Issue 5 | November 20, 2025
This article contains 3222 words, and reading it will take approximately 10 minutes.
On September 30, 2025, the Delaware court in the United States issued a first-instance judgment in the licensing dispute case between ARM and Qualcomm.
1. Background: Licensing, Acquisition, and Litigation
In 2021, Qualcomm announced its acquisition of chip startup NUVIA for $1.4 billion, with NUVIA aiming to develop high-performance, low-power ARM server processors to compete with Intel and AMD.
Developing ARM architecture chips requires obtaining a license from ARM, which mainly comes in two forms: Technology License Agreement (TLA), which authorizes the use of ARM’s existing processor cores; and Architecture License Agreement (ALA), which authorizes the design of cores based on the ARM instruction set architecture.
For better understanding, this can be likened to copyright: TLA is similar to reproduction rights, while ALA is akin to adaptation rights. Both NUVIA and Qualcomm hold these two types of licenses, and neither can transfer them without ARM’s approval.
After the acquisition, ARM claimed that Qualcomm used NUVIA’s designs in PC processors. Since NUVIA’s license was limited to server processors, Qualcomm needed to obtain a new license to use the relevant technology in PC processors.
In August 2022, after failing to reach an agreement, ARM sued Qualcomm and NUVIA in the Delaware District Court for infringement, seeking to destroy the designs and halt development.
In December 2024, the jury ruled that Qualcomm did not violate the licensing agreement, but did not make a determination on whether NUVIA breached the agreement.
On September 30, 2025, the court issued a first-instance judgment: neither Qualcomm nor NUVIA violated the ALA agreement, dismissing ARM’s claims.
Although ARM’s main lawsuit has concluded, Qualcomm’s counterclaims, including allegations of ARM breaching the agreement, interfering with customer relationships, and engaging in anti-competitive behavior, are still under review and are expected to go to trial in March 2026.
2. Core Disputes and Judgment Logic
ARM’s claim hinges on proving that Qualcomm’s PC processors indeed used NUVIA’s designs. Qualcomm, on the other hand, provided evidence that its designs were independently completed by its engineers based on its ALA license. The jury found Qualcomm’s evidence to be more convincing, and the foundational facts of ARM’s allegations were not established.
ARM also claimed that NUVIA used the server CPU license for notebook CPUs, constituting infringement and undermining its licensing ecosystem. However, the court pointed out that loss is a necessary element of infringement, and ARM failed to prove any loss, as licensing revenue continued to grow during the alleged infringement period.
For these reasons, the court dismissed ARM’s lawsuit.
3. Unresolved Issues: The Rights Boundaries of IP Cores Remain Unclear
The dispute between ARM and Qualcomm centers on the unclear rights boundaries of “IP cores.” IP cores have dual characteristics: they are critically important yet poorly defined.
They are important because IP cores are “ready-made design modules” (such as power management and input/output units) that can significantly enhance chip design efficiency, enabling small teams to complete complex designs, and are nearly unavoidable in actual chip design.
They are vague because IP cores are not defined intellectual property like patents, copyrights, or layout designs, but rather a “bundle of rights” that includes various forms of intellectual property. This composite nature leads to legal boundaries that are far less clear than traditional IP.
Clear rights lead to clear boundaries. For example, patents are limited by the principle of “exhaustion of rights”; once a product is legally sold, the rights holder cannot interfere with its subsequent use or resale. However, IP cores lack such statutory principles and cannot be completely disassembled into traditional IP for protection; their boundaries are more reliant on commercial agreements.
This case actually involves three issues regarding IP core licensing: Are the application scenario restrictions appropriate? After design completion, does the implementer still need a license? To what extent should the implementer’s license be restricted? Compared to the case conclusion, these three questions are more important for clarifying the boundaries of IP cores.
First, default restrictions on application scenarios may be unfair and hinder innovation
In licensing negotiations, both parties certainly have the freedom to define application scenarios, but determining that application scenarios constitute implied boundaries of the license clearly requires strict conditions. From a contract law perspective, in cases where the contract is not specified or is ambiguous, there are various interpretative methods based on purpose, transaction habits, and contracting background to reconstruct the licensing boundaries. Most civil law principles in various countries generally recognize these interpretative methods.
In this case, the licensing agreement did not explicitly limit application scenarios. ARM attempted to interpret the contract from a purpose perspective, claiming that NUVIA’s purpose in accepting the license was to develop server CPUs, so developing personal computer CPUs exceeded the licensing scope.
However, from a business logic standpoint, the fundamental demand of the licensee in contracting is to achieve profitability through chip development, and adjusting the R&D direction based on market dynamics is a routine operation in business, clearly aligning with the basic laws of commercial activity.
Moreover, the basic purpose of the licensor in contracting is to obtain licensing fees. ARM’s licensing agreement includes per-unit royalties, which only generate royalties when products are brought to market; a “steadfast” R&D approach that does not consider actual demand changes is not beneficial for product production and royalty increases.
If adjusting R&D requires obtaining a new license, this will undoubtedly increase R&D costs, which is detrimental to the licensee. But does ignoring the market and not adjusting direction through blind development benefit the licensor? Does it promote innovation? Does it benefit the allocation of social resources? The answer is clearly no.
Additionally, licensors often have more experience in negotiations than licensees and are also the drafters of the licensing agreement terms. In cases where the licensing agreement does not explicitly define application scenarios, interpreting the terms in a way that favors the licensor or requiring the licensee to re-accept the license would be unfair.
If both parties expressed an intention to limit application scenarios during the contracting process, interpreting it as a licensing boundary would be understandable. However, merely using the business that the licensee was engaged in at the time of signing the agreement as the licensing boundary is clearly unfair and detrimental to innovation.
It should be noted that licensing for IP cores has its particularities, generally having specific demands, and pricing often correlates with demand. Therefore, on one hand, it should not prohibit the agreement of application scenarios, while on the other hand, in the absence of an agreement, unless in special circumstances, there is no need to limit the licensing boundary based on application.
Second, after a new design is completed, implementation should still be restricted
This issue is relatively straightforward. Assuming A designs and then commissions B to produce, if the new design includes the original code, layout, or trade secrets of the IP core, B needs to obtain a license unless A’s license has extended to manufacturing.
This is significantly different from patent licensing. Patents are public technical solutions, and it lacks reasonableness for the rights holder to continue controlling the technical solution after granting a license for compensation. However, many technical solutions in IP cores are strictly confidential, and in some cases, certain hard IP cores are not even visible to the licensee.
In this case, producer B obtaining a licensing authorization is necessary, even for confidentiality reasons. Of course, royalties can be borne by designer A, which is not a significant issue in practice. Additionally, the license granted to the implementer should not unduly restrict the designer’s freedom to choose the implementer.
Third, if the implementer already has a license, are they still subject to application scenario restrictions beyond the license?
In this case, ARM’s expression is perplexing: both Qualcomm and NUVIA have full licenses, but Qualcomm needs to obtain a new license to use NUVIA’s design, which sounds somewhat unreasonable.
However, expressing it from another angle reveals that the issues may not be so black and white.
Using copyright as an analogy, if ARM is the rights holder of a literary work, both Qualcomm and NUVIA have licenses for reproduction and adaptation, but Qualcomm is adapting it into an opera while NUVIA is adapting it into a movie. If NUVIA writes the script and commissions Qualcomm to produce, does Qualcomm need a license? At this point, Qualcomm clearly needs to accept a license.
This is not a perfect analogy; IP cores are not independent copyrights, and ALA architecture licensing is not simply adaptation. But this analogy at least reflects that ARM’s lawsuit is not entirely performative; in specific circumstances, it is indeed necessary to consider the actual scope of the license.
Moreover, the above analogy is relatively extreme; if NUVIA exercises reproduction rights to print books and then commissions Qualcomm to sell, Qualcomm clearly does not need to obtain a license.
Using literary copyright to compare with IP cores is inherently inappropriate; artistic pursuits seek diversity, while technology pursues advancement. Requiring every author to create a novel from scratch benefits society, but requiring every engineer to reinvent the wheel is certainly nonsensical.
In fact, the third question is a continuation of the first; if, in the absence of an agreement, application scenarios cannot limit the licensing boundary, then the implementer’s license should also not be restricted by application scenarios. It is hoped that the court will provide clear opinions in the subsequent Qualcomm counterclaims.
4. ARM’s Litigation Motives and Market Impact
It should be noted that the fact of the acquisition in this case is merely a trigger for the case and has limited impact on legal analysis. Qualcomm’s acquisition of NUVIA was a reverse triangular merger, meaning it merged with NUVIA through a wholly-owned subsidiary to obtain all equity, while NUVIA’s original name, contracts, debts, and intellectual property remained unchanged.
ARM’s TLA and ALA licensing fee rates differ significantly. TLA directly using ARM core has a higher licensing fee, while designing cores independently through ALA has a lower fee rate. ARM certainly hopes for higher returns, so it seeks to renegotiate licensing fees with Qualcomm through litigation, or even compel Qualcomm to abandon existing designs and redesign.
Additionally, licensing fees and royalties for small companies and large companies clearly differ; the model of small companies developing and then being acquired by large companies may impact ARM’s revenue.
It is uncertain whether ARM will subsequently add application scenarios or re-pricing clauses after acquisition to contracts, but this will increase the cost of licensing negotiations. Meanwhile, ARM holds over 40% of the global IP core licensing market; if the market is narrowed to instruction set architecture licensing or processor core licensing, the share will be even higher.
With such a high market share, whether restrictions on licensing will weaken innovation and lead to monopoly risks is a question ARM needs to consider.
Furthermore, ARM is gradually transitioning from providing IP to providing chips. For example, the joint project “Stargate” between SoftBank and OpenAI plans to invest in 10 gigawatts of computing power, and CPU suppliers clearly benefit more than IP suppliers.
If ARM becomes a competitor in the chip market, imposing restrictions on the application scenarios of IP cores in licensing or differentiating pricing based on different application scenarios will clearly provide stronger incentives but also face more constraints.
Jin Qiang
Yuanhe Law Firm
Partner
Main Practice Areas:Intellectual Property and Competition Law Dispute Resolution.
Previous Issues Review
|
01 Yuanhe IC Observation: Four Sentences Explaining the Freezing of Anshi Semiconductor’s Intellectual Property |
|
02 Yuanhe IC Observation: The Yangtze Memory and Micron Defense Battle |
|
03 Yuanhe IC Observation: Semiconductor Case Biweekly Discussion |
|
04 Yuanhe IC Observation: Recent Case Review in the Semiconductor Field |
About the Firm
The intellectual property team at Yuanhe Law Firm has long provided comprehensive intellectual property legal services to globally renowned enterprises, research institutions, and technology and cultural innovation companies, successfully representing numerous high-profile intellectual property cases both domestically and internationally, and participating in many complex intellectual property transactions, demonstrating outstanding professional competence and rich practical experience.
Yuanhe has established long-term and stable cooperative relationships with many intellectual property agencies and law firms worldwide, forming a one-stop service system centered on international patent applications, international trademark registrations, and international intellectual property dispute resolution, capable of providing clients with rapid response and internationalized services for their global intellectual property legal needs.
If you have any needs, please feel free to contact us:：
Contact Address：
27th Floor, Fortune Financial Center, 5 Dongsanhuan Zhonglu, Chaoyang District, Beijing, China, 100020
电话：+86-10-5733-2388
Fax:：+86-10-5733-2399
Website:：www.yuanhepartners.com
Images in this article are sourced from: 588ku
Disclaimer:
This article does not represent any form of legal opinion or advice issued by Beijing Yuanhe Law Firm or its lawyers. For reprints or citations, please indicate the source as the public account “Yuanhe Legal Research Institute” and the author’s name at the beginning of the article.
– THE END –