MetaX - 沐曦
"Make the Difference. Mark the Future."
Chinese fabless GPU company developing full-stack high-performance GPGPU chips for AI training, inference and graphics rendering.
MetaX (Shanghai MetaX Integrated Circuit Co., Ltd.) is a Chinese fabless semiconductor company that independently develops full-stack, high-performance general-purpose GPU (GPGPU) chips and computing platforms. Listed on the Shanghai Stock Exchange STAR Market under stock code 688802, MetaX builds energy-efficient, versatile compute for intelligent computing, general-purpose computing and cloud rendering.
Its product portfolio spans four GPU accelerator-card families: the MXN (N-series, e.g. N100/N260/N300) for AI inference, the MXC (C-series, e.g. C500/C550/C600) for AI training and general compute, the MXG (G-series G100) for graphics rendering, and the MXX (X-series). MetaX also delivers complete GPU servers, GPU workstations and supernode systems, including liquid-cooled full-cabinet, optical-interconnect and 3D-mesh configurations for large-scale clusters. Its self-developed MXMACA heterogeneous computing platform provides the software stack and developer ecosystem, with documentation, SDKs and a DeepSeek inference zone offered through its developer community. MetaX targets data centers, finance, healthcare, energy, education and research, transportation, autonomous driving, digital twin and metaverse workloads. The company emphasizes mass-production GPU experience, a complete software ecosystem, a large innovative IP patent portfolio and a full-build R&D team. MetaX positions its compute as a foundational layer for the digital economy, supplying domestic high-performance AI compute developed on advanced process technology with self-owned IP.
Primary type & automation activities this supplier delivers:
MetaX C600
New-generation general-purpose GPU (GPGPU) for AI training and inference, built on domestic advanced process technology.
MetaX C500
MXC-series GPGPU accelerator card for large-scale AI training and general-purpose computing.
Contact MetaX
WEBSITE
https://www.metax-tech.com/HEADQUARTERS
China
Company Facts
Founded
-
Primary Role
Component Supplier
Company Size
-
Primary Region
Asia-Pacific
Annual Sales
-
Funding Stage
Public
Funding Total
-
Certifications & Memberships
ROBOTTODAY Weekly August 10 – 14, 2026
Your weekly robotics briefing: Unitree's IPO is swamped by retail demand, the US puts 100% tariffs on imported drones, LG and Nvidia commit to a 2027 humanoid, and Uber and Pony.ai bring 2,000 robotaxis to Europe.
Your Three-Shift Robot Is on a One-Shift Maintenance Plan: Phase 3 and the PM System
A three-shift robot on a one-shift maintenance plan ages at triple speed. RSF Phase 3: three PM tiers, three trigger models, environmental corrections.
WAIC Review 2/4: World Models vs. VLA: Robotics' Next 'ChatGPT Moment'
At WAIC 2026, world-model and VLA camps staked rival paths to generalist robots; Zhiyuan, SenseTime and NVIDIA predict a robot 'ChatGPT moment' in 2-5 years.
ROBOTTODAY Weekly July 20 – 24, 2026
Your weekly robotics briefing: AGIBOT unveils four new robots at WAIC 2026, the NDAA bars Chinese humanoids from the U.S. military, SoftBank eyes Gravis Robotics for its Roze vehicle, and Anduril reveals its Thunder autonomous attack tiltrotor at Farnborough.
AgiBot and Others Transition to Domestic Chips Amid Rising Costs in Robotics
As Chinese robotics companies ramp up production, domestic chips are increasingly replacing Nvidia's offerings due to cost pressures. On July 22, Wang Chuang, Senior Vice President of AgiBot, revealed that the company has begun integrating domestic chips into its robots, including the 'cerebellum' for joint motion control and communication chips, collaborating with various chip manufacturers. This shift is significant as it reflects a broader trend in the industry where high costs of foreign chips, which can account for one-third of hardware procurement costs, are driving companies like AgiBot and UBTECH to seek local alternatives. UBTECH has even formed a joint venture with domestic chip maker MetaX to focus on the research and mass production of intelligent chips. Looking ahead, the transition to domestic chips is expected to reshape not only supply chains but also the entire cost structure and competitive landscape of the robotics industry. By 2025, Chinese companies are projected to dominate the global humanoid robot market, with AgiBot leading the way with approximately 5,168 units shipped.
Humanoid Robots Domestic Chips Robotics Industry AI Cost Efficiency
AI for Science 2026: The Global Race for Scientific Discovery
2026 AI for Science landscape report: WAIC 2026 exhibitor breakdown, MIT Boltz-2 and CMU Coscientist case studies, US/China/EU policy comparison, Anthropic Claude Science, and US academic cautionary views.
Biren Develops Next-Gen Optical Links for Enhanced AI Chip Connectivity
Biren, a Shanghai-based startup, is pioneering optical technology to connect thousands of AI accelerators into unified computing systems. As AI models grow increasingly complex, the demand for efficient interconnectivity among GPUs has intensified, with traditional copper connections reaching their limits. Biren's innovative near-packaged optics (NPO) architecture aims to support clusters of up to 1,024 AI accelerator cards, significantly enhancing computing capabilities. This advancement is crucial for AI infrastructure companies striving to meet the needs of models with trillions of parameters. Biren's approach addresses the scalability challenges faced by conventional server architectures, which typically cap at around 128 GPUs. By positioning optical fibers closer to chips, the company enhances bandwidth and data transmission efficiency, positioning itself against competitors like MetaX and Alibaba. Looking ahead, Biren is testing its NPO interconnect system while also developing an orthogonal hardware architecture in collaboration with ZTE. Although the semiconductor industry is still in the early stages of optical interconnect technology, widespread adoption is projected for around 2028, contingent on further advancements and real-world validation of these systems.
AI and Robotics
Physical AI Landscape: From Digital Intelligence to the Embodied Physical World
A comprehensive industry analysis of Physical AI — covering Embodied AI, Field AI, Factory AI, VLM, World Model, VLA, and the VLA vs WAM architecture debate. Includes 20+ companies, 28 references, and the RobotToday Physical AI Landscape diagram. July 2026.
2026H2 Robotics Events Briefing: Tailored Recommendations for Robotics Readers
Discover the ultimate directory of global robotics events in 2026. Tailored briefings for engineers, investors, and founders on humanoid robots, AI, and industrial automation by RobotToday.
ROBOTTODAY Weekly June 30 – July 4, 2026
Unitree wins CSRC approval for China’s first humanoid IPO, Korea unveils ₩312T AI investment, RoboCup 2026 opens in Incheon, and Paris launches Europe’s inaugural Physical AI summit.
Automate 2026 | Ubicept Explores Photon-Level Machine Vision
Ubicept demonstrated single-photon SPAD imaging at Automate 2026, achieving 140 dB HDR and >10 dB SNR gains for industrial robotics edge AI.
Pittsburgh Regional Alliance Convenes Nine CMU-Rooted Robotics Firms at Automate 2026
At Automate 2026, the Pittsburgh Regional Alliance convened nine robotics firms — most spun out of Carnegie Mellon — spanning materials, perception and AMRs.
