Abstract:
In the legal dispute between Qualcomm and Arm over the licensing of processor architecture, the commercial competitiveness of RISC-V, an open instruction set architecture, became the focus of the court discussion. Judge Maryellen Noreika of the U.S. Federal Court in Delaware questioned the actual application of RISC-V during the trial, pointing out that this architecture currently has limited implementation in the enterprise data center market.
Relevant testimony also reflects that although RISC-V has long been regarded as an important alternative to Arm and x86, it still faces multiple challenges such as ecosystem, software compatibility and commercialization progress in order to truly enter the high-performance server market.
The background of this controversy can be traced back to 2021. That year, Qualcomm acquired Nuvia, a start-up that specializes in Arm architecture processor core design, for about US$1.4 billion, and then used related technologies to accelerate the development of its own Oryon CPU core. Arm believes that the architecture licensing agreement previously held by Nuvia cannot be automatically transferred to Qualcomm without permission, so Qualcomm needs to re-obtain a license or sign a new licensing agreement.
The two sides have been involved in a legal dispute that has lasted for many years. Previously, in a related case heard in the Delaware Federal Court, Judge Noreka supported Qualcomm’s position on key disputes. One important factor is that Qualcomm’s own architecture licensing agreement with Arm is still valid and the term extends to 2033.
However, the dispute between Qualcomm and Arm does not end here. In 2024, Qualcomm filed another lawsuit against Arm, accusing Arm of violating relevant obligations in the architecture licensing agreement and technology licensing agreement between the two parties.
The accusations raised by Qualcomm involve multiple aspects, including Arm's failure to provide architectural verification tools, software patches, chip testing and technical support as agreed in the contract; alleged violations of most-favored-nation pricing clauses; and Arm's alleged interference with potential chip cooperation negotiations between Qualcomm and Meta by leaking or allowing the disclosure of a letter threatening to terminate Qualcomm's architecture authorization. Qualcomm claimed that the potential damage caused by this incident reached hundreds of millions of dollars. In addition, Qualcomm also objected to the licensing fees involved in Arm’s next-generation v10 architecture, believing that the relevant fee requirements are too high.
These allegations still need to be examined separately in legal proceedings, and the relevant conduct cannot be deemed to have been proven based on one party's statements alone.
In the latest round of court proceedings, Arm called economist Timothy Simcoe as a rebuttal witness. Simcoe tried to explain that although Arm has a strong market position in the field of processor architecture licensing, long-term competitive pressure from Intel's x86 architecture and RISC-V can still constrain Arm's pricing and business practices.
However, Judge Noreka later questioned RISC-V’s realistic competitiveness.
According to Wccftech’s report on the court hearing, when Simcoe talked about the competitive pressure that RISC-V may pose, the judge asked whether this architecture has actual availability and expressed doubts about its application progress in the enterprise-level data center market. Faced with related issues, Simcoe mentioned Qualcomm’s investment in RISC-V related companies such as Ventana, as well as the related processor design work being carried out by Chinese companies.
But in the judge’s view, there is still a clear difference between potential technical routes and commercial competition that has already formed a large scale. Simcoe finally admitted that, except for the early designs being promoted by companies such as China's Alibaba, the application of RISC-V in the field of enterprise-level data centers is still in a relatively early stage.
There is another detail worthy of attention during the trial. Judge Noreka further asked Simcoe whether the competition theory he proposed means that even if Arm increases licensing fees or takes other practices that dissatisfy customers, it may prompt customers to increase investment in other processor architectures, thus forming a long-term constraint on Arm. Simcoe acknowledged that logic as part of his argument, and the judge ended the questioning.
It should be noted that this court discussion does not mean that the court has officially ruled that RISC-V lacks commercial value, nor can it be concluded that RISC-V can never become a competitor to Arm or x86. The judge’s questions mainly focused on the current market reality and whether competition constraints are effective enough, rather than making a final judgment on the technical potential of the entire architecture.
In fact, RISC-V has attracted attention from many countries and regions in recent years. Unlike Arm and x86, RISC-V is an open instruction set architecture specification that adopts a reduced instruction set computing concept. Enterprises can design their own processors according to relevant specifications without having to rely on a certain company's licensing arrangement just to use the instruction set itself, as is the case with proprietary architectures.
This openness has obvious strategic appeal. For companies that want to reduce intellectual property licensing costs, reduce dependence on a single supplier, or establish an independent processor industry chain, RISC-V provides another technical route. Companies and research institutions in China, Europe and other regions are advancing related projects, with application areas covering microcontrollers, embedded devices, Internet of Things, industrial control and artificial intelligence computing.
But opening the instruction set does not mean that it is easy to create products that can compete with mature commercial processors. The instruction set is only part of the processor technology system. To develop a truly competitive CPU, companies also need to invest a lot of resources in research and development in microarchitecture design, branch prediction, out-of-order execution, cache systems, memory management, multi-core interconnection, and power consumption control.
For server processors, the challenge is especially clear. Data center customers not only focus on the computing performance of the CPU itself, but also examine memory bandwidth, input and output capabilities, virtualization support, reliability, energy consumption, long-term supply capabilities, and the total cost of ownership of the entire platform. At the same time, the processor must also form a stable cooperative relationship with the operating system, compiler, database, virtual machine, container platform and various enterprise applications.
This is also one of the main obstacles facing RISC-V. After decades of development, Arm and x86 have accumulated a huge software ecosystem, mature development tools, rich system verification experience, and a broad commercial customer base. If an enterprise wants to migrate an existing data center to another architecture, it not only needs to purchase new hardware, but also may need to recompile or port software, verify application compatibility, and bear system migration and maintenance costs.
So even if RISC-V shows good performance or energy efficiency in some tests, that doesn't mean enterprise customers will adopt it immediately. For data centers that need to run year-round and carry critical services, maturity, stability and vendor support are often as important as peak performance.
The Xuantie C950 that Alibaba is promoting is a representative work of Chinese companies exploring high-performance RISC-V processors.
According to the introduction in the report, Alibaba announced the Xuantie C950 in March 2026 and positioned it in computing scenarios such as edge artificial intelligence. This 64-bit RISC-V processor adopts a 64-core design, the maximum frequency can be extended to 3.20GHz, and multiple clusters are connected through high-speed interconnect technology. Each cluster contains 8 cores and integrates matrix and vector acceleration engines to enhance the processing capabilities of specific artificial intelligence workloads.
These designs show that RISC-V is not limited to simple microcontrollers or low-power embedded devices, and relevant companies are also trying to advance it into more complex computing fields. However, the processor's published specifications, its performance potential for specific applications, and its proven commercial competitiveness in large enterprise data centers are three different issues. To prove that it can compete with mature server CPU platforms, more data on actual performance, software adaptation, reliability and large-scale deployment are needed.
In addition, although the openness of RISC-V lowers some of the barriers to adopting instruction sets, it cannot automatically eliminate the funding, talent and engineering experience requirements for high-performance chip development. Enterprises still need to establish complete chip design capabilities and invest resources in maintaining software tool chains, operating system support and developer ecosystems. For new processor platforms that lack a mature customer base, these efforts often take years to build scale.
Judging from the lawsuit between Qualcomm and Arm, the significance of RISC-V is not limited to whether it can replace existing processors on a large scale. Even if there are no mature enough products to directly replace Arm in the short term, RISC-V may still affect companies' investment decisions on future processor architectures by providing another optional technology route.
When customers believe that an alternative architecture is likely to become available in the future, they may increase relevant R&D investment or seek more favorable terms in licensing negotiations. For architecture licensees like Arm, this potential long-term competitive pressure may itself be of some significance. However, whether alternative technologies can truly form a constraint ultimately depends on product maturity, commercial deployment scale, and the actual cost for customers to switch platforms.
At this stage, Arm and Intel x86 have mature products and software ecosystems in the enterprise server market, while RISC-V is still exploring how to transform from an open technical specification into a high-performance commercial platform that can be widely deployed. The discussion during the trial highlighted this reality gap, but it cannot deny the application value of RISC-V in other markets, nor can it predetermine its future development limit in the server field.
For the RISC-V industry, the real test is no longer just proving that the open instruction set can design a executable processor, but how to continue to launch products with reliable performance, software compatibility, competitive cost, and the long-term trust of enterprise customers. Only when these conditions gradually mature, will RISC-V be able to grow from a potential alternative to a processor platform that can form more direct competitive pressure on Arm and x86.
Comments