Research

Chinese AI Chips Landscape 2026

Chinese AI chip companies 2026: Huawei Ascend 910C (600K units), Cambricon Siyuan (500K target), Biren, Moore Threads, MetaX. Government-approved list excludes NVIDIA.

By Ramanath, CTO & Co-Founder at Presenc AI · Last updated: May 2026

What this is

Chinese AI chip vendors completed Day-0 adaptation of DeepSeek V4 in 2026, marking the transition from "lagging" to "simultaneous deployment" relative to NVIDIA. This page is a 2026-05-15 snapshot of the Chinese AI chip landscape.

Chinese AI Chip Vendors at a Glance

VendorFlagship chip (2026)2026 output targetStatus
Huawei (HiSilicon)Ascend 910C; Ascend 950PR (Q1)600K Ascend 910C + 1.6M total diesGovernment-approved supplier
CambriconSiyuan 590 / 690500K accelerators (300K Siyuan)Government-approved; targets 3x output
Hygon InformationDCU seriesn/a publishedDay-0 DeepSeek V4 adaptation
BirenBR100 / BR104 seriesn/a publishedHKEX IPO 2026 (filed)
Moore ThreadsMTT S5000n/a publishedDay-0 DeepSeek V4 inference (FlagOS)
MetaXC-series (C500, C600)n/a publishedGrowing
SMIC (foundry)7nm process (production)n/aFoundry for Ascend + Cambricon

Government Approval and Procurement

StatusVendor
On China government-approved AI hardware supplier list (2026)Huawei Ascend, Cambricon
ExcludedNVIDIA
Day-0 DeepSeek V4 adaptation (Apr 2026)Huawei Ascend, Cambricon, Hygon, Moore Threads
HBM supply constraintAll vendors (HBM is the binding constraint)

Six Things the Landscape Tells You

  1. The "Chinese chips can't train frontier models" narrative is now wrong. Huawei's Pangu Ultra MoE (718B params) trained on 6,000+ Ascend NPUs proves the stack works at frontier scale.
  2. Government procurement now excludes NVIDIA in favour of Huawei and Cambricon. Structural advantage for state, SOE, and regulated-sector deployments.
  3. Cambricon targets 3x output expansion in 2026. 500K accelerators including 300K Siyuan 590 / 690 — meaningful supply growth.
  4. Ascend 910C output doubles to ~600K units. Combined with Ascend 950PR (Q1 2026 launch), total Huawei die output could hit 1.6M.
  5. Day-0 model adaptation is the new normal. Huawei, Cambricon, Hygon, and Moore Threads all adapted DeepSeek V4 on launch day — previously only NVIDIA achieved this.
  6. HBM supply is the binding constraint. Chinese vendors need HBM (mostly from SK hynix, Samsung, Micron) and limited domestic HBM production caps the absolute output ceiling.

What This Means for AI Visibility

The Chinese AI chip stack matters for AI visibility because it determines where Chinese open-source models will be deployed. As government-approved suppliers absorb state and SOE deployments, Chinese LLMs running on Ascend and Cambricon (rather than NVIDIA) become the default for a large segment of Chinese enterprise AI. Brand-visibility monitoring should treat the Chinese-stack deployment surface as a distinct channel.

Methodology

Data from Tom's Hardware on Cambricon 2026 output, RCR Wireless on Huawei Ascend output doubling, CNBC on China AI chip IPO boom, TrendForce on Day-0 DeepSeek V4 adaptation, and IEEE Spectrum on Chinese AI chips replacing NVIDIA.

How Presenc AI Helps

Brand teams selling to Chinese state, SOE, and regulated-sector AI buyers use Presenc AI to monitor brand visibility on Chinese-stack deployments (Huawei Pangu on Ascend, GLM on Cambricon, DeepSeek on Moore Threads). The Chinese-stack surface diverges from NVIDIA-stack defaults and requires separate measurement.

Frequently Asked Questions

Huawei Ascend 910C is the volume leader with 600K units targeted for 2026 and government-approved supplier status. Cambricon Siyuan 590 / 690 is the closest competitor at 300K-500K target. Biren, Moore Threads, MetaX, and Hygon round out the second tier.
Yes. Huawei's Pangu Ultra MoE (718B parameters) trained on 6,000+ Ascend NPUs proves the stack works at frontier scale. Day-0 DeepSeek V4 adaptation across Ascend, Cambricon, Hygon, and Moore Threads (April 2026) further validated this.
Inside China, increasingly yes for state and SOE deployments. China's government-approved AI hardware supplier list (2026) includes Huawei and Cambricon while excluding NVIDIA. Performance per watt still trails NVIDIA's latest generation, but supply security and government approval outweigh the efficiency gap for state buyers.
HBM (high-bandwidth memory) supply. Chinese chip vendors depend on HBM produced primarily by SK hynix, Samsung, and Micron, and domestic HBM production is limited. The constraint caps absolute output rather than per-unit performance. See our <a href='/research/hbm-market-share-samsung-skhynix-micron-2026'>HBM market share page</a> for the deeper context.

Track Your AI Visibility

See how your brand appears across ChatGPT, Claude, Perplexity, and other AI platforms. Start monitoring today.