Company at a glance
Scale
Revenue up ≈31× since FY2023
Growth prospects
Expanding customer base and product roadmap
via metax-tech.com ↗2020-09 in Shanghai by Chen Weiliang (陈维良), a GPU veteran with 20+ years of design and mass-production experience at AMD; core team carries 5nm tape-…via cls.cn ↗2025-12-17 on STAR Market (688802), issue price RMB 104.66; first-day close up about 692.95%, market value near RMB 300-330 billion; company also pur…China competition
Competes with Huawei Ascend, Biren, Moore Threads, Cambricon in AI accelerators — peers listed at sector level.
SubstituteGlobal comparables
Global reference set: Nvidia, AMD, Google TPU.
Nvidia (A100/A800, H-series)Technical gap
Sector rated substitute vs the frontier.
via stcn.com ↗Mass production from 2024-02; benchmarked against Nvidia A100/A800, 64GB HBM per card, supports 2-64 card interconnect topologies for 100-billion-par…Obstacles
Sector bottleneck applies
CUDA-like software ecosystem, HBM supply, advanced packaging, leading-node access, and large-scale deployment proof.All sourced facts (8)
- Founded: 2020-09 in Shanghai by Chen Weiliang (陈维良), a GPU veteran with 20+ years of design and mass-production experience at AMD; core team carries 5nm tape-out and 7nm production experience via metax-tech.com ↗
- IPO: 2025-12-17 on STAR Market (688802), issue price RMB 104.66; first-day close up about 692.95%, market value near RMB 300-330 billion; company also pursuing a Hong Kong listing via cls.cn ↗
- Product families: 曦思 N series (inference), 曦云 C series (training/inference), 曦彩 G series (graphics rendering), 曦索 X series (science AI), all paired with the in-house MXMACA software stack via metax-tech.com ↗
- 曦云 C500: Mass production from 2024-02; benchmarked against Nvidia A100/A800, 64GB HBM per card, supports 2-64 card interconnect topologies for 100-billion-parameter LLM training; contributed over 70% of revenue (~98% including the full C-series in Q1 2025) via stcn.com ↗
- 曦云 C600: Tape-out deliveries from 2024-10, unveiled at WAIC July 2025; newer process node, higher memory capacity, FP8 mixed-precision support, and a design-manufacturing-packaging-test supply chain fully domestic via news.cn ↗
- Customers: Shortlisted for China Telecom centralized procurement and participating in China Mobile and China Unicom procurement tests; reported hundreds of millions of RMB in operator orders; order backlog about RMB 1.43 billion at IPO review stage via 21jingji.com ↗
- H1 2025 revenue: RMB 915 million, growing rapidly as products pass customer validation cycles via tmtpost.com ↗
- Next generation: 曦云 C700 in design and performance-tuning stage, expected to drive 2026 growth via cls.cn ↗
Tracked chips: MetaX C500 (曦云C500)
About
Fabless GPU design selling training (曦云 C), inference (曦思 N), graphics (曦彩 G) and science-AI (曦索 X) GPU lines plus the MXMACA software stack into data centers, telecom-operator procurements and intelligent-computing centers.
Founded 2020
Why it matters
MetaX is the second domestic GPU maker to reach the public market (after Moore Threads) and one of the few with volume training-GPU revenue; operator procurement wins and a fully domestic supply chain for the C600 make it a central beneficiary of China's AI compute localization.
Where it sits in the stack
Key semiconductor products
- 曦云 C-series training/inference GPUs·C500 in volume, C600 ramping, C700 in development
- 曦思 N-series inference GPUs
- 曦彩 G-series graphics GPUs and 曦索 X-series science-AI GPUs
- MXMACA self-developed software stack·compute ecosystem for AI and graphics
- GPU / AI accelerators
Chip pricing
- MetaX C500 (曦云C500)~$6.6K · 2025-06-30
Financials
688802.SS · Shanghai STAR Market 688802 (IPO Dec 2025) · via Yahoo Finance ↗
Financial trajectory via annual filings via Yahoo Finance
RMB billionsMarket data refreshed daily; annual figures from exchange filings. Figures in the reporting currency shown on the chart.
Ecosystem
Global peer set
- Nvidia (A100/A800, H-series)
- AMD (Instinct)
- Moore Threads 摩尔线程 (domestic peer)
Competitive position
- Positioned as a domestic alternative to Nvidia A100/A800-class training GPUs for Chinese cloud and operator data centers
- Reported as the only domestic GPU firm on operator centralized-procurement shortlists, with China Mobile/Telecom orders
- Second domestic GPU company to list on STAR Market; market value rivaling Moore Threads on debut
- Global comparables: Nvidia (A100/A800, H-series)
- Global comparables: AMD (Instinct)
- Global comparables: Moore Threads 摩尔线程 (domestic peer)
Milestones and signals
- 2020—MetaX was founded.
- 2025—IPO Raise / Issue Price: RMB 104.66 per share
- 2026-10-04—Market cap: RMB 199B
- FY2025—Revenue: RMB 1.6B
- FY2025—Net Profit: RMB -789M
What to watch
- C600 ramp and C700 launch in 2026
- Hong Kong IPO progress after the December 2025 STAR listing
- Operator procurement volumes in 2026 and supply-demand tightness
- Revenue concentration in the C500/C-series line
Questions
Last reviewed 2026-07-11Answered questions
What is the measurable gap versus global reference companies?
Positioned as a domestic alternative to Nvidia A100/A800-class training GPUs for Chinese cloud and operator data centers.
Open questions
- Which products are qualified beyond pilots in fabless?
- Which customers, fabs, or end markets show real adoption rather than policy interest?