Huaxi Securities: OpenClaw Ignites AI Agent Era, Market Scale May Exceed 350 Million by 2031

On March 13, early 2026, the open-source AI project OpenClaw gained popularity for its ability to autonomously perform complex tasks — it is no longer just a conversational chatbot but a “digital employee” capable of local file operations and workflow automation.

Behind this trend is a dual resonance of demand and supply: IDC predicts that the number of agents in Chinese enterprises will exceed 350 million by 2031, growing at an annual rate of over 135%. Meanwhile, OpenAI’s latest release, GPT-5.4, has enhanced the cross-software execution capabilities of intelligent agents. Advances in underlying technology are accelerating the deployment of OpenClaw.

OpenClaw is an open-source autonomous AI virtual assistant software project. It was developed by software engineer Peter Steinberger and initially released on GitHub at the end of 2025 under the name Clawdbot, later renamed Moltbot, and finally settled on the current name.

In early 2026, the project attracted attention for its ability to autonomously handle complex tasks within applications and online services based on user instructions. OpenClaw is designed as an autonomous AI virtual assistant that can replace users in executing tasks, rather than just a conversational chatbot.

OpenClaw stores configuration data and interaction history locally, giving it a more persistent memory. Driven by natural language commands, it performs tasks such as file operations, workflow automation, browser automation, and multi-IM platform interactions locally or in private cloud environments. It bridges the gap from “dialogue-based suggestions” to “automated execution,” serving as a self-hosted AI digital employee for individuals and enterprises.

On the demand side, according to IDC estimates, the scale of agents in Chinese enterprises is entering an acceleration phase. With continuous upgrades in domestic model capabilities, rapid maturation of intelligent agent technology and application ecosystems, and the influence of industry policies, the number of enterprise agents in China is expected to surpass 350 million by 2031, with a compound annual growth rate of over 135%. Additionally, as the density and complexity of agent tasks increase, there will be an exponential rise in intelligent agent token consumption, potentially over 30 times annually. It is believed that agents could become one of the earliest application scenarios to be widely adopted.

On the supply side, on March 5, OpenAI announced the release of the latest upgrade to the GPT-5 model, GPT-5.4. One of the most significant changes in this update is the enhancement of agent capabilities. In the API and Codex, GPT-5.4 has achieved native “computer operation” functions for the first time, supporting intelligent agents in executing complex workflows across software. As a leading product in the large-scale model industry, the technological development trend indicates that strengthening agent capabilities through foundational large models will be a key focus of iterative improvements, likely continuing to enhance core technology and accelerating the commercialization of OpenClaw.