"Lobster" Craze Sweeps In: Public Funds Break Down AI Industry Investment Opportunities

Securities Times Reporter Wang Xiaoqin

Recently, a “digital lobster” called OpenClaw has quickly become popular, pushing artificial intelligence from content generation to a new stage of task execution.

Many public fund institutions believe that as domestic tech companies rapidly follow up with “lobster” products and create a trend, agent (intelligent agent) applications are gradually being integrated into real work scenarios. The frequency of AI calls and token (model call volume) consumption are expected to significantly increase. Under this change, the focus of the AI industry chain is also shifting. From application-level intelligent agent tools, to underlying computing infrastructure, and to new infrastructure construction such as computing power collaboration, related industry opportunities are gradually emerging.

HuaXia Fund stated that the emergence of OpenClaw has evolved artificial intelligence from “just chatting” to “actually doing work.” OpenClaw is like a super intern, capable not only of understanding instructions but also of planning steps independently, such as opening software, searching for data, analyzing and comparing, and generating reports. This shift from content generation to task execution is considered a qualitative leap.

From an industry trend perspective, this change also indicates a profound adjustment in AI application forms. Guolian Fund believes that AI applications are moving from single-point functions to system-level collaboration. Many AI products in the past only enhanced single functions, such as copywriting, summarization, or image generation; the core change with agents is connecting multiple actions to form complete workflows. Moreover, user demand for AI has shifted from “just trying” to “can it help me do work.” This means that the standard for evaluating an AI product’s value is shifting from showmanship to efficiency improvement and task delivery.

Under this trend, the space for AI application implementation is further expanding. HuaXia Fund predicts that in the future, industries such as finance, healthcare, and manufacturing may see a large number of major applications based on OpenClaw, fundamentally changing work methods. “This is not just a technological victory but a leap in overall social productivity.”

The profound change in AI application forms also has a significant impact on the AI industry chain and brings new investment opportunities.

First, the demand for computing power is greatly increasing. As agent applications gradually enter productivity scenarios, token consumption scales rapidly, and inference computing power demand also rises. This means that the change in AI application forms further emphasizes the importance of computing infrastructure.

ICBC Credit Suisse Fund stated that as deployment thresholds continue to lower, agent tools used in productivity scenarios are expected to explode this year. Agent applications are likely to cause multiple-fold increases in token consumption, amplifying inference computing power demand and benefiting domestic computing power industry. Investment in computing power is expected to remain high over the next two years, with a long-term positive outlook on domestic computing power sectors.

Second, at the industry chain level, “computing power application— infrastructure investment” is gradually forming a positive cycle. The growth in computing power demand will drive the prosperity of related hardware segments.

ICBC Credit Suisse Fund analyzed that with the significant increase in computing infrastructure, the total communication bandwidth demand for AI computing power is also expected to expand substantially in the next two years. As transmission speed requirements increase, the market share of PCB (printed circuit boards) and optical modules in capital expenditure may gradually rise, continuing to achieve high growth.

HuaXia Fund also believes that for large companies and government departments, there is a need for dedicated, secure “private lanes” to operate their AI, which drives huge demand for high-end computing clusters and customized cloud services. Those who can provide faster, more stable, and safer operating environments will become the “infrastructure kings” of the AI era.

Additionally, with the continuous expansion of AI computing power demand, the electricity required for computing operations remains a key market focus. The linkage between computing power and electricity is becoming closer.

HTP Asset Management pointed out that applications like OpenClaw consume massive tokens daily, and the growth in token consumption directly drives up computing power demand. Meanwhile, the operation of computing power depends on electricity support. As AI paradigms shift from “dialogue AI” to autonomous AI agents capable of executing complex tasks around the clock, token consumption shows a clear exponential growth trend. Data shows that China’s overall daily token consumption will rise from about 100 billion at the beginning of 2024 to over 30 trillion by mid-2025, reaching 180 trillion by February 2026.

In this process, computing services are becoming a new energy demand source. HTP Asset Management stated that when overseas users invoke Chinese large model APIs, data completes inference at domestic intelligent computing centers using Chinese computing power and electricity, and the resulting tokens are sent back overseas. This mode is called “Token Going Overseas,” essentially using computing services as a carrier to achieve an “invisible” export of electricity.