Author: Nancy, PANews

In March, the air is filled not only with the scent of spring but also a hint of lobster from the AI realm.

As everyone begins to "raise lobsters," the open-source framework OpenClaw is stepping out from the screen into reality, penetrating and reconstructing the hardware world. Meanwhile, a lobster installation movement is rapidly spreading in China, stirring up the landscape of domestic large models.

Agent steps out of the screen, assembling hardware like building with Legos

Just like Lego blocks, an increasing number of hardware products are being combined with OpenClaw to assemble various intelligent devices.

The Rokid Glasses, an AI eyewear with display capabilities, recently launched a seemingly simple yet highly imaginative feature - customizable agents. Through a standard SSE interface, users can connect the glasses to any backend system, such as OpenClaw. In other words, these glasses have become a programmable device that can connect to various AI capabilities.

The Apple Watch transformed into an AI control terminal after connecting to OpenClaw, allowing users to view their inbox, receive notifications, approve/reject actions, quickly reply, and even send commands directly from the watch without continuously relying on a smartphone.

The wearable device company WHOOP, after integrating OpenClaw, enables AI to automatically read key data such as sleep scores, recovery, and HRV, and pushes personalized exercise and daily routine suggestions based on these metrics, upgrading from a recording tool to a health decision assistant.

If wearable devices are just the beginning, the robotics field shows even more imagination.

The seven-axis robotic arm NERO launched by Songling Robotics, after integrating OpenClaw, no longer requires users to write complex control programs; they can simply issue commands in natural language, such as asking the robotic arm to move an object to a specified location. OpenClaw will automatically parse the semantics, plan the motion path, generate control scripts, and execute them.

Some companies are making even bolder attempts. For example, the AI company DeepMirror has integrated OpenClaw into its core physical AI products and connected it to the middleware of Yushu Technology's robotics software, enabling robots to understand space and time. The system can recognize people, remember object locations, and record the sequence of events, thereby forming a kind of world memory. This means AI not only understands language but is continuously comprehending changes in the real world.

In addition to hardware manufacturers, some individual developers/players are also using OpenClaw for various hardware device modifications.

For example, the recently popular open-source project VisionClaw combines Meta's Ray-Ban Meta smart glasses with OpenClaw. Users just need to tap the glasses or say a command, and the system can understand, explore, or even execute tasks based on what the camera sees, becoming a visual AI Agent.

After Vbot Super Dog connected to OpenClaw, it is no longer limited to preset programs. In the past, robotic dogs were more like automated devices carrying out fixed actions. After integrating Agent, it can autonomously plan actions based on tasks, such as patrolling indoors, searching for targets, or reporting environmental conditions.

Moreover, on platforms such as GitHub, X, and Xiaohongshu, OpenClaw is continually "transforming." Some have inserted OpenClaw into second-hand mobile phones, others have integrated it into smartwatches, or even fully control smart homes. These attempts are breaking existing boundaries of smart hardware, opening up more possibilities and imaginative space.

The continuous exploration of these OpenClaw-based smart hardware will promote the widespread adoption of Agents among the public. However, the security risks posed by system permissions and the ongoing token costs incurred from model calls are real challenges that need to be addressed in the future.

Lobsters "coming ashore," domestic tokens accelerating "going overseas"

Currently, the OpenClaw trend is rapidly sweeping through China, not only attracting big companies to get involved but also prompting policy impositions, with player enthusiasm on the rise.

In Shenzhen, this fever has turned into an out-of-town urban landscape. A "free lobster installation" booth appeared at the entrance of the Tencent Building, with lines stretching over a thousand people, including elementary school students and elderly attendees. Related photos flooded social media; various lobster-themed gatherings have been frequently held, with many events packed to capacity and even emergency additional seating arranged; the local government is proposing new policies to support "raising lobsters."

Even the world-renowned "hardware Silicon Valley" Huaqiangbei has ventured into the OpenClaw business, where various modified lobster boxes are becoming new hot-sellers.

The OpenClaw storm brewing in China has unexpectedly brought a new growth curve to domestic large model manufacturers, accelerating the overseas expansion of Chinese tokens.

According to the latest data from OpenRouter, the world's largest API aggregation platform, the global consumption of tokens for the top ten models this week approached 8.5 trillion, with Chinese models accounting for half of the global actual call volume. Numerous Chinese models hold core positions on the list, especially MiniMax M2.5, which firmly occupies the top spot with nearly 2 trillion calls this week, showing a growth rate of 21%.

As call volumes exploded, MiniMax also became one of the hotly discussed lobster concept stocks, with its stock price rising over 200% this year. Dramatically, just two months post-listing, MiniMax's market value exceeded Baidu’s; its founder and CEO, Yan Junjie, previously interned at Baidu and received the second Baidu Scholarship. Recently, its stock price further climbed due to its appearance on the PinchBench list and being shared by OpenClaw's creator, Peter Steinberger, on social media. From MiniMax's first financial report, despite still being in a loss state, its revenue structure significantly improved, with contributions from international markets exceeding seventy percent.

Notably, OpenClaw is OpenRouter's largest single application. This month, its usage surged to the top globally, over twice that of the second-place Kilo Code, and is a key driver for domestic large models going overseas.

Among the top five models in OpenClaw, three are from domestic large models. Notably, Kimi2.5 is far ahead; OpenClaw announced last month that it would be the first officially designated free main model, further promoting its usage scale. Over the past two months, orders from paid users of Kimi have surged sharply, directly driving the income curve to rise vertically, with revenue in the past 20 days exceeding the entire year of 2025, while overseas revenue has notably surpassed domestic for the first time.

It is particularly important to note that the data from OpenRouter does not cover all real usage globally. Due to its integration convenience and low-cost model access capabilities, it has become quite popular among developers, and OpenRouter has received attention from Peter Steinberger multiple times. Nevertheless, OpenRouter's user base has surpassed 5 million, primarily concentrated overseas, making this data even more valuable as a reference and somewhat reflecting the "offensive" of Chinese large models in the overseas market.

In the new track of Agents, the boundaries between various roles are being redefined. Hardware manufacturers, developers, and large model companies are all being drawn into the same ecosystem.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。