The video codec patent licensing mistake and how to avoid it
Licensing an open standard on FRAND terms is not the same as licensing a proprietary technology and the two approaches must never be blurred. Nowhere is this more the case than with VP9, AV1 and now AV2
By Mattia Fogliacco
Sisvel launched its Video Codec Licensing Platform, covering AV1 and VP9, in 2019. We did so in the face of a great deal of misinformation concerning the ‘royalty-free’ nature of the technology. The reaction was extremely positive.
The AV1 pool provides one-stop access to over 1,900 patents relevant to the AV1 video coding specification, owned by more than 20 world-class innovators. So far, it has licensed approximately 50% of the AV1 finished product market, notwithstanding that this is a young technology. The VP9 pool, meanwhile, has now licensed almost the entire addressable market.
But Markets for Technology never stand still. We are currently finalising our plans for AV2, the specification for which was released in June 2026. Our preparations are at an advanced stage. Sisvel’s technical and programmes teams have been working on the codec since January 2025. As a result, we are well on our way to creating an AV2 licensing solution that balances the interests of both licensors and licensees.
Recent press reports indicate that Sisvel is not the only pool operator looking at AV2. This competition is welcomed wholeheartedly. Over the past nine years, Sisvel has gained unparalleled expertise not only in video coding technology, but also in understanding the most effective ways to make the patents that underpin it available as transparently and efficiently as possible. We are confident that the offer we make will be the most compelling from Day 1 because it will best reflect the market realities.
It is clear that the race to license AV2 will be framed by some as a contest of scale – in other words, who can bolt the most codecs into the biggest pool. However, we believe that this framing fundamentally misses the point. The real issue is whether a programme respects the nature of the technology it licenses.
That means this is the moment for principled choices.
Not all pools are the same
The publication of the AV2 specification and the licensing initiatives now forming around it have reopened one of the most important (and most misunderstood) questions in our field: are all patent pools the same? The convenient answer, offered by those pitching a single vehicle to license every video codec together, is yes. But the correct answer is no.
A licensing programme built around a standard adopted by a standards development organisation (SDO) and a licensing programme built around proprietary technology developed inside a private consortium are not two flavours of the same basic product. Instead, they rest on different origins, different legal obligations, different safeguards and different remedies. Treating them as interchangeable is not a simplification; it is a category error – and one that can quietly transfer the wrong obligations onto the wrong assets, to the detriment of licensors, licensees and the wider Markets for Technology.
Everything starts with how a technology comes into the world.
The SDO bargain
A standard is the product of a deliberate, pro-competitive, collective process. Within SDOs such as ISO/IEC MPEG, ITU’s VCEG, DVB, ETSI or IEEE, firms, universities and research centres compete to include the best technical solution in a standard, and agree on a single, interoperable specification that the whole market can implement. This process is governed by principles that are largely derived from competition law: openness, transparency and the availability of the essential patents on fair, reasonable and non-discriminatory (FRAND) terms.
In exchange for having their technology included in a standard that others implement to ensure interoperability, contributors accept an ex-ante FRAND commitment. This is the price of standardisation. It is what allows a consumer to buy any television or smartphone and be confident that it will interoperate with every broadcaster and every network in the world. It is a process that time and again has proved its worth and delivered major societal benefits.
The proprietary consortium
A proprietary codec, such as VP9, AV1 or AV2, follows a fundamentally different path. It is not developed within an accredited standards body but inside a private consortium, in this case the Alliance for Open Media (AOM). Its specifications, as Sisvel has consistently noted since 2019, have not been adopted by any standard-setting organisation or been subject to any IPR policies set by such organisations.
That single fact changes the legal landscape entirely. There is no standardisation bargain, no SDO IPR policy and therefore no FRAND commitment attaching to patent owners that have not joined the consortium and thus have not pledged anything.
The ‘royalty-free’ claim and who it binds
The defining marketing claim of the open-media movement is that its codecs are ‘royalty free’. It is essential to be precise about what this means.
A royalty-free pledge is a promise made by the members of a closed consortium in respect of their own patents only; nobody else’s. It is a contractual undertaking made by parties within a self-defined group. It cannot, as a matter of law, curtail the rights of a third party which is not a member and has made no such promise. Many of the most significant patented inventions practised by these codecs are owned by innovators – universities, research institutes, companies – which are outside the consortium.
There is also a commercial logic that explains the divide. Consortium members can afford to give away their technology because they monetise it elsewhere, through chips, devices, platforms and services that they sell in the hundreds of millions. They cross-finance their R&D from adjacent revenue. The independent innovator which invents but does not sell the downstream product has no such outlets. For that innovator, a royalty-bearing licence may be the only way to recoup and reinvest. This is the mechanism that keeps its innovation cycle alive.
Knowing versus discovering: a fundamental asymmetry
There is a further difference between the two worlds that is too often overlooked but is arguably the most consequential of all. It concerns a deceptively simple question: does the patent owner even know that its technology is being used?
In a standard: knowledge by design
In the SDO/SEP context, the patent owner knows that its patents are practised by the standard precisely because it contributed those very technologies during the standardisation process. It sat at the table, made technical proposals and saw which of them were adopted into the specification. The link between a specific contribution and a specific patent is transparent because the patent owner created both. Knowing that the standard reads on its patents, the owner can then choose whether to seek royalties on FRAND terms – and can do so with a clear line of sight from patent to standardised feature.
In this way, standardisation delivers knowledge by design. Essentiality is comparatively straightforward to establish because the patented technology entered the standard through an open, documented process in which the owner was an active participant.
In a proprietary consortium: discovery from the outside
For a proprietary codec, such as VP9 or AV1, the outside innovator’s position is the opposite. A patent owner that did not participate in the private consortium has no visibility whatsoever into whether its inventions have been used, how they have been implemented or when this began. The technology was assembled behind closed doors, by a group the owner never joined, using processes the owner never saw.
In fact, if a patent owner external to the private consortium were to take the claims of the consortium’s founders at face value, it would assume that its technologies have not been used. After all, a group claiming a royalty-free regime should not, by definition, integrate technologies patented by others – at least, not without asking.
But as Sisvel found out with AV1 and VP9, this is not how it works in practice. One may anticipate that the same will happen with AV2. If this is the case, it means that many owners of technologies at the forefront of advancements in video coding will not realise that their inventions are being used by these proprietary specifications.
As a result, these patent owners must discover the use of their own technology from the outside. This will involve reverse engineering a specification and mapping it, claim by claim, against a portfolio, without any of the contribution record that standardisation would have provided. This is not a clerical exercise. It requires deep technical expertise and very substantial, sustained investment simply to reach the starting point of knowing that an infringement exists at all.
Why this changes both cost and equity
An asymmetry such as this reshapes two things at once:
The cost profile is fundamentally different. The very fabric of the investment required to run a programme based on proprietary technology such as AV1 or AV2 is fundamentally different to what is required for a programme based on FRAND-encumbered SEPs. The heavy lifting and the hefty expenditure reside in identification and evidence of use, not merely in administration. A licensing solution for proprietary technology must fund and sustain a genuine technical capability to find, analyse and prove essentiality where no contribution record exists. That is a materially larger and riskier undertaking, and it is a principal reason why credible proprietary programmes cannot be improvised.
The equity of the situation is, frankly, terrible for third-party patent owners. Consider what actually happens in a private consortium: third parties’ technologies may be practised by the codec even though those third parties are not aware of it, were not consulted and – crucially – may never have agreed to make their inventions available at all, let alone on a royalty-free basis. Their property is used without their knowledge and without their consent, and they are then told that the result is ‘free’. Compare that with standardisation, where the patent owner knowingly and willingly brings its technology forward under a transparent policy. On any honest reading, it is the proprietary consortium model, rather than the SEP one, that raises serious fairness concerns.
So, the difference is not marginal. In knowledge, in cost and in equity, the two situations are poles apart. By pricing and structuring proprietary technology licensing as if essentiality were as self-evident as it is in a standard, a programme misunderstands the work involved and, worse, obscures where the real fairness deficit lies.
Different obligations, different safeguards, different remedies
Because the two programmes are born differently, they carry different legal DNA. This distinction is not academic; it drives the terms, the pricing discipline and the enforcement tools available in each case.
A standard-based (SEP) programme involves the following:
Rate discipline: Pricing is anchored to FRAND.
Non-discrimination: Similarly situated licensees must be treated similarly. This is a direct constraint on how terms can differ.
Constrained enforcement: The right to exclude is exercised within a judicially defined framework of good-faith negotiation steps before an injunction is available.
Governed pooling: SEP pools operate inside a body of competition law guidance built up since the 1990s, from the MPEG-2 clearance onwards.
These safeguards are not burdens imposed at random; they are the quid pro quo for the extraordinary market position that a standard confers.
Where the technology is proprietary and no FRAND commitment attaches, the picture is different. Adoption is a market choice. Licensing is an ordinary, voluntary commercial transaction governed by ordinary patent law. The implications are as follows:
The rate reflects the market value of the patented technology and is not capped by a FRAND ceiling or an aggregate royalty theory borrowed from standardisation.
There is no external non-discrimination duty imported from an SDO IPR policy that the patent owner never signed.
The full spectrum of patent remedies remains available. Patent owners would be under no obligation to license at all. Instead, they could just exercise their right to exclude.
Why the two must never be commingled
It follows that, however convenient it may sound, combining standard-based codecs and proprietary codecs into a single, undifferentiated licensing vehicle is a mistake of principle, not merely of housekeeping.
Blend them and you create confusion in both directions:
Onto proprietary assets, a combined vehicle risks importing FRAND obligations, non-discrimination duties and remedy constraints that simply do not apply, gratuitously surrendering rate freedom and enforcement tools that the patent owner is entitled to keep.
Onto standard-essential assets, it risks diluting the carefully balanced safeguards that give the SEP system its legitimacy – the very framework that regulators, courts and implementers rely upon.
A single pool that licenses HEVC and VVC (true SDO standards) together with AV1 and AV2 (proprietary consortium codecs) as if they were one homogeneous mass is not “a one-stop shop for the market”. It is a blurring of two philosophies that regulators – which are already monitoring open-media licensing under EU competition law and the Digital Markets Act – have every reason to examine closely. Efficiency for the licensee should never come at the cost of legal coherence.
Discipline by choice, not by compulsion
None of this means that a proprietary technology programme should be a free-for-all; quite the opposite. The best proprietary programmes voluntarily adopt the good governance that makes pools work; but they do so as a matter of self-imposed discipline, not legal compulsion, and in return for prompt and efficient licensing.
Sisvel’s Video Coding Licensing Platform has demonstrated exactly this since 2019. VP9 and AV1 were offered as two separate programmes, each supported by:
independent, accredited third-party essentiality evaluation before launch;
transparent, published terms;
a one-stop shop that reduces friction and transaction costs; and
a rate structure that takes into account how the technology is actually used – for example, the distinction between display and non-display devices.
None of this was compulsory. It was all done on an entirely voluntary basis, to make access to the technology as simple, transparent and efficient as possible.
The market responded: Sisvel’s programmes have licensed the overwhelming majority of the VP9 market and a substantial share of AV1, while avoiding the litigation and disruption seen elsewhere.
This is the model that we are extending to AV2: a balanced, efficient programme, covering both devices and content, built on years of technical analysis to identify the genuinely essential assets – and offered on terms appropriate to a proprietary technology, not on terms borrowed from a standard that never was.
Staying honest
A standard carries FRAND obligations and the safeguards that come with them and tells the innovator exactly what was used. A proprietary technology imposes no such obligations and leaves the outside innovator to discover the use of its own inventions at great cost – often without ever having consented to their deployment in the first place. More than anything else, this is a matter of equity. It is simply not right de facto to impose potentially substantial obligations and costs on a patent owner without its knowledge or permission.
Claiming that standards and proprietary-based pools are essentially two peas in a pod that can be combined into one is misguided. Instead, the course that best serves innovators and implementers alike – and to which Sisvel is committed – is to treat them independently: standard-essential programmes governed by the discipline of standardisation; and proprietary technology programmes governed by the discipline of the market.
Two philosophies, two programmes. That is not a limitation. It is how the system stays honest.
Mattia Fogliacco is President and CEO of Sisvel International
This article sets out a general perspective on licensing philosophy and is intended for public discussion. It does not disclose confidential portfolio information or the essentiality analyses that underlie any specific programme
