China Certifies Nine Domestic AI Processors for State Procurement Under New Anke Security Framework

China Advances Domestic AI Chip Certification Through Anke Framework

China has taken a significant step toward bolstering its domestic AI semiconductor industry by certifying nine homegrown AI processors for state procurement under a newly established “AI training and inference chips” category within the Anke security certification framework. This development, reported by Tom’s Hardware, marks a critical milestone in Beijing’s ongoing efforts to secure key technology supply chains and reduce reliance on foreign suppliers amid intensifying geopolitical tensions.

The certification, valid for three years, includes flagship AI chips from major Chinese technology players such as Huawei’s Ascend 310 and 910, Alibaba’s T-Head Zhenwu M530 and M890, as well as processors from Biren, Hygon, Iluvatar CoreX, MetaX, and Moore Threads. Notably absent from the list are AI chips from Cambricon and Baidu-backed Kunlunxin, two prominent names in China’s AI chip ecosystem.

Understanding Anke Certification vs. Xinchuang Procurement Replacement

To grasp the full strategic significance of this certification, it is essential to differentiate between the Anke and Xinchuang initiatives, both of which play important but distinct roles in China’s technology self-reliance agenda.

Anke (安克) refers to a security certification framework focused on evaluating and verifying the security attributes of information technology products, including chips. The certification process scrutinizes chips for compliance with national cybersecurity standards, ensuring that products meet rigorous requirements for security, reliability, and trustworthiness before being eligible for government and critical infrastructure procurement. The new AI chip category within Anke explicitly targets processors designed for AI model training and inference workloads, reflecting the growing importance of AI in national strategic sectors.

In contrast, Xinchuang (新创), also known as the “procurement replacement” program, is a broader government initiative aimed at substituting foreign hardware and software products with trusted domestic alternatives to build a fully indigenous supply chain. Xinchuang focuses on procurement policy and ecosystem development, encouraging government agencies and state-owned enterprises to prioritize Chinese-designed solutions. While Xinchuang promotes domestic adoption, Anke certification ensures that these domestic products meet strict security benchmarks.

The Anke certification of these nine AI processors thus represents a crucial security stamp that enhances their eligibility and attractiveness for state procurement under Xinchuang’s replacement policies. It signals Beijing’s confidence that these chips can safely underpin critical AI applications in sensitive sectors such as defense, finance, telecommunications, and public administration.

China’s Growing Share of AI Server Shipments

The certification comes at a time when Chinese AI chip makers have gained substantial market traction domestically. In 2025, Chinese firms accounted for approximately 41% of local AI server shipments, translating to around 1.65 million GPUs deployed across data centers nationwide. Huawei alone shipped about 812,000 units, underscoring its dominant position in the Chinese AI infrastructure landscape.

This growing market share is a testament to years of strategic investment in semiconductor design, manufacturing, and ecosystem cultivation. It also reflects Beijing’s aggressive push to cultivate a robust domestic AI hardware industry as part of its broader digital economy ambitions and national security imperatives.

The Nine Certified AI Processors: Players and Capabilities

The nine AI processors certified under the Anke framework represent a spectrum of design philosophies and target applications within China’s rapidly evolving AI chip market.

• Huawei Ascend 310 and 910: Huawei’s Ascend series has been a flagship in China’s AI chip race, with the Ascend 910 targeting high-performance AI training workloads, while the Ascend 310 is optimized for edge inference applications. Both chips leverage Huawei’s Da Vinci architecture and have been widely integrated into Huawei’s AI cloud solutions.

• Alibaba T-Head Zhenwu M530 and M890: Alibaba’s semiconductor arm T-Head specializes in AI inference and training processors designed to power the company’s cloud infrastructure as well as commercial applications. The Zhenwu series reflects Alibaba’s commitment to AI acceleration and cloud optimization.

• Biren: A newer entrant focused on high-performance AI training chips, Biren has been gaining attention for its cutting-edge architecture and performance metrics tailored to large-scale AI model training.

• Hygon: Originally a joint venture with AMD, Hygon focuses on server CPUs and AI acceleration chips designed for enterprise and government cloud services.

• Iluvatar CoreX: Iluvatar specializes in AI acceleration products optimized for both training and inference, with a focus on energy efficiency and scalability.

• MetaX: A rising AI chip startup, MetaX targets AI inference integration, especially in edge computing and real-time applications.

• Moore Threads: Moore Threads is developing AI GPUs and accelerators aiming to compete with both domestic and international alternatives in the AI server market.

The absence of Cambricon and Baidu-backed Kunlunxin from the certified list raises questions about either the certification compliance or strategic positioning of these firms. Both companies are recognized leaders in China’s AI chip space, with Cambricon known for its long-standing AI chip development and Kunlunxin powering Baidu’s AI cloud infrastructure. Their exclusion may indicate ongoing certification processes, strategic decisions to focus on other markets, or challenges meeting the stringent Anke requirements.

Strategic Implications for China’s AI Chip Ecosystem

The formal certification of these nine AI processors under the Anke framework has several far-reaching implications for China’s AI chip industry and broader national technology strategies.

First, the Anke certification enhances supply chain security by ensuring that only verified, secure domestic AI chips are procured for sensitive and government applications. This reduces systemic risks associated with imported components, which could be subject to supply disruptions or embedded vulnerabilities amid geopolitical tensions and export controls.

Second, the certification is likely to accelerate the adoption of domestic AI chips in China’s government, defense, and critical infrastructure sectors. Given the procurement preferences under Xinchuang, certified processors stand to benefit from preferential policies, increasing their market penetration and revenue opportunities.

Third, the certification list reflects a relatively diverse competitive landscape, with established giants like Huawei and Alibaba coexisting alongside emerging startups such as Biren and MetaX. This diversity encourages innovation, as companies compete on architecture design, performance, and power efficiency to capture state procurement projects.

On the other hand, the absence of notable players like Cambricon and Kunlunxin highlights that Anke certification may serve as a significant market barrier, potentially excluding otherwise capable processors that have not met security or compliance benchmarks. This underscores the importance for AI chip firms to invest not only in technical performance but also in security and standards alignment.

Broader Industry Context and Future Outlook

China’s move to institutionalize AI chip certification within the Anke framework reflects a growing global trend where countries seek to secure AI hardware through rigorous security protocols as AI becomes central to national competitiveness and security. This development comes amid an intensifying U.S.-China technology rivalry, where semiconductor supply chains are increasingly bifurcated.

For China, cultivating a trusted pool of domestic AI processors is critical to sustaining its ambitious AI ecosystem, which spans cloud computing, autonomous vehicles, smart cities, and defense applications. The certified chips will underpin next-generation AI models and services, supporting China’s goal to lead globally in AI innovation.

Looking ahead, the interplay between Anke certification and Xinchuang procurement policies will shape the trajectory of China’s AI chip industry. Firms that secure certification will gain privileged access to government projects, while those lagging may need to innovate or pivot to non-governmental markets.

As China continues to expand its AI infrastructure, the certification of these nine processors marks a pivotal moment in aligning technological capability with national security imperatives. It also signals a maturing domestic AI chip market ready to challenge international incumbents and reduce dependence on foreign technology, a critical factor in the evolving global technology landscape.

For additional insights on China’s AI chip strategy and certification developments, see the detailed coverage by Tom’s Hardware and South China Morning Post.