Guolian Minsheng Securities stated that the current phenomenon of "Eastern Cultivation of Western Lobster" superficially reflects major domestic companies' FOMO-driven anxiety regarding the OpenClaw trend, but actually indicates an ecosystem dislocation between domestic and overseas markets in the current wave of AI application implementation. Compared with overseas markets, China possesses cost advantages, traffic advantages, and policy advantages in this "Lobster Wave." From a cost perspective, driven by OpenClaw, token consumption for domestic models is accelerating. OpenRouter data shows recent large model consumption has hit record highs with a steep growth curve, and Chinese models rank high in token consumption. The rapid surge in token consumption also serves as proof of the cost advantage. The main views of Guolian Minsheng Securities are as follows:
Event: On March 10, Knowledge Atlas officially launched AutoClaw (Chinese name: Australian Lobster) – a local version of OpenClaw that can be installed with one click, possessing the full native capabilities of OpenClaw. Knowledge Atlas currently offers certain free quotas for users to experience AutoClaw's capabilities. Knowledge Atlas introduced that AutoClaw comes pre-installed with over 50 mainstream Skills, ready to use out-of-the-box, covering high-frequency scenarios such as content creation, office work, coding, marketing, and financial investment research, and supports one-click integration with instant messaging tools like Feishu. It can connect to the CodingPlan/API of any model, including APIs from model providers like DeepSeek, Kimi, MINIMAX, and GLM.
In fact, besides Knowledge Atlas, many major domestic companies are actively responding to this "Lobster Wave," participating in the localization process of OpenClaw. Since February, large model companies like Knowledge Atlas, MINIMAX, and major cloud providers have successively released related Claw products. Comparing the product characteristics of OpenClaw and domestic participants reveals differences in two aspects: deployment method and model source.
From the deployment perspective, the participation methods of various entities in this "Lobster Wave" show a trend of "a hundred flowers blooming." Unlike OpenClaw, which focuses on local deployment characteristics, the "lobster products" from major domestic companies feature three deployment methods: local deployment, cloud deployment, and hybrid deployment, offering greater selectivity. As an open-source AI agent, OpenClaw was originally designed to emphasize localization and flexibility. According to its founder Peter Steinberger's idea, the safest and most flexible way to use OpenClaw is local model deployment + OpenClaw + integration with existing instant messaging chat windows. Guolian Minsheng Securities believes this phenomenon occurs primarily because the current domestic "lobster products" focus more on deployment convenience and security.
From the model source perspective, most current participants primarily support multi-model matching, but some companies choose to use only their own models as the main source. A horizontal comparison shows that OpenClaw supports a wider variety of large models, involving both overseas and domestic mainstream models; whereas domestic "lobster products" mainly support calling domestic mainstream large models, with some currently supporting only their own models as the primary source. This might be due not only to major companies leading the "lobster product" layout being more concerned about the calling costs and operability of domestic large models but also includes the competition for "new distribution rights" in the Agent era.
Why does the "Lobster Fever" triggered by this round of OpenClaw exhibit the characteristic of "Eastern Cultivation of Western Lobster"? Although the rise of OpenClaw as a product originated overseas, domestic attention has rapidly increased since February, with numerous participants entering the field, showing the characteristic of "Eastern Cultivation of Western Lobster." Combined with the above comparison, the current "Eastern Cultivation of Western Lobster" superficially reflects major domestic companies' FOMO-driven anxiety regarding the OpenClaw trend, but actually indicates an ecosystem dislocation between domestic and overseas markets in the current wave of AI application implementation in the Agent era. Compared with overseas, China possesses cost advantages, traffic advantages, and policy advantages in this "Lobster Wave."
From a cost perspective, the domestic large model ecosystem has formed a unique "low-price API." In terms of calling prices, the API calling price for domestic models is approximately one-sixth that of similar overseas products. This advantage stems from cheaper electricity behind domestic computing power, more flexible hardware configurations, and price competition among model vendors. Driven by OpenClaw, token consumption for domestic models is accelerating. OpenRouter data shows recent large model consumption has hit record highs with a steep growth curve, and Chinese models rank high in token consumption. The rapid surge in token consumption also serves as proof of the cost advantage.
From a traffic perspective, this "Lobster Fever" is inseparable from the active participation of major companies. For these companies, deploying OpenClaw may not only relate to revenue growth from the Agent business itself but also includes the competition for potential "new distribution rights" in the Agent era. Looking ahead, as Agent deployment continues to be implemented, the interaction method between humans and the digital world may undergo a new reconstruction – when the operating entity transitions from humans to Agents, new traffic entry points might also shift. Although the current OpenClaw may still face technical immaturity, its transformation of AI from "Chat" to "Work" could raise market concerns about changes in new platform entry points, leading to widespread "FOMO" and active deployment among major companies this round. Furthermore, major companies represented by Tencent might further amplify traffic attention through their social fission effects, promoting the further spread of societal attention to the "Lobster Wave."
From a policy perspective, recent attention from local governments in China towards OpenClaw has also catalyzed sentiment. Shenzhen and Wuxi took the lead in introducing special policies, becoming pioneers of the "Lobster Fever." 1) On March 8, according to Shenzhen Release, recently, the Shenzhen Longgang District Artificial Intelligence (Robot) Department publicly solicited opinions on the "Several Measures to Support the Development of OpenClaw & OPC (Draft for Comment)." It proposed encouraging market-oriented, professional platform carriers to launch "Lobster Service Zones," providing free OpenClaw deployment services, and offering certain subsidies to eligible entities. 2) On March 9, Wuxi High-tech Zone released the "Several Measures on Supporting the Integrated Development of Open Source Community Projects like OpenClaw and the OPC Community (Draft for Comment)." The 12 "lobster farming" policies, from basic support to industrial implementation, from talent introduction and cultivation to safety and compliance, release a package of tangible monetary benefits, with single-item support reaching up to 5 million yuan. At the deployment level, full subsidies of up to 1 million yuan are provided for local cloud platforms offering free deployment and development toolkits.
Regarding investment targets, it is recommended to focus on: 1) Companies with model or application capabilities, benefiting from potential accelerated growth in API call-related revenue: MINIMAX-WP (00100), Knowledge Atlas (02513); 2) Platform-based internet companies with synergistic computing resources, model capabilities, and application scenarios: Alibaba-W (09988), Tencent Holdings (00700); 3) Companies involved in massive computing power and Token consumption, expected to drive demand for related cloud services: Kingsoft Cloud (03896), Baidu Group-SW (09998), Kingdee International (00268); 4) AI computing hardware layer companies with strong certainty: Iluvatar CoreX (09903), Biren Technology (06082), Hua Hong Semiconductor (01347), Lankeep Technology (06809); 5) US-listed related targets: SanDisk (SNDK.US), Micron Technology (MU.US), NVIDIA (NVDA.US), Intel (INTC.US), Lumentum (LITE.US), Cheetah Mobile (CMCM.US).
Risk warnings: AI model technology development falls short of expectations; AI application commercialization progress falls short of expectations; Deterioration of AI industry competition landscape.
Comments