth and WLAN have been here for ages. They offer ample connectivity for IoT. Not to even mention the fixed connectivity technologies. They have been ripe for IoT for a few decades too. So, the connectivity technologies, both wireless and wireline, required for the IoT market to take off have been around for a decade or two already.

SENSORS FOR IOT

As to sensors, they are mainly decades old technology too. While there of course are some more recent technological inventions, such as measuring blood sugar levels through the skin, most sensor technologies used in the IoT devices are several decades old. That is not to say that there has not been much focus lately on improving them. By making them more reliable, more affordable, less energy consuming, more miniaturized etc. This has all been necessary to implement those decades old sensor technologies at scale for the consumer IoT market. Even then, it is clear that breakthrough innovations in sensors are generally not what is driving the sudden emergence of the IoT market. That is manifestly not the case.

CLOUD COMPUTING FOR IOT

So what then is driving the IoT finally happening, while its earlier iterations, such as M2M, never really did so at any meaningful scale? Simple, and what I alluded to above: ubiquitous and almost free cloud computing and storage. On the server side, everything has changed. Today, the other "machine" at the server end is a vast framework of datacenters that can analyze, process and store anything in a very reliable and affordable manner. This suddenly allows for the utilization of generally much older connectivity and sensor technologies at scale for IoT purposes. Make no mistake about it: IoT is happening now because of the tremendous recent improvements in cloud computing, not in connectivity or sensors.

Again, there will always be minor exceptions to every general rule. That applies here as well. While, say, heart rate monitors have been around for a long time, sensors measuring things like blood sugar through the skin have not. So, yes, there will some IoT devices based on entirely new types of sensors hitting the market soon. Still, it is the emergence of cloud computing that is causing the IoT market to finally take off at scale.

THE PATENT LICENSING LENS

Now, let us turn to the questions: 1) who will have the most IP and 2) who is best positioned to monetize such IP in this brave new IoT world? The answer may surprise you, but I will not share it yet. The conclusion will be further down.

Anyhow, as people truly versed in patent monetization know, it is pointless to look at individual patents before understanding the broader R&D landscape (i.e. who even can have much meaningful IP in a given space?) and the prevailing licensing landscape (i.e. what are the licensing practices? who has already licensed out everything they have? And, is there a legacy of licensing or even anyone to talk to regarding licensing in a given industry?). Then you need to further understand, who is both willing and able to license and, more importantly, to litigate? And who has the licensing and litigation history to back up the same?

Put differently, understanding the licensing and overall R&D history and context reigns supreme for understanding the IP landscape for any industry, IoT included. Patents, paradoxically, are less informative for analyzing the patent landscape. Not to say they do not matter at all, but whether and to what extent they matter depends on other things. So, a broader analysis is needed.

WHO CAN HAVE LICENSABLE PATENTS FOR THE IOT SPACE?

First, it is crucial to understand that today′s patent landscape reflects the R&D that was generally done 10-15 years ago. More recent R&D rarely matters to any major degree. This because it generally takes 5-10 years for patents to be issued after the application and then some additional time for the corresponding technologies to be implemented in a scalable, affordable, reliable and energy efficient manner. My best quick and dirty proxy for patent quality and relevancy remains the patent having at least 15 years of age. It has worked remarkably well over the years for both litigation and licensing purposes.

In this regard, I personally find it amusing, or frankly downright hilarious at times, that the press is in the habit of writing articles about freshly published recent patent applications, particularly from the likes of Apple. Only the passage of time, normally at least 10-15 years, will later tell whether those patent applications have value or not. Patenting and patent monetization is a very long term business. The R&D of today is generally irrelevant for current licensing.

CONNECTIVITY PATENTS FOR IOT

This area is simple. 10-15 years ago the R&D in the cellular space was very concentrated. Really mainly only Nokia, Ericsson and Qualcomm were doing large scale R&D. Of course there were others still around, such as Motorola and Siemens, but they had already fallen very far behind. Then there is the licensing lens that needs to be applied as well. Unless you know about the so called "Qualcomm passthrough rights", please take time to familiarize yourself with the concept. Their net impact is that almost all companies have practically already licensed (at least) their cellular connectivity patents for the lion's share of the cellular device market using Qualcomm chipsets. Put differently, it does not much matter, whether most companies have patents for cellular connectivity or not. This still seems to surprise many out there. You may want to read up on the topic.

NOKIA CORPORATION

Nokia has been twice confirmed in public litigation not to have subjected its patents to these Qualcomm passthrough rights. This is rare and very important. It also possesses the strongest cellular connectivity patent portfolio according to most high quality third party studies. So, Nokia clearly stands out as a company with a valuable cellular patent portfolio for the IoT space, both because it clearly did the largest amount of R&D in that space 10-15 years ago, but even more importantly as its patents are not covered by the Qualcomm passthrough rights. Nokia also has a large and capable patent licensing and litigation operation and has recently announced that it has exceeded the important 1 billion dollar annual licensing income threshold (go Eeva, Susanna, Robert, Rob, Jan, Aura, Sonja, Teemu, Mika, Patrick and others!). So the history, skills and reputation are there and can be leveraged into IoT as well.

QUALCOMM INC.

The same of course applies to Qualcomm itself. And Qualcomm did quite a bit of R&D in the cellular space 10-15 years ago. So, it clearly ranks quite high in terms of having patents to be licensed for the IoT space. Qualcomm has an unparalled licensing track record, which more than makes up for its more modest litigation track record (although it did sue Meizu recently). In terms of annual licensing revenue, Qualcomm is in a class of its own with more than 8 billion USD in annual patent income, if I recall things right. Some of you, I know, will be quick to point out to their recent regulatory challenges in this context. Still, the reputation, the skills and the track record are there. Eric, Fabian and their troops are very seasoned veterans of patent licensing, to make a massive understatement.

TELEFONAKTIEBOLAGET L.M. ERICSSON

Then there′s Ericsson as the third major connectivity IP owner. But that story is a little more complicated and I cannot share it here. Interestingly, Ericsson's strategy to tackle the emerging IoT patent licensing market comprises of setting up a new NPE/patent pool for the purpose, led by its well known former head of IP Kasim A. Ericsson has an increasingly impressive litigation history and has been battling Nokia for the second place in annual licensing income over the last decade or so, until Microsoft bypassed them both a while back. So, once again, the skills, the reputation and the track record are there. But, as the launch of their new IoT licensing platform shows, they themselves must have some reservations about being able to leverage their IP at scale for the IoT space, to the extent possible given their existing licensing encumbrances. In any event, it will be very interesting to see how Kasim's new IoT licensing platform fares. He also landed Luke M. to work for him, so the pieces are starting to fall in place it seems.

THE OTHERS

As to the rest, the patents for cellular connectivity are tremendously spread out and almost entirely encumbered by the Qualcomm passthrough rights. This largely moots the discussion. As an example, some Chinese companies may, or may not, have meaningful cellular connectivity patents portfolios about 10 years from now. Their R&D has started only very recently (two decades too late for 2G, one decade too late for 3G and generally too late for 4G also) and their patenting has been massively tax subsized by the Chinese government, which supports very low quality patent filings. And then there are the Qualcomm passthrough rights. Only time will tell, if they ever become a major patent holder for the cellular connectivity space. While they are trying, I would not necessarily bet my own money on that, knowing what I know about them and their practices. But the government over there has a "5 year plan" to try to make it happen.

I will not comment in this context on things like fixed or WLAN connectivity, as those markets lack meaningful patent monetization history due to things like extensive cross licensing and the operation, or at least alleged operation, of the underlying SSO IPR policies.

SENSOR PATENTS FOR IOT

As mentioned, many of the sensor technologies are several decades old meaning that the corresponding core patents have long since expired. The ones in force are held by a very diverse and large group of companies ranging from large consumer electronics companies to phone manufacturers to semiconductor companies to smaller startups.

As an example, a relatively small, but old and well established, Finnish company called Suunto is a pioneer in heart rate monitors and holds some early IP in that sensor space. All the other types of sensor spaces tend to have their own Suuntos as well, i.e. pioneer companies having done the R&D in that space the required 10-15 years ago. Expect tremendous patent acquisition interest in their patent portfolios, if broadly unlicensed.

Then there are the semiconductor companies. Many of them indeed did R&D in the sensor space 10-15 years ago, but their patent portfolios tend to be either broadly cross licensed already or then subject to complicated detente type standoffs really based on balance of terror between various semiconductor companies. For them, sensor patents are just a tiny piece of a far bigger game, or a far bigger mess really.

And, finally, the consumer electronics and phone giants having been active in that space 10-15 years ago, do have relevant and valuable sensor patent portfolios as well, but tend to monetize them as parts of other more core patent offerings, such as connectivity or audiovisual codec patents.

The licensing and litigations skills and track records vary wildly across this huge group of companies. As does the licensing track record in the form of annual patent licensing income. But, for the most part, not a lot of patent monetization has happened or is happening in the sensor space. That probably will not change.

So, to conclude, the sensor patents relevant for the IoT market are tremendously spread out depending on the type of the sensor in question and, as regards most types of sensors, the most important patents have long since expired. But, again, new sensor types will continue to emerge and there will continue to be patents for improved versions of the old sensor types.

CLOUD COMPUTING PATENTS FOR IOT

MICROSOFT

Well, this is quite simple. Who was doing large scale R&D in the computing space 10-15 years ago? Microsoft. Simple as that. Their market share in computing was like 95-98% back then. They pioneered the whole client-server type distributed computing paradigm, which still underlies the emerging cloud computing market. Understand also that Microsoft stands way above everyone else in terms of having computing patents to license for the client side of the IoT market, as evidenced by their tremendous success in Android/Chrome/Linux licensing. Microsoft also has a strong patent licensing group with quite impressive annual licensing incomes recently (go Nick, Yoshi, Geoff, Rob and Andy !!!). There is some patent litigation track record as well, most recently against Kyocera.

THE OTHERS

That is not to say that there cannot be other companies with meaningful cloud computing related patents too. Many folks in this context would like to single out IBM, but when you then apply the licensing lens and understand how broadly their portfolio has been cross licensed and divested out already as well as how broadly it is subject to complex open source and like encumbrances, it becomes quite obvious that IBM has little unlicensed IP left. If you need more evidence, just look at the huge drop in their annualized IP income. Really just a shadow of its former glory. This supports the view that most of the value in their IP portfolio has been extracted long ago. Most industry insiders would agree with that.

I could run through a similar logic for portfolios such as the former Sun Microsystems (currently Oracle) patent portfolio, but it really is not necessary. Microsoft stands so very clearly from the crowd. So much so, that they will essentially inform the prevailing patent licensing paradigm for the cloud space, based on their own aspirations and strategies. Also note that companies like a certain cloud computing giant, who were not doing meaningful amounts of computing R&D back then, cannot have very valuable cloud computing IP portfolios - at least yet. It remains to be seen in about 10 years from now, what kind of a cloud computing portfolio they will have. Same generally applies to the emerging Silicon Valley cloud players.

THE EVER IMPORTANT VERTICAL SPECIFIC PATENTS

Finally, let us not forget that, while IoT devices are cloud connected devices loaded with sensors, those devices normally serve a larger and more traditional purpose. A purpose specific to the vertical industry in question. This is to say that a car, even if a connected IoT car, is still first and foremost a car. With massive patent portfolios to be cleared for anyone entering the connected car industry. The patents relevant to engines (whether combustion, electric or both), batteries, airbags, chassis, drive train, gearboxes, audio/entertainment etc. will remain hugely important. And, what is more, there is very complex history and associated practices for clearing all that IP, not just between the car brands, but also by their so called tier one suppliers such as Delphi and Autoliv, who have traditionally played very important roles. Really a unique industry from the licensing perspective, as anyone having dealt with the car industry knows.

Similarly, a connected refrigerator is still first and foremost a refrigerator. And if it is not working well as a refrigerator and with the latest bells and whistles applicable to that vertical (high tech water filters, ice making, energy efficiency etc.), then the customers will not be happy. Even if it has nice WLAN based cloud connectivity and dozens of sensors allowing for seamless IoT interaction with the device. Similarly, a connected home is still mainly a house and a connected drone/helicopter is still mainly a flying drone/helicopter. If it does not fly well and for a long time in a well controllable manner, then it really fails irrespective of possible IoT connectivity. I could go on, but I trust you get the point. IoT, in many ways, is just one feature among many, although an important one.

Each of these verticals (cars vs. refrigerators vs. drones vs. houses vs. health care devices etc.) have their old and well established IP and licensing landscapes and practices that are not going anywhere simply due to the addition of an IoT cloud stack, some wireless/wireline connectivity and a bunch of additional sensors. So, the vertical industry relevant patents will remain very important also in an IoT context, but there are too many industries to be meaningfully addressed in one article. Still, that underlying complexity is not going anywhere - IoT or not.

CONCLUSIONS

So, a few points to restate. And the promised perhaps surprising conclusions.

The emerging IoT market is mainly driven by recent improvements and advances in cloud computing. The connectivity and sensor technologies have been ready for IoT market for at least a decade, maybe longer.

Nokia, Qualcomm and, perhaps to a lesser extent, Ericsson, will continue to be giants in connectivity patent licensing, also for the IoT space, over the foreseeable future. Pay attention to how Kasim′s new IoT licensing platform fares. That could be interesting to watch.

The sensor patent landscape will be much more complicated and dispersed. Given the age of most key sensor inventions, it is quite possible that no independent licensing paradigm will emerge for sensor patents used in IoT. Expect no big IP splashes or battles from this space.

As to computing, Microsoft clearly stands out from everyone else in the space of cloud computing patents and will, so I believe, largely inform the prevailing licensing paradigm for the cloud patent space through the courses of action it chooses, with Satya and his troops choosing the path forward.

Finally, the traditional patent and licensing landscapes and practices relevant to the vertical industry underlying the IoT device in question will remain highly significant. For example, you cannot become a connected car maker just by clearing the IP rights for the IoT stack, the connectivity and the sensors needed to make it a modern connected car. After all, it is still first and foremost a car. And must function as one, even if IoT connected. And the car R&D history goes back more than a century, with all that it entails.

So, does it surprise you? Not much will change. Mostly same old, same old. Even for the IoT space. Yes. It will be the Nokias, Microsofts, Qualcomms and Ericssons still in the limelight in terms of IP licensing, perhaps with some new entrants from some specific cloud computing or sensor angle. And then the Hondas, the Toyotas, the VWs, the GMs, the Whirlpools, the Boschs and so on, who are not going anywhere as holders of massive vertical specific patent portfolios. The more things change, the more they stay the same, or so they say.

I personally believe that it will be impossible for anyone to overturn that conclusion just by looking at patent portfolio data and individual patents, as many in the patent profession are traditionally tempted to do. But you are, of course, welcome to try, if you have several years of time to use for such analysis.

Even then, you would need to apply licensing and business understanding on top of those results. Meaning that the outcome of such micro to macro analysis would have to be adjusted to align with the macro level realities. And in the patent space, that reality is not only informed by patents themselves, but even more by the licensing and litigation landscape that has been created on top of such patents over the past decades.

This is by no means unique to the IP market. The same is also true, e.g., with the financial markets. They are not only driven and informed by some objective analysis of the underlying assets by some genius quant person somewhere. No. Instead, it is the actors on the markets, their histories and actions (and even emotions) and the dynamics between them as well as even things like outright government intervention that shape the market. The assets themselves are just the foundation, on top of which a much bigger and more complicated show is played out. Which is really ... what makes it all so interesting.

Thank you! And stay tuned for more.

Niklas ?stman

[email protected]