Chinese chip designer Loongson Technology has shipped more than 1 million units of its flagship 3A6000 desktop processor, marking a milestone for China’s efforts to build a self-sufficient semiconductor industry and move domestic central processing units beyond basic usability toward broader commercial adoption. The announcement was reported by the South China Morning Post on May 13.
The 3A6000 processor was built on Loongson’s self-developed LoongArch instruction set architecture, proprietary IP cores, and a domestic manufacturing supply chain. The company has said the chip’s performance is broadly comparable to Intel desktop processors released around 2020, a benchmark that positions it as a credible alternative for government and enterprise users who do not require cutting-edge performance but do require supply chain independence.
The XinChuang Tailwind
Since its launch in November 2023, the 3A6000 has been adopted in China’s government-backed IT replacement program known as “XinChuang”, which mandates the substitution of foreign technology in sensitive public-sector systems with domestic alternatives. Deployments have included education, taxation, and other public-sector systems. Loongson founder Hu Weiwu told investors that 2025 to 2027 would represent a concentrated procurement period for XinChuang projects, with 2026 likely to become a particularly strong year for orders.
The XinChuang program has been a critical enabler for Loongson’s commercial trajectory. The company reported revenue of 635.3 million yuan ($93.4 million) in 2025, up 26% from a year earlier, while its net loss narrowed to 455.1 million yuan from 625.3 million yuan in 2024. Its gross margin rose to 47%, reflecting stronger demand from its XinChuang projects. The company remains loss-making, but the trend lines are moving in the right direction.
Founded in 2001 as a research project under the Chinese Academy of Sciences and then spun off in 2010, Loongson became the first CPU-focused company listed on Shanghai’s Star market in 2022. It remains one of China’s most prominent efforts to build a domestic alternative to Intel and AMD. Loongson founder Hu Weiwu has described the company’s culture in distinctly ideological terms, he is known for frequently wearing Mao suits and Mao badges and displaying Mao quotations on office walls, but the commercial ambition is pragmatic: compete on price-performance in a market where foreign-brand alternatives are becoming increasingly expensive.
Moving Beyond Policy Procurement
Loongson is now seeking to expand beyond policy-driven procurement and compete in the broader commercial market. At an investor meeting in April, Hu said more Chinese customers were testing the company’s chips as foreign-brand alternatives became increasingly expensive, a shift driven partly by tariffs and partly by the broader geopolitical pressure on Chinese companies to reduce their dependence on US-origin technology.
Hu also said the company had completed a strategic shift in manufacturing, moving to domestic production lines. He added that Loongson faced relatively limited cost pressure because its processor IP cores were developed in-house, giving it more control over its cost structure than companies that license architecture from foreign IP holders.
After years of being overshadowed by graphics processing units during the AI boom, CPUs are once again drawing attention as companies race to optimize AI infrastructure costs and reduce dependence on Nvidia. The 1 million unit milestone for the 3A6000 arrives at a moment when that attention is intensifying.
A Broader Localization Push
The 3A6000 milestone is part of a wider effort to close gaps across China’s semiconductor supply chain. On the same day as the Loongson announcement, a research team led by the Shanghai Artificial Intelligence Laboratory said it had used an “AI decision-making plus automated synthesis” system to achieve stable production of high-end KrF photoresist resin, a critical chipmaking material long dominated by a small group of overseas suppliers. The announcement highlights how China is attempting to address vulnerabilities not just at the chip design level but across the entire upstream materials and equipment ecosystem.
China has set a target of 80% chip self-sufficiency by 2030, a goal that analysts widely regard as ambitious given the technology gaps that persist at the leading edge. But the Loongson milestone illustrates that progress is being made in the segments of the market where domestic alternatives are most viable, mature-node, general-purpose computing for government and enterprise use cases where performance requirements are modest and supply chain security is paramount.
The 1 million unit figure is modest by global CPU standards, Intel ships hundreds of millions of units annually, but it represents a meaningful step for a domestic industry that was producing essentially no commercially viable alternatives to foreign CPUs a decade ago. For the XinChuang program and for China’s broader technology independence ambitions, the 3A6000’s trajectory is a data point worth watching.
The LoongArch Ecosystem Challenge
Loongson’s long-term commercial ambitions depend not just on the chip itself but on the software ecosystem built around LoongArch, its proprietary instruction set architecture. ISA transitions are notoriously difficult: the value of any processor architecture is determined as much by the software that runs on it as by the hardware’s raw performance. Intel’s x86 and Arm’s architecture dominate the market in part because decades of software development have been optimized for them.
Loongson has made progress on this front. The LoongArch architecture has been accepted into the mainline Linux kernel, meaning that Linux distributions can now run natively on LoongArch hardware without requiring custom patches. A growing number of open-source applications have been ported to the platform. But the ecosystem remains thin compared to x86 and Arm, and enterprise customers evaluating Loongson hardware for commercial deployments must weigh the software compatibility risk alongside the supply chain security benefits.
Hu has acknowledged the challenge. At the April investor meeting, he said the company’s strategy is to focus first on use cases where the software stack is relatively simple and well-defined — government administration, taxation, education — and to expand into more complex enterprise environments as the ecosystem matures. That sequencing aligns with the XinChuang program’s own priorities, which have historically focused on replacing foreign hardware in government and public-sector systems before tackling more demanding commercial applications.
The competitive landscape is also evolving. Phytium Technology, another Chinese CPU designer, is pursuing a similar strategy with its Arm-based processors, which benefit from the broader Arm software ecosystem even as they use domestically designed cores. Hygon, which produces x86-compatible processors based on licensed AMD technology, offers a different trade-off: better software compatibility but a more complicated IP situation given US export restrictions on the original AMD license. Loongson’s fully proprietary approach is the most independent of the three but also the most demanding in terms of ecosystem development.