IP monetisation: what might the future hold?

Three fundamental technologies are emerging that will shape the future of all business in all markets, creating an economic value proposition of trillions of dollars in the next decade. These are:

• artificial intelligence (AI) (specifically machine learning);
• robotics; and
• distributed ledger technologies (blockchain).

Each of these technologies is rooted in a fourth technology – the Cloud. Such technologies are collectively referred to as next generation software infrastructure (NGSI).

NGSI is already an integral part of our daily lives, from the product choices that Amazon offers us to surveillance by the National Security Agency (NSA). It is poised to revolutionise many other aspects of our lives (eg, transportation, autonomous vehicles, accounting and aerial delivery drones). Home service robots will clean, provide security and play with your dog. Smart sensors will monitor people’s blood sugar and organ functions, robotic tutors will augment human instruction and AI will lead to new interactive ways to deliver media. Robots and smart systems will move inventory in warehouses, schedule meetings and offer financial advice. McKinsey estimates huge markets for NGSI adoption of between $3.5 and $5.8 trillion. The greatest NGSI markets are expected to be in marketing and sales ($1.4 to $2.6 trillion) and supply chain management and manufacturing ($1.2 to $2 trillion). McKinsey expects NGSI retail to reach $800 billion in market size and NGSI healthcare, banking and automotive to each reach over $300 billion in market size.

Former Kleiner Perkins partner Mary Meeker’s annual internet trends report, revealed at the Code Conference, disclosed some important themes for NGSI in 2018. The highlights were:

• massive growth in voice and natural language processing with human-level quality;
• the huge importance of China for innovation and sales (and intellectual property);
• the convergence of business models and revenue streams around data monetisation;
• the increasing speed of adoption and market growth everywhere; and
• the commoditisation of pricing as a result of competition, scale and the dominance of a handful of global firms.

What emerges from this picture is an accelerating market opportunity in NGSI that is of staggering proportions, based on a vast ocean of incredibly valuable intellectual property, primarily embodied in code, architecture and data. The intellectual property in NGSI is potentially the most valuable cluster of intellectual property ever assembled. This is because unlike prior fundamental technologies (eg, the internal combustion engine, the electric light and silicon chips) NGSI products and services are practically 100% virtual. Moreover, the commercial value of NGSI technologies lies primarily in the data being generated by such systems, as opposed to the systems and methods used to generate the data in the first place, which makes NGSI an unprecedented phenomenon from an IP perspective.

This article explores how IP monetisation professionals will use IP rights to protect investments in NGSI businesses and how IP monetisation will evolve as a result.

Three falsehoods

An understanding of how IP monetisation will evolve in NGSI industries, begins with a dismissal of three widely discussed falsehoods.

Déjà vu no more

The first of these is that IP monetisation in the NGSI fields will play out just as it did for PCs, the Internet and smartphones during the early part of this century. This patent-heavy, US-centric NPE monetisation – as chronicled in the pages of IAM between 2006 and 2016 – is now on life-support and practised profitably by a select few. It will remain so with respect to NGSIs. There are a multitude of reasons for this, including the following:

• Migration of value from the physical and operating system layers of computing platforms to the less tangible application layer, where patents have always been harder to assert.

• The continuing growth of open source ecosystems, whose ostensibly free use makes them difficult to associate with large litigation damages models and thus more difficult to profitably associate with the NPE model.
• The difficulty of steering software patent applications past 35 USC Section 101 in the United States and especially Article 52 of the European Patent Convention in Europe.
• The migration of processing functionality behind the impenetrable curtain of the Cloud, which affects the ability to determine which patents are being infringed.
• The shift in the balance of power from the West to the East, which affects litigation venues and therefore the possibility of outsized damages awards, that was relied on by the NPE model.
• The shift in value from processing apparatuses and methods to the data generated by those apparatuses and methods.

This last trend is especially important and deserves more analysis.

From intangible to even more intangible – Aristotle revisited

The history of computers is often told as a history of objects, from the abacus to the Babbage engine up through the code-breaking machines of World War II. In fact, it is better understood as a history of ideas, mainly ideas that emerged from mathematical logic, an obscure and cult-like discipline that first developed in the 19th century. Mathematical logic was pioneered by philosopher-mathematicians, most notably George Boole and Gottlob Frege, who were themselves inspired by Leibniz’s dream of a universal ‘concept language’, and the ancient logical system of Aristotle.

(Chris Dixon, “How Aristotle created the Computer”, The Atlantic, 2017).

In his master’s thesis at the Massachusetts Institute of Technology, it was Claude Shannon who first grasped the fundamental idea that the true values in Boolean logic (true and false) could be represented by a circuit with an open and closed gate. As Ben Thompson has noted in a 2017 article in The Atlantic (“The Arrival of Artificial Intelligence”, 2017) this insight was the bridge that connected the logical/intangible layer of mathematical logic with the physical layer of early computing engines. This insight also made it possible for attorneys to try to patent intangible software and eventually led to business ideas, such as OneClick Purchasing. Thus was spawned the industry of software patent licensing and much of the PC, internet and smartphone patent wars.

As we now enter the next cycle of innovation, another bridge is being built between mathematical logic, as embodied by computer programs, and the data generated by that logic. As machine learning becomes ubiquitous, value is migrating rapidly from programmatic algorithms that iterate in a predictive fashion, to the vast lakes of data created by those algorithms. The owners of these data lakes can optimise their business models in a manner that cannot be replicated without access to the same volume and quality of data.

This is shifting and consolidating the power of those who, because of their business model, can collect and collate useful data about consumers, geographies, organisms or any other potential focus of commerce. Uber is gathering data about people’s whereabouts. Google is gathering data about the information people seek out on the Internet. Amazon is gathering data about the items people are buying. Facebook is gathering data about people’s social interactions. Apple is gathering data about our vital health signs. The widely available machine-learning language algorithms are infinitely more valuable when executed by these companies because they have the data to iteratively train and optimise their implementations of algorithms, which in turn yields still better data which can still further improve the algorithms. With better data, Uber better anticipates where you want to go, Google better anticipates what you are seeking out on the Internet, Amazon knows what items you might want to buy, Facebook better guesses persons with whom you might want to reconnect and Apple better understands what new medical apps you might need.

This data is not patentable, but it is protectable both physically and legally (primarily through trade secret law), thus providing an impenetrable barrier to entry for others who lack the data of Uber, Google, Amazon, Apple and Facebook, even if they have access to the same machine-learning algorithms. This data barrier is so tall and wide that many of these data-rich companies can afford to place their NGSI code on open source platforms (eg, Google placed TensorFlow on open source platforms), knowing full well that doing so poses no risk to their business models, because the value of their data dwarfs all other sources of value in the NGSI ecosystem, including the value of the algorithms used to generate the data.

It is this shift in value away from patentable processes and apparatuses that implement data-processing logic, towards the unpatentable data they generate, that will primarily minimise the impact of NPEs on NGSI businesses. More generally, this shift will have a significant impact on IP monetisation activity and strategy.

This article explores how IP monetisation professionals will use IP rights to protect investments in NGSI businesses and how IP monetisation will evolve as a result.

The open source bonanza

Open source is not about giving away the ‘software machine’ to users: it is about sharing its design and components among producers (‘free as speech, not as a beer’).

Alberto Pianon, “Trade Secret vs Open Source: and the Winner is”, Erasmus Law and Economics Review 1, February 2004

A second falsehood about the future of NGSI IP monetisation, which runs contrary to the first falsehood discussed above, is that there will be no NGSI blocking patents or intellectual property, because NGSI is generally being delivered using open source software.

We do not think that the open source delivery of NGSI algorithms (as opposed to the open source delivery of NGSI data, which we do not expect to occur) will render NGSI patents useless. Before explaining our prediction, it is worth noting that according to data from the National Bureau of Economic Research, the biggest NGSI companies are spending hundreds of millions of dollars filing massive numbers of NGSI patents, even as NGSI code is increasingly placed on open source platforms that at least discourage the assertion of such patents.

We believe that so many NGSI patent applications are being filed, and so many NGSI patents will continue to issue, precisely because there will be so many open source implementations of NGSI.

Specifically, we believe that although it will be difficult to apply the NPE model to NGSI, it will be possible to apply other more strategic IP monetisation models to open source NGSI implementations.

Strategic licensing and open source platforms

The PC and internet era demonstrated that PC and internet patents became more important as strategic tools when incumbent companies started to see more of their previously trade secret innovations appear in open source implementations created by competitors.

This rise in open source implementations gave rise to three classic IP strategies that could well play out again in NGSI:

• Toll – some early PC and internet pioneers used their intellectual property to maintain lucrative positions in some markets, even as more nimble competitors leap-frogged them commercially. For example, by 2004 Microsoft was accumulating 3,000 patents a year for licensing and blocking. In 2010 Oracle launched a multibillion-dollar lawsuit against Google for copyright and patent infringement on 37 application program interface packages and 7,000 lines of code. The addressable market for Oracle was $42 billion in advertising revenue.
• Contribute – some pioneers, such as Novell, used their patents to thwart the predations of other pioneers, which tried to slow down the adoption of open source. The formation of the Open Invention Network (OIN) by IBM, Novell, Philips, Red Hat and Sony was a direct response to the open source wars, which eventually resulted in Microsoft adopting the same strategy – as it realised that suing the customer, developer or distributor ecosystem was not a winning strategy and released thousands of patents into OIN. The 2018 $34 billion acquisition of Red Hat by dominant patent filer IBM provided further evidence that patent strategies for business models in open source environments are much more sophisticated than file-license-litigate.
• Monetise – other early PC and internet pioneers used their intellectual property to get a consolation cheque, even as they were squeezed out of their markets or failed. AOL’s 2012 sale of 800 (former Netscape) patents to Microsoft for $1.1 billion and HP’s $1.2 billion purchase of Palm and subsequent sale of its patents to Qualcomm are examples of this. The $4.5 billion purchase of Nortel Networks patent portfolio by Rockstar Consortium and $12.5 billion sale of Motorola Mobility to Google remain the landmark transactions in this space.

Microsoft, Novell and IBM each leveraged their PC, internet and smartphone patents, even though those patents were being implemented on open source platforms. In our view the NGSI patent holders will also eventually leverage their patents using similar IP strategies, especially if new open source platforms emerge that are in no way based on code contributions from those original NGSI patent holders.

In summary, it is fair to say that thanks to the toll, contribute and monetise IP strategies, the IP system worked well for PCs and internet pioneers even as much of this technology migrated to open source platforms. This in turn has given investors more comfort as they pour billions of dollars into AI and cloud technologies. The mix of IP protections available 15 to 20 years ago (including software patents) were part of the reason that the PC and internet revolution worked out well for just about everyone, even as open source implementations became increasingly popular for many critical markets, such as operating systems and application servers. We see this mix repeating itself with NGSI, even as the value of NGSI intellectual property shifts away from algorithms to data.

Software patents are dead

The third false theory of how NGSI IP monetisation will play out is that it will become entirely divorced from patents because of the recent death of software patents.

A Washington Post article in September 2013 summarised the ongoing animus that has attached to software patents. “Software patents are particularly prone to such abuses because software is inherently conceptual,” wrote James Bessen of Boston University School of Law.

Enemies of software patents have long proselytised to technology companies about the preferability of trade secret law to protect inventions implemented as software. It is worth noting that while trade secret protection has been a mainstay of software developers for decades, it is a poor cousin of patents for the following reasons:

• trade secrets do not confer freedom to operate;
• trade secrets are difficult to protect; and
• trade secrets hamper innovation largely because of the measures required to protect them.

These reasons mean that from a value and productivity standpoint patents are more desirable than trade secrets in software development since they add greater economic value and are more efficient in producing innovation, especially in an industry such as NGSI, where many interventions must be exposed through open source implementations that destroy trade secrets. Patent-based innovations also allow for better incremental innovations in situations where the entire market decides, through public disclosure of patented inventions, to optimise a technology through the development of industry standards (eg, LTE). This reduces industry costs and benefits consumers.

The over-exaggeration of the death of software patents, as applied to the sorts of software patents being spawned by NGSI businesses, is explained below.

NGSI software as especially patentable machines

Notwithstanding the evident challenges of establishing valid NGSI software patents, the USPTO has been hyperactively issuing patents for NGSI software in particular, and the rate of NGSI patent filing has significantly increased.

For example, in 2010 there were 145 US patents filed that mentioned ‘machine learning’ – six years later this number had soared to almost 600. This pattern of filing inflation has been repeated with neural networks and other AI techniques. In 2017 IBM alone reported that it had secured over 1,400 patents for AI-related innovations.

Blockchain patent filings are also increasing, led by banks and credit card companies. In 2017 there were 97 blockchain patent filings, more than all previous filings in history in this area. This number is expected to be more than 10 times greater in 2018. More than 50% of these filings originate from Chinese inventors. The story is the same with robots, where China and Japan rival the United States for ownership of patents in such diverse areas as huggable robots for healthcare, robot bees for pollination, automotive sensors and LIDAR for autonomous vehicles.

Much of this increased patent filing in NGSI technologies is attributable to the increased standardisation of the building blocks of software. As more and more NGSI programs are built on common platforms such as Docker Containers and Kerberos, software can be more readily understood as a machine that is made up of standardised components. Containers, processes and objects are to software systems what bolts, screws and plates are to mechanical systems. This means that the language used to describe software innovations can be more precisely expressed, and is therefore less ambiguous.

Moreover, a larger proportion of judges, juries, lawyers and patent examiners better understand software and its building blocks, as compared to 20 years ago. These more sophisticated audiences have driven patentees into claiming software more precisely, which in turn has enabled the courts to more readily treat software as ordered combinations of known elements that can be patentable.

All this means that NGSI patents are here to stay. NPEs might still have difficulty monetising NGSI patents that are primarily implemented on open source platforms as explained above, but everyone else should find it straightforward to obtain NGSI patents, and then apply them in some other manner to NGSI open source platforms.

The job to be done

People don’t want quarter-inch drills. They want quarter-inch holes.

(Theodore Levitt)

Having dismissed three popular falsehoods regarding the monetisation of NGSI intellectual property, it is important to return to a theme that recurred as we dealt with each falsehood: the value in NGSI intellectual property primarily lies in data and not algorithms.

Harvard Business School professor Theodore Levitt famously said that when people buy a drill, they do not want a drill, they want a hole. This idea has been developed further by fellow Harvard academic Clayton Christensen into the theory of ‘jobs to be done’. In this way of thinking it is the job, not the tool that is the fundamental unit of analysis for the innovator seeking to develop products that customers will buy. The corollary for IP monetisation is that going forward IP monetisation will focus on the commercial use case (output) rather than the specifics of the tool necessary for the job (input), as has been the case in PC and internet-era IP monetisation.

In IP terms, the last two decades have focused on protecting Levitt’s “drill” using IP rights such as patents and then licensing others to make copies of the drill. As NGSI technologies become dominant; however, the focus is shifting to Levitt’s hole.

In this respect, the most valuable NGSI intellectual property that is being created by innovators is the data and the business model that extracts value from the data. IP rights in the data will fall squarely into the realm of trade secrets, which represents a shift away from the more patent-centric IP realm of the PC, internet and smartphone era.

This shift will have two significant effects on IP monetisation strategies.

First, it will make damages for using NGSI technologies, as opposed to the data generated by NGSI technologies, relatively small, which will kill the old NPE patent lawsuit model as discussed above. The trigger for patent monetisation for a licensor is usually to identify an infringement that is generating substantial revenues for a third party at the innovator’s expense. While the rate of adoption of NGSI technologies is unprecedented, the revenues generated from their deployment remain less impressive when that deployment is decoupled from pre-existing lakes of data that can be used to train the algorithms underlying those technologies.

Second, it means that only a few companies – those that own huge data lakes – will be making most of the money from NGSI, which in turn means that all IP monetisation strategies will need to intersect with the business plans of this small handful of data-rich companies. The market capitalisation of this group (Facebook, Apple, Amazon, Netflix and Google (FAANG)) is more than 50% of NASDAQ-listed companies – an unprecedented level of dominance. A large part of this value derives from these companies’ exploitation of NGSI technology.

The impact on IP monetisation

Focus on the job to be done: focus on monetising data

Going forward, data will form the most valuable part of the IP equation. Whether we are talking about the data Amazon or Google collect as we use their services, the data Uber collects as we move around or the data robotics platforms gather as they clean our houses and service our cars, the real money to be made will be from controlling this data, which in turn will enable the provision of services (eg, self-driving cars and online shopping), which is where the opportunities are.

IP strategy for data protection has and will continue to hinge on trade secret and confidentiality protections and some copyright. Such protections broadly fall into the following categories:

• schemes designed to protect the privacy of certain data and its dissemination; and
• schemes designed to protect the investment companies make in gathering the data.

To address the latter concern, in 2016 the US government enacted the Defend Trade Secrets Act, a milestone for trade secret protection. The act creates a federal civil cause of action so that companies will no longer need to navigate a maze of state laws to enforce their rights when their trade secrets are stolen. It also provides ex parte seizure orders to enable data owners to rapidly limit further disclosure of their prize assets. We expect more IP monetisation strategies to be anchored in the protections provided by such statutes.

Monetisation strategies become focused on rights in data

Since the most important source of value in NGSI intellectual property will be the data locked away in the servers of the largest NGSI companies, we predict that NGSI IP monetisaton strategies will focus on perfecting rights in such data sets and then seeking commercially optimal ways to transfer those data rights to the highest-value users.

There are some interesting AI technologies reaching the market that may accelerate such data rights transfers. These technologies leverage metadata to accelerate development and paths to commercialisation:

• Transfer learning – the ability for AI to be trained in one domain (eg, classifying the sentiment of Amazon reviews) and then transferring and adapting that knowledge to a new domain (eg, classifying the topic of conversation in financial services). This leads to an interesting shift where both AI toolkits and pre-trained AI models become useful and are released under open source licences.
• Active learning – getting AI to ask for an answer when it is uncertain, thus reducing the amount of data required.
• Reinforcement learning – by interacting with its environment and receiving some reward for each action, the AI figures out what it is supposed to be doing.
• Inverse-reinforcement learning – when this ‘reward’ is uncertain, the AI will observe another agent (human or otherwise) and infer what the reward is supposed to be.

Another interesting development that could spur an emerging data rights monetisation industry happens to be one of the central premises of blockchain. Blockchain puts control of data back into the hands of the end user, so companies only have the minimal required information about their end users (ie, their customers). If blockchain technology continues to proliferate, data rights monetisation schemes featuring pools of customer data that can be disaggregated, catalogued, re-combined into novel sets and then made available to any company, become feasible.

All these technologies open up possibilities for new sorts of IP monetisation strategies, which focus on rights in data rather than inventions. Monetisation strategy will evolve to develop new techniques of monetising data, such as data pooling arrangements among smaller companies having complimentary data sets and novel usage of public data and data leasing agreements to enable newer companies to more rapidly train their copies of algorithms on reliable data.

As the value of NGSI intellectual property migrates to data, M&A activities will also be affected. Domain-specific data that is used for training AI, metrology and testing together with metadata derived from significant deployment and other confidential customer information will present investment bankers with a challenge in valuation terms. The monetisation of this critical asset will drive corporate success in the long run and may present the most valuable asset for future acquisitions, in the same way as acqui-hires drove some significant deals in this decade (eg, Google’s acquisition of DeepMind).

A China (and Europe) strategy is more important than ever

Protecting an NGSI business using rights in data poses additional challenges because (notwithstanding the EU General Data Protection Regulation) data can move freely and instantaneously, and because it retains its value even as it crosses national borders, which is not the case for patented inventions. Given the central role that China is playing in NGSI technology, any IP strategy or monetisation scheme must take into account the likelihood of data flowing between the United States, Europe and China.

Regulations pertaining to NGSI data will vary extensively across all three jurisdictions. In previous IP monetisation eras, only a light degree of inconvenience was experienced as a result of cross-border IP flows, mostly in the form of export control restrictions that were rare given the public disclosure of patents and an occasional antitrust authority review for big-dollar IP licensing deals (especially in China).

Given the non-public nature of NGSI data, the complicated and greatly varying rules, as well as the growing government interest in the regulation of such data (as compared to patented inventions), jurisdiction-specific IP monetisation strategies will be more important than ever.

Expect continued event-driven monetisation of patents

In many respects the PC, internet and especially smartphone patent wars of the past 20 years were the result of some significant business events rather than an expected result of some grand pre-meditated IP monetisation scheme. The noteworthy business failures of Nortel and Motorola, the decline of Nokia’s once-dominant handset business and the slide into obscurity of AOL and Yahoo – together with the capitulation of early smartphone innovators like Siemens – drove IP monetisation opportunities that were primarily the products of business strategy failures, and not pre-meditated IP monetisation schemes. These events coincided with the rise of Google, Apple and Samsung and the resulting battle for market share further turbo-charged these monetisation opportunities.

Such significant business events will continue to occur and open up IP monetisation opportunities. It is a venture capital axiom that more than 90% of all investments fail. NGSI innovators will follow the same trend and when this happens (or perhaps an antitrust decision breaks up the FAANG oligarchy) patents could become very important. The event-driven importance of patents is recognised – as evidenced by the tremendous amount of patent filings occurring in the space. When the next large failure does occur we do think that some more traditional monetisation activities will take place.

The scale and scope will be far smaller than in the past; however, because of the smaller number of patents proportionately, the stronger position of dominant incumbents, the greater relative importance of data and the likely lack of skills and experience in a retrenched NPE industry.

There will be very few FRAND patents involved in NGSI monetisation strategies

As mentioned above, the building blocks of software have become standardised as software has come to be conceptually understood as a machine that is made up of fungible components. For example, Dockers Containers and Kerbeross processes are ubiquitous, are the subjects of extensive open source libraries that have become de facto standards and were jointly developed by a number of contributing companies. Normally, that would mean that companies would file SEPs before contributing to a common project (ie, 3GPP 5G), the project would react by imposing FRAND obligations on any of its technology contributors as a condition precedent for incorporating those contributions into the project and, ultimately, FRAND licensing and litigation activities would result.

However, for NGSI technology the open source agreements governing the formation of the open source project using a royalty-free licensing scheme will continue to stop FRAND licensing from ever catching hold.

Reliance on governments to protect investments in data

For well-heeled US NGSI companies prepared to lobby extensively in Washington, and large NGSI Chinese companies like Tencent that are themselves quasi-state actors, expect their legal and security efforts to protect their data lakes to be augmented by efforts to enlist national governments to help them grow and strengthen that barrier, with an objective of forming a great wall of protection around this precious data. An ability to exclude competitors from accessing data, (eg, through lobbying or to access competitors’ data through the provocation of antitrust actions) will become a valid IP strategy for competitive advantage. That ability will need to factor in a government’s objectives in regulating the given set of data at stake (eg, health data pertaining to its individual citizens), which can range from maintaining the individual privacy of citizens, to tracking their individual activities for security or political purposes, to myriad other public policy objectives.

Facebook and Aldous Huxley

The most striking difference between IP monetisation of the past and that of the future is the pervasiveness, ubiquity and utility of data about virtually everything, and society’s interests in regulating its dissemination and use. The information that is collected from the public and refined through iterative use has unprecedented value. Control of and insight into this data is not just a commercial opportunity but increasingly a societal and political issue.

Unlike the electrification of the United States, the establishment of the railway systems or telecoms infrastructure, the innovations now taking place are 100% intangible. This means that they are almost instantly portable and of immense value to both the oligopoly in which they reside and to government agencies at home and abroad.

Imagine for a second that Facebook stumbles a little more and collapses – what would happen to the vast data lake that Mr Zuckerberg and his organisation has collected and curated? Can we imagine Tencent bidding freely for the assets in liquidation? Or Google? Or perhaps the NSA?

The big data companies have grown so fast and become so valuable that notwithstanding the occasional public pantomimes involving security agencies, it is clear that these are not ordinary companies and the intellectual property that they hold is not theirs alone to monetise.

In this respect, the future for IP monetisation in NGSI technologies will undoubtedly be entwined with government and security policy to an extent hitherto unseen. Licensing, enforcement and the commercial exploitation of primary data and metadata used to train AI and deploy through the Internet of Things in worldwide software updates will become necessarily multi-jurisdictional, quasi-governmental and may stray beyond the remit of corporate IP departments, hedge funds or litigation finance firms.

Matthew Vella is also a partner at Prince Lobel LLP. The authors wish to thank Professor Bill Moloney at the University of Massachusetts for his insights into the technical aspects of NGSI.