BEIJING, September 13 — China’s artificial intelligence industry is changing direction. According to a new report on the sector, the country’s AI companies are shifting their strategic focus away from building ever-larger foundation models and toward deploying AI agents — software designed to plan and carry out multi-step tasks with limited human supervision. The report describes agents as distinct from the chatbots that have dominated the generative AI era.
Where a chatbot mainly responds to prompts, an agent can execute tasks. That difference is now shaping where Chinese firms place their bets.
A maturing market
The pivot reflects a maturing market in which companies increasingly prioritize practical, task-executing software over raw model scale. Competition among domestic model developers has intensified, and pressure is mounting to demonstrate returns on heavy investment in computing infrastructure. Enterprises, meanwhile, are seeking automation of workflows such as customer service, software development, and data analysis.
China has assembled one of the world’s largest AI ecosystems, anchored by technology hubs in Beijing, Shanghai, Shenzhen, and Hangzhou. Major domestic players include Baidu, Alibaba, Tencent, ByteDance, and Huawei, alongside a wave of startups that emerged after 2022, when the public release of OpenAI’s ChatGPT triggered a global generative AI race.
Chinese firms responded with their own large language models, including Baidu’s Ernie Bot, Alibaba’s Tongyi Qianwen, and models from the startup DeepSeek, which drew international attention for competitive performance at lower reported training costs.
Following a global pattern
The turn toward agents follows a broader industry pattern seen worldwide, where developers have moved from releasing standalone models to packaging them into products that can browse the web, call software tools, and complete tasks autonomously. The report indicates Chinese companies are following this trajectory, integrating agents into consumer apps, enterprise platforms, and cloud services. The economic logic is straightforward.
Building and operating frontier models requires expensive GPU clusters and large energy inputs, and investors have questioned when such spending will translate into revenue. Agent products are viewed as a nearer-term path to monetization because they can be sold as subscriptions or embedded in existing business software.
Governance questions accompany the shift. The report notes that agent deployment raises issues of reliability, security, and accountability when autonomous systems take actions on a user’s behalf.
Regulation and what comes next
China has implemented rules governing generative AI services and algorithmic recommendations, requiring providers to register certain services and comply with content and data requirements. How those frameworks apply to autonomous agents remains an open question that the report identifies as a factor shaping the industry’s next phase. The report does not specify a single publication date or quantify the size of the agent market.
Its central claim is directional: China’s AI sector, once defined by a race to build bigger models, is now organizing around agents as the primary product frontier. What to watch next is whether Chinese firms can convert agent products into recurring revenue — and how regulators decide that rules written for chatbots should apply to software that acts on its own.
