China adds AI chips to secure tech list as US export curbs bite
Synopsis
For the first time, China has added AI training and inference chips to its official Xinchuang secure-technology certification list — a move that effectively mandates domestic alternatives to Nvidia across government and state enterprise AI procurement, directly accelerating the homegrown chip ecosystem amid tightening US export controls.
Key Takeaways
China included AI training and inference chips in its 'secure and reliable' technology assessment framework for the first time on 27 May 2026 .
The certifications were issued by the China Information Technology Security Evaluation Centre and the National Secrecy Science and Technology Evaluation Centre , and are valid for three years .
Approvals function as the de facto procurement catalogue for party and government agencies , central state-owned enterprises , and other state-linked buyers under the Xinchuang initiative.
Beijing 's technology replacement drive is pivoting from legacy CPU and database vendors ( Intel , AMD , Oracle ) toward AI accelerator suppliers, primarily targeting Nvidia .
Domestic beneficiaries include Huawei Technologies , Cambricon Technologies , Kunlunxin , MetaX , Moore Threads , Iluvatar CoreX , Biren Technology , Hygon Information Technology , and Alibaba Group Holding 's Ascend chips.
Successive rounds of US export controls on advanced GPUs have accelerated the urgency of this domestic substitution push.
China has, for the first time, included artificial intelligence chips in its official 'secure and reliable' technology assessment framework, marking a significant expansion of Beijing's state-backed drive to replace Western technology with domestic alternatives. The China Information Technology Security Evaluation Centre and the National Secrecy Science and Technology Evaluation Centre released the latest batch of certifications on Tuesday, 27 May 2026, introducing a new category specifically for 'AI training and inference chips'. The certifications carry a validity period of three years, according to the official announcement.
What the new category means
The two bodies serve as China's authoritative assessment authorities for information security and secrecy-related technologies. Their approvals function, in practice, as the definitive procurement catalogue for party and government agencies, central state-owned enterprises, and other state-linked customers operating under the Xinchuang initiative. Inclusion on this list is widely viewed as a prerequisite for vendors seeking contracts with sensitive government and strategic-sector clients.
Xinchuang — formally translated as information technology application innovation — is Beijing's long-running campaign to purge foreign hardware and software from sensitive and strategically important information systems. The programme has, until now, concentrated heavily on phasing out US suppliers such as Intel and Advanced Micro Devices for central processing units (CPUs), and Oracle for databases.
Why it matters: AI infrastructure now in the crosshairs
The extension of Xinchuang assessments to AI chips signals that Beijing is now systematically targeting Nvidia and other foreign suppliers of AI computing power — not merely legacy enterprise hardware. This pivot reflects how China's technology replacement drive has evolved alongside the rapid buildout of domestic AI infrastructure. Domestic chipmakers including Huawei Technologies, Cambricon Technologies, Kunlunxin, MetaX, Moore Threads, Iluvatar CoreX, Anke, Biren Technology, Hygon Information Technology, and Alibaba Group Holding's Ascend-branded chips are among those now positioned to benefit from state procurement mandates.
The competitive backdrop: US export controls accelerate the shift
The move has gained urgency following successive rounds of US export controls that have progressively restricted Chinese access to advanced graphics processing units (GPUs). Each new round of restrictions has effectively closed off procurement channels for frontier AI accelerators from American vendors, pushing state-linked buyers toward homegrown alternatives — whether or not those alternatives match the performance benchmarks of restricted products. The formal inclusion of AI chips in the assessment framework now provides institutional backing for that substitution at scale.
What's next
With the certification framework now extended to AI training and inference chips, domestic vendors that secure approvals will gain a structural procurement advantage across the vast network of government bodies and state enterprises bound by Xinchuang mandates. The three-year certification window means the current cohort of approved chips will shape public-sector AI infrastructure investment through at least 2029. Analysts and industry observers will be watching closely to see which domestic chipmakers achieve certification first — and whether their products can meet the computational demands of large-scale AI workloads in production environments.
Point of View
Locking in domestic vendors through procurement mandates before the performance gap with Nvidia can be fully measured. What mainstream coverage underplays is the compounding effect — once state enterprises standardise on certified domestic chips, the switching costs rise sharply, creating a captive market that insulates Chinese AI chipmakers from competitive pressure even if their products underperform. This mirrors the earlier Xinchuang playbook for CPUs, where domestic vendors like Hygon and Kunlunxin secured government footholds that commercial benchmarks alone would not have justified. The longer-term risk for global AI supply chains is a bifurcated hardware stack: one optimised for Western hyperscalers, another entrenched across a vast state-directed economy — with diminishing interoperability between the two.
NationPress
12 Jul 2026
Frequently Asked Questions
What is China's Xinchuang initiative and why does it matter for AI chips?
Xinchuang, or information technology application innovation, is Beijing's state-backed campaign to replace foreign hardware and software in sensitive government and strategic-sector systems with domestic alternatives. Its certification list now includes AI training and inference chips for the first time, meaning domestic AI chipmakers can qualify for procurement contracts across party agencies, central state-owned enterprises, and other state-linked buyers.
Which Chinese AI chip companies could benefit from the new certification?
Domestic chipmakers positioned to benefit include Huawei Technologies, Cambricon Technologies, Kunlunxin, MetaX, Moore Threads, Iluvatar CoreX, Anke, Biren Technology, Hygon Information Technology, and Alibaba Group Holding's Ascend chips. Vendors that secure formal certification will gain a structural procurement advantage across the government and state enterprise sector.
How do US export controls connect to China's AI chip certification move?
Successive rounds of US export controls have restricted Chinese access to advanced GPUs from American vendors, most notably Nvidia, accelerating Beijing's urgency to develop and mandate domestic alternatives. The new Xinchuang category for AI chips formalises that substitution at an institutional level, providing state procurement backing for homegrown accelerators.
How long are the new AI chip certifications valid?
The certifications issued by the China Information Technology Security Evaluation Centre and the National Secrecy Science and Technology Evaluation Centre on 27 May 2026 are valid for three years, according to the official announcement. This means the current cohort of approved chips will shape public-sector AI infrastructure investment through at least 2029.
What does this mean for Nvidia's position in China's government and enterprise market?
The new certification framework effectively creates an institutional barrier to Nvidia and other foreign AI chip suppliers in China's state-linked procurement ecosystem. Combined with existing US export restrictions that already limit which Nvidia products can be sold in China, the Xinchuang mandate further narrows the addressable market for foreign AI accelerators among government bodies and central state-owned enterprises.
