H.264 Streaming Fees: What Changed, Who's Affected, and What It Means
H.264 (AVC) streaming royalties have been a solved problem for a long time. In August 2010, MPEG LA, the predecessor organisation to the Via Licensing Alliance (Via), announced that internet video delivered free to end users would carry no royalties for the life of the licence, a move widely read as a response to Google's open-sourcing of VP8. Subscription and pay-per-view content remained royalty-bearing, but the annual cap for large SVOD services was $100,000, a number so modest that most platforms treated it as a rounding error in the licensing budget. That picture changed at the start of 2026.
Via's new AVC Streaming Licence Fee structure, which applies to new licences beginning in 2026, replaces the old single-cap model with a tiered system that scales sharply with platform size. For Tier 1 OTT platforms, or those with 100 million or more subscribers, the annual list fee is now $4,500,000. The same figure applies to Tier 1 FAST services (100M+ daily users), Tier 1 social media platforms (1B+ monthly active users), and Tier 1 cloud gaming platforms (15M+ monthly active users). Tier 2 and Tier 3 fees run $3,375,000 and $2,250,000, respectively. Only the smallest category, platforms Via defines as small or nascent, retains the $100,000 fee.
Table 1. Via's new rate structure for H.264.
The jump from the floor to the Tier 1 ceiling is 45x. For large platforms that previously viewed H.264 licensing as a negligible line item, which is a material change, provided, of course, it actually applies to them.
This change triggers several obvious questions. Who is actually affected? Why wasn't there a public announcement? How does the new fee structure work administratively? And what about the widespread belief that H.264 patents have largely expired? To address these questions, we sent two sets of questions to Via and a separate set to patent licensing attorney Jim Harlan, JD/MBA. Their responses are woven throughout the article, and Jim's complete Q&A appears as an appendix.
Who Is Actually Affected
The first and most important question is scope. Via's response to my questions was clear on this point: all AVC licencees with an active licence as of the end of 2025 were entitled to retain the streaming terms of their original agreement. The new fee structure applies only to previously unlicenced implementers seeking a licence starting in 2026.
That narrows the universe considerably. I asked Via about their outreach regarding this transition, and they replied that it had contacted unlicenced media companies in 2025 to inform them of the change and give them a window to secure a licence under the old terms.
I also asked whether there was any published notice of the change, such as a press release, public announcement, or similar. Via stated that it communicated directly with unlicenced AVC companies rather than issuing a public statement. That approach may have left some smaller or peripheral players unaware. If you're one of those companies, or a company that didn't respond to Via's outbound communications, a $100K problem may have grown considerably.
Why did Via make the change? In its reply to this question, Via noted that the original streaming terms were defined more than a decade ago, when AVC streaming was not yet a significant business category. The company characterised the restructuring as a reflection of current market conditions, not a break with established practice.
How the New Structure Works
The fee schedule divides the market into two broad segments: Video Content Streaming, which covers OTT, FAST, social media, and cloud gaming, and Other Video Content Delivery, covering cable/satellite television and OTA networks. Each service type uses its own unit of measurement. OTT fees are based on subscriber counts; FAST on daily users; social media and cloud gaming on monthly active users; cable and satellite on customer counts; and OTA on household viewable reach.
A few operational details matter for anyone trying to model their exposure. On the stacking question, whether a company running OTT, FAST, and social video under one roof gets charged in multiple categories, Via says no. A service is assessed once, in one category. Fees are also assessed per legal entity rather than per parent company, meaning affiliates each need their own licence.
However, an enterprise structure is available that allows multiple affiliates to be covered under a shared cap. Notably, Via confirmed that the AVC licence includes two separate enterprise caps: one for devices and one for streaming services. That means streaming fees are not offset by device royalties, and a company that already hits the device cap still faces the streaming fees independently.
Figure 2. The Via H.264 licensing page has two caps: one for devices, and this one for streaming.
When asked why the list price had not come down, given the aging of the AVC patent portfolio, Via pointed to the codec's continued relevance: AVC still dominates streaming traffic globally, and the pool covers the large majority of essential patents required to implement the standard.
The Patent Tail and the Bigger Picture
By way of background, when I first noticed the royalty change on the Via site, I posted about it on LinkedIn. Not surprisingly, the post generated significant engagement, including many versions of this question. "I assumed that patents for H.264 (AVC) have expired, and since January 1, 2024, licences are free."
Figure 3. Jim Harlan provided insights into royalties when patents are expiring.
Rather than rely on the multiple LinkedIn lawyers who commented without the credentials, I decided to pose a series of questions to Jim Harlan, a real attorney with close to ten years of experience with InterDigital and over 25 years of patent licensing experience. Jim was kind enough to respond; I'll summarise his answers for this article and append all of his answers below.
I first asked how we should think about a mature standard like H.264 in 2026, where many but not all patents have expired. "Maturity changes the economic context, not the legal framework," he said. "The analysis shifts from an emerging innovation platform to a legacy but embedded infrastructure. That shift can affect valuation, but it does not nullify licensing obligations." Courts evaluating FRAND rates still look to comparable licences and consider the composition, strength, and remaining life of the patents still in force.
Cherry-Picking Expired Profiles
Several LinkedIn posters recommended limiting H.264 implementations to the Baseline or Main Profiles, where most patents have expired, and avoiding the High Profile, MVC, and SVC, where they have not. I asked Harlan if that was a realistic strategy for a commercial service. He responded, "In theory, a company could attempt to design around specific patented features. In practice, which requires a detailed claim-by-claim analysis of the relevant patents and the specific implementation. Patent claims do not necessarily map neatly to profile labels, so determining whether a subset is truly non-infringing typically requires careful legal and technical analysis."
I followed up with, "If a company deliberately stays on Baseline/Main profiles and operates mostly in countries where High-profile patents appear to be expired, how much comfort—if any—does that give you as counsel? Harlan's answer placed doubts on this strategy as well. "Operating within certain profiles or jurisdictions may reduce risk in some circumstances, but it does not automatically eliminate exposure. The analysis would still require evaluating the remaining patents that may cover the relevant implementation and the jurisdictions where those patents remain in force."
I next asked how a global streamer should think about tail patents that only exist in a handful of countries, and whether having users in a country with a live patent creates meaningful risk even if the rest of the world is clear. Harlan said patent rights are strictly territorial, and the infringement analysis typically turns on three factors: where encoding occurs, where content is transmitted from, and where it is received or used. A service can be legally clear in most markets and still face real exposure in others.
He noted that companies sometimes evaluate technical design choices or geographic considerations to manage this risk, but the analysis often comes down to practical business and operational realities. The presence of an enforceable patent in even a single key market can sustain a licensing obligation regardless of what the rest of the world looks like.
Late-Stage FRAND Pricing
On whether list prices should track the size of the remaining patent portfolio, Harlan said there is no legal requirement for automatic rate reductions as patents expire. That said, there is a meaningful distinction between prices holding steady and prices increasing sharply late in a standard's lifecycle. "A sharp late cycle increase that is not supported by portfolio strength, remaining patent life, or comparable licences is more likely to attract scrutiny under a FRAND analysis," he said. Whether the new Via structure clears that bar is a question that would require access to the full patent list, a technical analysis of remaining claims, and a comparison against market benchmarks, none of which can be completed from public information alone.
For large services looking at multi-million-dollar list fees, Harlan outlined the realistic options: negotiate a licence, explore design alternatives that may avoid specific patent claims, accept the risk of operating without a licence, or litigate. In practice, he said, negotiation is the primary avenue. Companies may also seek to align terms with their specific business model, expected usage, and the remaining life of the relevant patents.
Harlan also identified the two biggest misunderstandings he sees in technical discussions about expiring codec patents. The first is that the widespread expiration of patents necessarily eliminates licensing obligations. The second is that FRAND rates must track the number of patents remaining rather than the portfolio's overall value and strength. Both assumptions are incorrect, he said, and both lead implementers to underestimate their actual risk exposure.
Where This Leaves You
If you are already licenced under Via's pre-2026 AVC terms, none of these changes affect your situation today. Your existing rates are grandfathered, and your licence structure remains in place.
If you are not licenced and are delivering AVC content at meaningful scale, the new fee structure is the starting point for any conversation with Via. The company frames the pool as providing a centralised solution for full compliance, and it is actively encouraging unlicenced implementers to come to the table.
It is also worth keeping this in proper context. The H.264 situation does not exist in a vacuum. Nokia has secured significant licencees for HEVC content. Access Advance and Avanci have published rates for a pool asserting content royalties across AVC, HEVC, VP9, VVC, and AV1 that could push major platforms toward nine-figure annual exposure. Against that backdrop, the question of whether a $4.5 million H.264 streaming fee is high, low, or reasonable depends considerably on what the rest of your codec licensing stack looks like and what your legal and technical analysis shows about the patents that remain.
Full Q&A with Jim Harlan
Here the unedited Q&A with Jim Harlan addressing FRAND analysis, safe-subset strategies, territorial risk, and litigation thresholds in detail appears as an addendum to this article.
- From a patent and FRAND perspective, how should we think about a mature standard like H.264 in 2026, where many—but not all—patents have expired?
From a FRAND standpoint, maturity changes the economic context, not the legal framework. The analysis shifts from an emerging innovation platform to a legacy but embedded infrastructure. That shift can affect valuation, but it does not nullify licensing obligations. Courts evaluating FRAND rates will still look to comparable licences, but those comparisons need to account for the maturity of the portfolio. Where a standard’s patent base is shrinking over time, the analysis should consider the composition, strength, and remaining life of the patents that remain.
Pricing and End-of-Life Behaviour
- As the underlying patent portfolio shrinks over time, is there any general expectation that licence prices should come down, or is that purely a business decision as long as FRAND is respected?
Legally, there is no automatic downward adjustment as individual patents expire. But economically, as a portfolio materially shrinks over time, questions can arise about whether the remaining set of patents still justifies the same rate under a FRAND analysis, particularly in light of comparable licences and the technical contribution of the remaining patents.
- Is it typical, in your experience, for licensors or pools to raise prices or restructure fees near the end of a standard’s life, once most patents have expired?
It is not typical as a general pattern to raise prices simply because a standard is aging. What sometimes occurs is a restructuring of licensing models or a shift in how royalties are allocated across different products or services. A sharp late cycle increase that is not supported by portfolio strength, remaining patent life, or comparable licences is more likely to attract scrutiny under a FRAND analysis.
- Are there examples in other standards (telecom, audio, etc.) where courts or regulators have said “at this stage, those prices are no longer FRAND” because too few patents remain?
There is no bright-line precedent where a court has said that prices are no longer FRAND simply because too few patents remain. Courts generally evaluate FRAND through factors such as comparable licences, portfolio strength, and the overall reasonableness of the rate in the relevant market context.
- When a pool charges the same (or more) as the patent set shrinks, what factors would you look at to decide if that is still “fair and reasonable”?
Relevant factors would include comparable licences in the market, the technical importance of the remaining patents, whether those patents continue to read on core implementation features, the remaining life of the patents, and the overall strength and scope of the portfolio relative to earlier licensing benchmarks.
“Safe Subset” and Working Around Patents
- Engineers sometimes say “H.264 is free if you avoid High Profile, MVC, SVC, and stick to a safe subset.” From a legal point of view, is that ever a realistic strategy for a commercial service, or is it mostly wishful thinking?
In theory, a company could attempt to design around specific patented features. In practice, which requires a detailed claim-by-claim analysis of the relevant patents and the specific implementation. Patent claims do not necessarily map neatly to profile labels, so determining whether a subset is truly non-infringing typically requires careful legal and technical analysis.
- If a company deliberately stays on Baseline/Main profiles and operates mostly in countries where High-profile patents appear to be expired, how much comfort—if any—does that give you as counsel?
Operating within certain profiles or jurisdictions may reduce risk in some circumstances, but it does not automatically eliminate exposure. The analysis would still require evaluating the remaining patents that may cover the relevant implementation and the jurisdictions where those patents remain in force.
- Are there good precedents where implementers successfully argued, “we only used a non-infringing subset of the standard, so we don’t need a licence”?
Courts have recognised that implementers may avoid infringement if they do not practice the patented claims, but successfully demonstrating that a particular implementation falls entirely outside the scope of all relevant patents can be complex and fact specific.
Territorial Issues and the Tail
- How should a global streamer think about tail patents that only exist in a handful of countries (for example, Brazil or a few remaining US patents)? Is that the kind of thing you try to route around, or do you assume worldwide risk?
Companies typically evaluate both the geographic scope of the remaining patents and the operational realities of their services. In some cases, it may be possible to manage exposure through licensing, technical design choices, or geographic considerations, but the analysis often turns on practical business and operational factors.
- If your encoders are in the US or EU, but you have users in a country with a live H.264 patent, does that create meaningful infringement risk in that country, even if most of the world is clear?
Potentially. Patent rights are territorial, and infringement analyses can depend on factors such as where encoding occurs, where content is transmitted from, and where it is received or used. The specific facts and the law of the relevant jurisdiction would determine the level of risk.
Strategies vs “Grin and Bear It”
- For a large streaming service looking at multi-million-dollar list fees late in a standard’s life, what are the realistic options? Negotiate, redesign around a subset, stay unlicenced and accept risk, or litigate?
The typical options include negotiating a licence, exploring design alternatives that may avoid certain patent claims, accepting the risk of remaining unlicenced, or litigating to resolve disputes over infringement or FRAND terms. The appropriate course often depends on the magnitude of the potential exposure and the strength of the legal positions on both sides.
- Are there practical strategies you’ve seen to minimise risk and cost—short of just paying list price—or is the honest answer that most big services either negotiate or grin and bear it?
Negotiation is often the primary avenue. Companies may also seek to align licensing terms with their specific business models, explore cross-licensing where relevant, or structure agreements that reflect expected usage or remaining patent life.
- At what rough scale—revenues, proposed royalties, remaining patent life—do you start telling clients, “It might actually be worth fighting this” instead of settling?
That assessment usually involves evaluating the expected financial exposure, the strength of the legal and technical defenses, the remaining life of the patents, and the broader business implications of litigation.
FRAND and Portfolio Licensing
- Courts have previously found the AVC pool licence to be FRAND in handset cases. How much room does that give a pool to change pricing or structure later in the life of the standard?
Prior findings that a particular licence structure was FRAND can provide useful benchmarks, but they do not necessarily determine future outcomes. Courts evaluating FRAND rates may still consider the current composition and remaining life of the portfolio, as well as comparable licences in the market.
- Is there any FRAND principle that says licencees should get some benefit as patents expire, or is it accepted practice that portfolio rates can stay flat even as patents are added or drop out?
There is no FRAND principle that requires automatic rate reductions as patents expire. Portfolio licences often reflect the value of the overall portfolio at a given point in time. However, FRAND analysis may still consider whether the strength and scope of the portfolio that remains supports the rate being charged.
- If a pool introduced significantly higher streaming fees late in the lifecycle, do you see a plausible FRAND challenge there, or is that a very high bar in practice?
A challenge could arise if the increase appears inconsistent with comparable licences, the strength of the portfolio, or the overall economic context. Whether such a challenge succeeds would depend on the specific facts and the evidence presented.
Helping Calibrate the Article
- When you look at this specific situation—shrinking patent set, new higher streaming fees—what’s the fairest way for someone like me to describe it in an article without overstating the legal risk or implying bad faith?
One way to describe the situation is to note that as the essential patent base narrows over time, questions may arise about how licensing terms relate to the value of the remaining portfolio. Whether rates remain FRAND depends on factors such as the technical importance and remaining life of the surviving patents and comparable licences in the market.
- What are the one or two biggest misunderstandings you see in engineer / LinkedIn discussions about expiring codec patents and end-of-life licensing?
Two common misunderstandings are that the expiration of many patents necessarily eliminates licensing obligations, and that FRAND rates must track the number of patents remaining rather than the overall value and strength of the portfolio.
Related Articles
Following on the news that Via Licensing has acquired codec patent pool administrators MPEG LA to form Via LA as a "unified entity," Via LA President Heath Hoglund discusses how the pool will operate, as well as resulting synergies including streamlined IP distribution and economies of scale in this interview with Streaming Learning Center's Jan Ozer.
08 May 2023