Xiaomi’s latest 3nm flagship chip, the Xuanjie O1, has recently been embroiled in a “self-development” controversy. The Arm official website initially referred to it as a “custom chip,” raising questions among netizens about whether Xiaomi relies on Arm’s existing solutions. Xiaomi quickly refuted this, emphasizing that the chip is entirely self-developed, prompting Arm to revise its wording and delete the original article. This incident reflects the struggle for technological discourse power in domestic chips and highlights the challenges of self-developed chips—Lei Jun stated that Xiaomi has invested over 13.5 billion yuan in chip development and will continue to increase investment in the future. Can the Xuanjie O1 truly compete with Apple’s A18 Pro? The answer may lie in the details amidst the controversy.
On May 22, Xiaomi founder Lei Jun grandly announced the Xuanjie O1 chip—Xiaomi’s first 3nm flagship SoC—and claimed its performance rivals that of Apple’s A18 Pro. However, just a few days later, a press release from Arm’s official website plunged the chip into a whirlpool of public opinion.
Initially, the article published by Arm was titled “XRING O1 Custom Silicon from Xiaomi is Powered by the Arm Compute Platform,” which translates to “Xiaomi’s XRING O1 custom chip is supported by the Arm compute platform.” This statement quickly sparked heated discussions in the tech community, with many netizens questioning: Is the Xuanjie O1 merely a ‘semi-custom’ solution based on Arm’s CSS (Compute Subsystem)? How substantial is Xiaomi’s claim of ‘self-development’?
In response to public pressure, Xiaomi urgently replied on the evening of May 26, denying that the Xuanjie O1 is an Arm custom chip and emphasizing that it is entirely designed and developed by Xiaomi’s Xuanjie team. Subsequently, Arm deleted the original article and reissued a revised version, changing “Custom Silicon” to “Xiaomi Self-Developed Chip.” Although the storm has calmed, the underlying issues are worth pondering: What misunderstandings and challenges does the path of ‘self-development’ for domestic chips face?
To understand this controversy, it is essential to clarify the division of labor in the chip R&D industry chain. Currently, global mobile SoC manufacturers (such as Qualcomm, MediaTek, and Apple) all use Arm’s CPU/GPU IP licensing, but each company will conduct secondary design based on its needs, such as adjusting architecture, optimizing energy efficiency, and integrating self-developed modules.
Xiaomi detailed the R&D process of the Xuanjie O1 in its statement:
-
Based on Arm’s standard IP, but not limited to ‘copying’
-
Utilizing Armv9.2 Cortex-X925/A725 CPU and Immortalis GPU’s standard IP licensing, but the core multi-core scheduling, memory access system, and physical layer design are all completed by the Xiaomi team.
-
For example, the CPU’s super-large core frequency reaches 3.9GHz (exceeding Arm’s reference design), relying on Xiaomi’s redesign of 480 types of standard cell libraries, as well as self-developed edge power supply technology and high-speed registers.
Essential differences from Arm’s CSS service
-
Arm CSS (Compute Subsystem) is a “semi-custom” service where customers can directly use Arm’s pre-configured CPU/GPU/interconnect solutions with minimal adaptation.
-
Xiaomi clearly stated that the Xuanjie O1 did not use CSS services, but started from IP licensing and independently completed backend implementation and system-level optimization.
In short, the R&D model of the Xuanjie O1 is closer to Huawei’s Kirin or Apple’s A-series chips, rather than a ‘white-label’ product.
Xiaomi’s experience is not an isolated case. In recent years, any domestic chip claiming ‘self-development’ has been accompanied by controversy. For example:
-
Huawei’s Kirin chips were early criticized for being ‘modified from ARM’s public version’;
-
A certain automotive company’s ‘self-developed chip’ was later revealed to be an integration of off-the-shelf IP.
These controversies stem from three public misconceptions about the chip industry:
-
Misconception 1: ‘Self-development = 100% starting from scratch’
-
In reality, modern chip R&D relies on global industrial chain collaboration. Even Apple’s A-series chips depend on Arm’s instruction set and TSMC’s foundry. True ‘self-development’ is reflected in architectural design, system optimization, and breakthroughs in core technologies.
Misconception 2: ‘Using Arm IP = low technical value’
-
Arm IP licensing comes in different levels, with significant differences between standard IP and architecture licensing (like Apple’s). Xiaomi’s use of standard IP licensing this time, but achieving performance surpassing through deep optimization, is a testament to its technical strength.
Misconception 3: ‘Domestic chips must replace all foreign technologies’
In a globalized division of labor, building everything in-house is unrealistic. The breakthrough path for domestic chips should be ‘first participate, then lead’—just as Xiaomi progressed from the Surge S1 trial to the Xuanjie O1’s 3nm advancement.
Despite the controversies, the release of the Xuanjie O1 is still of milestone significance:
-
Technical aspect
-
It becomes the fourth company globally, after Apple, Samsung, and Huawei, to master 3nm mobile SoC technology.
-
The “2+4+2+2” ten-core architecture design and 3.9GHz ultra-high frequency demonstrate Xiaomi’s accumulation in chip design.
Industrial aspect
-
Lei Jun revealed that the Xuanjie project has invested over 13.5 billion yuan, with a 2024 R&D budget exceeding 6 billion. This long-term investment is a necessary condition for breaking through “bottleneck” technologies.
-
Through self-developed chips, Xiaomi can reduce dependence on suppliers like Qualcomm and enhance product differentiation (such as deep optimization of imaging and AI capabilities).
Public opinion aspect
-
This incident encourages the public to view the definition of ‘self-development’ more rationally and provides public relations experience for other domestic chip manufacturers.
The controversy surrounding the Xuanjie O1 reflects a reality: The growth of domestic high-end chips will inevitably be accompanied by skepticism, as this path is inherently challenging.
-
Short-term challenges: The cost of 3nm technology is extremely high (the single-chip tape-out cost may exceed 1 billion yuan), and Xiaomi must ensure yield and market returns after mass production.
-
Long-term test: Can it gradually shift from IP licensing to architectural innovation (such as self-developed NPU, ISP, and other modules) like Huawei?
Lei Jun stated that Xiaomi’s chip development is a “marathon of at least ten years and 50 billion yuan,” and the Xuanjie O1 is just the starting point. For consumers, it is advisable to adopt a more open mindset towards the progress of domestic technology—after all, no giant’s chip journey is achieved overnight.
The ‘mislabeling’ incident by Arm has become a footnote, but the true value of the Xuanjie O1 will be tested in the future performance and energy efficiency of mobile phones. Regardless of the controversy, Xiaomi’s courage to challenge the 3nm high-end chip market has injected a dose of confidence into the domestic semiconductor industry.