Opinion
Earlier this week, a dispute over a Qualcomm-Huawei license agreement broke out, but only in the court of public opinion (October 5, 2026 ip fray article):
After Huawei made an announcement and various media (including ip fray) reported, Qualcomm issued a public statement according to which Huawei was allegedly not the net licensor under the deal.
Sadly, the interpretation of complex commercial agreements in this field is nowhere near as mathematical as one might think. And purposely so. That’s why we don’t have a reason to believe that anyone lied. We actually assume that there’s simply more than one reasonable and defensible position that one can take on this sensitive question.
We perfectly understand that both companies legitimately seek to maximize the perceived value of their patents. This is not a PR question. In future negotiations, litigations, or arbitration proceedings over fair, reasonable, or non-discriminatory (FRAND) licensing rates for their standard-essential patents, this is potentially about a ton of money. So Qualcomm has an obvious interest in arguing that Huawei pays a higher royalty for its use of Qualcomm’s patents than the other way round. As does Huawei vice versa.
Three deals in one
Huawei and Qualcomm agreed on three deals in one:
- a license from Huawei to Qualcomm;
- a license from Qualcomm to Huawei; and
- a sale of patents by Huawei to Qualcomm.
The above list has no particular order. We have no idea which component is the largest. All we know is that the overall transaction volume was high enough to meet the thresholds for regulatory review in the U.S., but that doesn’t mean too much here.
If this deal is, as we assume, just one agreement, then forget about that contract providing a clear breakdown, or a bottom-up derivation. That’s not what industry players do. They love obfuscation if it means that either one can argue in FRAND negotiations and disputes what suits them best.
Even simpler deals are hard to unpack
We watch FRAND hearings all the time, and we read the related decisions. There’s a fairly high degree of transparency in the English courts. The two FRAND rate matters that have been litigated most extensively there are InterDigital v. Lenovo and Optis v. Apple. Others settled out sooner.
In those cases, it was not just about which license agreements with third parties actually were useful points of reference (so-called “comparables”). Even for a given comparable, it was often still the question what it really meant.
You don’t need a cross-license, much less a cross-license combined with a patent transfer, to have uncertainty about the economic implications. In such a scenario, the parties will usually always know how high the balancing payment is. For short-term financial purposes, that’s what matters.
But even a one-way license (say, a deal between InterDigital, which does not need cross-licenses, and an implementer of a standard) can have more than one component. One common example that has kept UK courts and lawyers busy more than once is the unpacking of back and future royalties:
- As UK judges have noted, there obviously is an incentive for the patent holder to make the prospective royalty as high as possible, and it’s easier to make concessions on royalties for past unlicensed use.
- For the implementer, this isn’t irrelevant, but it won’t be too important in practice until they negotiate the next deal with the same licensor. And by then, a lot of things will have changed, especially if it’s a long-term agreement.
Patent transfers have become a popular component of license deals
It is seen increasingly often that months after the announcement of a license deal, some patent assignments between the parties show up in assignment databases. For example, Philips received 5G patents from OPPO two years ago (August 8, 2024 ip fray article). It’s a safe assumption that the reason was a license deal with a patent purchase component.
If parties agree on the amount of a balancing payment, but they can’t agree on a derivation that works for both, adding a patent transfer to the package can solve the problem. Patent valuation is even harder — far harder — than FRAND determinations.
Qualcomm’s version of the story is that it is the net licensor, the one who receives a balancing payment as far as the licensing part of the deal is concerned. That’s my interpretation of Qualcomm’s statement. Qualcomm denies that Huawei is the net licensor, and it touts the value of its portfolio, so the only reasonable inference here is that Qualcomm suggests Huawei is the net licensee.
But that doesn’t mean Huawei isn’t receiving money on the bottom line. There’s also this patent purchase part.
Patent transfers obfuscate licensing terms. That doesn’t mean to say that those are sham deals. There probably wsere good reasons for which Philips wanted to shore up its 5G portfolio, and OPPO has sold 5G patents to others, such as Toyota, because it has so many. Similarly, Huawei may hold a lot of patents in some fields in which Qualcomm wanted to own more and Huawei was prepared to divest. But it’s rather likely that the patent transfer components of those deals made it a lot easier to agree on something that both parties were able to commercially accept with a view to the bilateral deal at hand and any future FRAND disputes involving one party or the other.
License deals are sometimes also combined with purchases of goods or services
Another impediment to comparability lies in commercial relationships that go beyond patents. For example, the 2017 Nokia-Apple settlement involved more than just patent licenses. Apple was a reseller of some Nokia products and a business-to-business customer of others.
With such deals it’s possible that the agreements formally state separate numbers: a purchase price for certain goods and services, and a license fee. But if there is a dependency such as a condition precedent, you’ll never know whether someone agreed to overpay for products in order to pay a lower royalty rate, or to massively discount products in order to increase the license fee.
The key takeaway from the Federal Trade Commission v. Qualcomm decision by the United States Court of Appeals for the Ninth Circuit is that this is simply in the discretion of negotiating parties. The FTC (and not only the FTC) had argued that Qualcomm’s “No License, No Chips” policy inflated royalty payments by tying chipset sales to license agreements. In other words, they said the chipsets were sold at a nominally lower rate because Qualcomm wanted to be able to attribute a greater portion of the overall transaction value to the license. The district court took issue with it, but the Ninth Circuit said this was not an antitrust problem.
The problem between Huawei and Qualcomm presumably was that due to U.S. sanctions (which Qualcomm actually lobbied against), Qualcomm can’t sell anything to Huawei at the moment. But Huawei can sell patents to Qualcomm.
Judicial FRAND rate-setting is not getting easier
The trend toward “license-plus” agreements is good for industry because it often enables the parties to a negotiation to overcome an impasse.
It’s bad news for judicial or quasi-judicial FRAND determinations because the number of directly comparable license agreements shrinks. Ever more often, the intermediate step of trying to unpack a license agreement is involved, and that task, in and of itself, can be a lot of work at and after trial as well as on appeal.
There’s no reason to distrust either Huawei or Qualcomm. Let’s wait and see if their agreement gets discussed in future court proceedings and rulings. And then the complexity and inevitable subjectivity of the unpacking of license-plus agreements will be easy to see. If you ask three different judges to decide on unpacking, chances are that no two of them will agree. Much less all three.
As that late-1970s song goes: “Everyone’s a winner, baby, that’s the truth.” Here, it means everyone can legitimately claim to be a winner. That’s the truth.