Global AI · China

AI in China

Large-scale platforms, fast-improving models, open releases, manufacturing, robotics, cloud ecosystems, and hardware constraints.

China combines a very large domestic market, strong engineering talent, mobile and commerce platforms, cloud providers, manufacturing, robotics, electric vehicles, state-supported research, and rapidly improving model teams. Its ecosystem is not merely a copy of the United States: distribution, regulation, language, hardware access, and industrial deployment create a distinct path.

What China contributes to the AI system

Foundation models

Qwen, DeepSeek, ERNIE, Hunyuan, GLM, and other model families compete across language, coding, vision, video, reasoning, and agents.

Consumer distribution

Alibaba, Tencent, Baidu, ByteDance, Huawei, Xiaomi, and other platforms can place AI inside commerce, communication, search, cloud, and devices.

Manufacturing and robotics

Factories, logistics, drones, electric vehicles, batteries, and electronics create large physical-AI deployment environments.

Open-model influence

Chinese teams release widely used open-weight models, code, datasets, and developer tools that spread beyond China.

Cloud and enterprise

Alibaba Cloud, Tencent Cloud, Baidu AI Cloud, Huawei Cloud, and others offer domestic model and deployment stacks.

Constraint-driven efficiency

Restrictions on access to the most advanced foreign chips increase pressure for model efficiency, domestic accelerators, software optimization, and supply-chain substitution.

Leading organizations and teams

Alibaba · Models and agents

Qwen

A broad model ecosystem spanning language, coding, vision, image generation, robotics, world models, APIs, and open releases.

Frontier models

DeepSeek

Known for efficient reasoning and coding models, open releases, developer APIs, and strong attention to training and inference economics.

Models, search, cloud

Baidu ERNIE

A model family and platform connected to Baidu search, PaddlePaddle, AI Studio, enterprise cloud, agents, and multimodal applications.

Models and product integration

Tencent Hunyuan

Develops language, agent, image, video, 3D, translation, and research systems integrated across Tencent products and cloud services.

Open model ecosystem

ModelScope

A large platform for models, datasets, demonstrations, tools, and developer collaboration.

Framework

PaddlePaddle

Baidu’s deep-learning platform for research, training, deployment, and industrial applications.

Cloud and domestic compute

Huawei Cloud AI

Cloud, model, accelerator, data, and enterprise systems connected to Huawei’s broader hardware and software ecosystem.

Research

Chinese Academy of Sciences

A major national research network spanning computing, automation, robotics, materials, biology, and scientific AI.

People and teams to follow

Multimodal and agent systems

The Qwen research team

Follow for model families spanning code, vision, image, robotics, agents, world models, and open releases.

Models and industrial stack

The Baidu ERNIE team

Follow for model architecture, PaddlePaddle, multimodal systems, agents, search, and enterprise deployment.

Product integration

The Tencent Hunyuan team

Follow for language models, agents, 3D generation, media, cloud APIs, and deployment across a large product ecosystem.

Official and primary resources

Models and research

ERNIE official blog

Official releases, technical reports, open models, and platform announcements.

Models and tools

Tencent Hunyuan

Official Hunyuan model, agent, media-generation, and research information.

Open research platform

OpenI

Models, code, datasets, competitions, and collaboration in the Chinese open AI community.

Strategic opportunities and constraints