AI "Lobster Farming" Goes Viral! OpenClaw Concept Gains Strength, Multiple Listed Companies Respond to Latest Developments in Related Business

OpenClaw continues to ignite the tech industry and capital markets.

Recently, OpenClaw released its core version 2026.3.7. It not only introduced the “Context Engine Plugin Interface” to enable features like “Memory Hot Swapping,” but also achieved native support for GPT-5.4. Data shows that OpenClaw has over 280,000 stars on GitHub, making it the most watched open-source project in the global developer community.

Meanwhile, in the capital markets, AI “lobster farming” concept stocks are defying the trend and performing strongly, becoming a market highlight. According to reporters from the Science and Technology Innovation Board Daily, many listed companies involved in related businesses have been surveyed. Despite intensive industry chain deployment, most companies have stated that “there has been no obvious increase in related business so far.”

▍OpenClaw’s Five Major Breakthroughs Drive “Lobster” Concept Stocks Higher

The term “lobster farming” is an illustrative nickname for users’ personalized training and deployment of the open-source AI agent framework OpenClaw. The project was released as a beta by the overseas open-source team ClawAI in December 2025, with a formal version launched in January 2026. Because the official project icon features a red Boston lobster, and users need to feed data, configure environments, train commands, and adapt scenarios to customize the AI agent, the process closely resembles “farming,” leading domestic netizens to call it “lobster farming.”

Bai Wenxi, Chief Economist of the China Capital Alliance, told the Science and Technology Innovation Board Daily that compared to previous products like AutoGPT and GPT-4o, OpenClaw has five core breakthroughs:

Architectural Philosophy: From “Framework” to “Product” — OpenClaw adopts a layered architecture of Gateway-Agent-Workspace, positioning itself as a “self-hosted AI gateway” and a ready-to-use “digital employee.” File-driven design: managing AI behaviors through Markdown files like SOUL.md (personality), MEMORY.md (memory), AGENTS.md (behavior norms), ensuring transparency and controllability.

Execution Capability: From “Passive Response” to “Proactive Execution” — enabling 24/7 autonomous background operation via heartbeat and cron mechanisms, without human commands, capable of monitoring and scheduling tasks; high local permissions: direct control over file systems, browsers, code editors, terminal commands, enabling real hands-on work rather than just text generation.

Memory System: From “Goldfish Memory” to “Permanent Archives” — dual-mode memory architecture: short-term memory (context cache for ~72 hours) plus long-term memory (local SQLite database), making the system smarter with use; privacy and deployment: from “cloud black box” to “local sovereignty”; data stored entirely on user devices, no cloud dependency, open source and free (MIT license).

Interaction and Ecosystem: From “Single Interface” to “Global Connectivity” — supports integration with over 50 platforms including WhatsApp, Telegram, Discord, Feishu, Slack; flexible expansion via Skills and MCP protocols, with over 3,000 skill plugins on ClawHub; model-agnostic: switch between Claude, GPT-4, DeepSeek, Zhipu GLM, and other mainstream large models.

Privacy and Deployment: From “Cloud Black Box” to “Local Sovereignty” — ChatGPT Agent: data uploaded to the cloud, runs in browser sandbox, with monthly subscription fees of $20–$200; OpenClaw: prioritizes local deployment, data stored entirely on user devices, zero cloud reliance, open source and free (MIT license).

Bai Wenxi believes that OpenClaw’s popularity is creating a “shrimp farming” boom in the A-share market, with clear industry chain benefits. Cloud deployment and ecological services are the most directly affected; increased agent penetration drives token consumption surges; the dual demand for local operation and cloud invocation makes computing infrastructure a rigid requirement. Additionally, as a “digital employee,” OpenClaw is rapidly penetrating scenarios in office work, finance, research, and more, potentially triggering explosive growth in AI applications and edge deployment.

Wang Peng, Associate Researcher at the Beijing Academy of Social Sciences, told the Daily that OpenClaw’s popularity will promote AI from the cloud to edge and terminal devices, significantly benefiting related industry chains. High-performance computing hardware, such as AI PCs, smartphones, and private servers (e.g., AI Boxes), will see increased demand for edge-side computing power due to emphasis on local operation. In industrial AI application development, internet platforms and vertical software providers will integrate OpenClaw capabilities to improve automation efficiency and reshape service processes.

Looking at today’s (March 9) market, the overall A-share market remains sluggish, but lobster concept stocks are defying the trend. The power computing leasing sector surged to daily limit-ups. UCloud Data, Borre Data, QingCloud, Shunwang Technology, and others hit the 20% daily limit; TuoWei Information, Great Wall, Meiliyun, YunSai Zhilian, and others closed with 10% limit-ups; Hand Enterprise rose over 17%.

▍Cloud Service Providers Take the Lead, Becoming the Core Beneficiaries of the First Wave

In response to the opportunities brought by OpenClaw, several listed companies have disclosed related business deployments and technical adaptations across the entire industry chain, including cloud deployment, enterprise applications, hardware compatibility, and cybersecurity.

Currently, cloud deployment is the main way for individual and small-to-medium enterprise users to “farm lobsters.” Due to high hardware requirements and complex environment setup for local deployment, ordinary users prefer to use one-click deployment images or dedicated cloud hosts provided by cloud service providers, making cloud providers the earliest and most benefited players in this wave.

UCloud Data is among the earliest to implement related business. Recently, the company announced on investor platforms that it launched images for deploying OpenClaw (Clawdbot) in late January, becoming one of the first to enable cloud deployment in the industry. Its services are now available in the US, Singapore, Japan, and other overseas nodes, helping users quickly build “24/7 personal AI super assistants.”

Notably, UCloud Data stated today (March 9) that, in response to investor inquiries about using OpenClaw to assist investment research, the development team has engaged with several well-known public and private fund managers to summarize their investment practices and identify the most valuable application scenarios and boundaries for OpenClaw in research. Based on UCloud’s strong cloud capabilities, they have developed a practical AI research framework. However, the Daily reports that this article has since been deleted, reasons unknown.

Meanwhile, the Daily learned from the securities department of Wenshuidata that their OpenClaw images are mainly customized based on customer needs for local deployment, and due to the short time since launch, no significant business growth has been observed yet.

Other cloud service providers are also rapidly deploying. QingCloud recently announced full adaptation of OpenClaw on its QingCloud platform, launching dedicated application templates for one-click deployment, with edge computing nodes also adapted; Wangsu Technology disclosed that leveraging over 2,800 global edge nodes, it has launched an OpenClaw edge deployment solution with security features, currently in testing with some enterprise clients.

Regarding OpenClaw-related business, the Daily contacted Wangsu’s securities department, which stated that they can provide deployment environments and platforms for OpenClaw but have not yet seen a significant impact on their business.

The staff further explained, “Currently, our main business is CDN-based network acceleration, related to traffic stability and efficiency. Products like Doubao and DeepSeek mainly replace traditional traffic flows. We haven’t yet seen a significant new traffic increase from OpenClaw’s popularity. Its future impact on our business is still under observation.”

▍Intensive Enterprise Application Deployment, Most Still in Early Implementation Stage

If cloud deployment is the traffic entry point for OpenClaw, then enterprise scenario deployment is key to scaling revenue and building long-term competitiveness. Compared to entertainment-oriented deployment for individual users, enterprise clients have stronger willingness and rigid demand for AI agents, with higher customer stickiness, making this the core track for listed companies.

Hand Enterprise announced on March 5 that its Large Customer Digital Intelligence Business Unit launched the AI-CAMP AgentHub platform. It is described as a “unified application platform built on the AI middle platform,” aiming to integrate diverse AI capabilities—from open-source communities, third-party professional middle platforms, to popular “execution-type” AIs like OpenClaw—into a complete, controllable, and operable enterprise AI solution.

Beijing XinYuan Information Security, a leader in digital security solutions, stated on investor platforms that its self-developed XinYuan MiXin product has completed OpenClaw integration and can provide solutions and services for related business needs.

Zhongkechuangda, a provider of intelligent operating systems and technology, has achieved full-stack adaptation of OpenClaw across multiple platforms. Recently, it announced the completion of deep integration and large-scale deployment of OpenClaw on its Magic Cube PAI 3 and AIBOX platforms as of February 11.

The Daily learned from Zhongkechuangda’s securities department that OpenClaw’s business was launched recently, and no significant impact on their operations has been observed yet.

Recently, software and digital technology service provider Ruantong Power integrated computing infrastructure, token economy, and the OpenClaw framework to launch the Ruidong AI Agent Cloud Platform. The company plans to deeply embed OpenClaw into products like industrial IoT, smart parks, and home automation, with applications in office automation, smart manufacturing, energy, and healthcare. In industrial scenarios, OpenClaw-driven inspection robots can operate 24/7, with token consumption per task potentially 40% lower than traditional manual processes.

▍Domestic Hardware Adaptation Continues to Break Through, Business Growth to Be Observed

As user demands for data privacy and information security increase, full-stack adaptation of domestic hardware is crucial for achieving independent control in “lobster farming” and key for domestic manufacturers to differentiate competitively. Currently, chip and server vendors in A-shares are actively deploying full-stack adaptation from underlying chips to upper-layer applications.

Leading domestic CPU manufacturer Loongson has achieved OpenClaw operation in a domestic environment. Recently, it announced on official platforms that the “Little Box” based on Loongson 3B6000M chip successfully completed local deployment of OpenClaw. Developers no longer need to purchase expensive Mac minis or dedicated servers; they can run OpenClaw smoothly on fully domestic hardware.

Rockchip has implemented lightweight deployment of OpenClaw on consumer-grade devices. As a top domestic AIoT chip vendor, its RK3588 chip is one of the most widely used high-end ARM architecture chips in China. The company recently stated on investor platforms that its RK3588 and RK182X chips are suitable for personal deployment of OpenClaw and other AI agent applications, with solutions already in use that invoke cloud large models.

The RK182X co-processor offers high computing power and bandwidth, supporting local deployment of large models on various smart devices to quickly enable edge AI agents, better meeting user needs for data privacy, low latency, and autonomous multi-tasking.

The securities staff of Rockchip told the Daily that “RK3588 currently calls cloud large models; for local deployment in the future, RK182X will be used.” They further noted that “OpenClaw’s popularity has not yet brought significant business growth for us; overall, it remains normal.”

In the server and complete machine sector, Inspur has launched AI servers optimized for OpenClaw, supporting multi-node clusters for deploying AI agents, meeting large-scale computing needs, and has supplied to multiple cloud providers and enterprise clients.

China Merchants Securities commented that the full-scenario personal assistant Agent similar to lobster is accelerating deployment, with 24/7 operation creating massive token demand. Currently, this is still in the early stage of token volume expansion. Looking ahead, as Agent adoption spreads, tokens—considered the “basic energy”—may see further development. Overall, AI is entering a stage of application deployment, demand acceleration, and industry chain prosperity, with China’s rapidly increasing token consumption indicating a fast-growing demand for computing inference.

(Article source: Science and Technology Innovation Board Daily)