After Tencent's "shrimp farming" caused a surge similar to Baidu's, Baidu was so anxious that it hurriedly started "installing shrimp" at the entrance.

On March 11th at 10:40 AM, during the peak work hours at the Haidian West Erqi Software Park in Beijing, the square in front of the Baidu Technology Park K1 building had already become lively.

The news of “Baidu offering free installation of OpenClaw” spread, and people wearing various badges flocked in from all directions. Starting at 10:30, lines began to form on-site. By 11:00, the line had stretched across the entire entrance of Baidu K1.

A rough estimate indicated that over 20 technical staff and nearly a thousand new users were simultaneously participating in this special “installation service.” The installations were divided into two parts: PC and mobile, with each booth crowded with people.

This was Baidu’s rapid follow-up “defensive battle” after Tencent recently launched a similar event. On the 6th, Tencent introduced a free installation event and announced the launch of the Lobster Family Bucket, prompting media reports claiming “Tencent’s valuation has raised an entire Baidu.”

The two giants, Tencent and Baidu, turned what was originally an online cloud computing customer acquisition battle into an offline “street fight.”

As Baidu and Tencent clashed, AI native companies such as Zhipu AI, Kimi, and MiniMax had quietly laid out their own “lobster ecosystem.” Recently, Zhipu opened a free quota for the GLM-4 API, Kimi was reported to be testing a lightweight computing package for developers, and MiniMax was offering computing subsidies to enterprise users through “Conch AI”—they do not sell servers directly but are bundling computational resources with model capabilities, participating in this war in another way.

The battlefield in West Erqi is far busier than it appears. On the surface, this is a “free installation” welfare activity. However, being in the midst of it, one can clearly feel that this is a meticulously planned customer acquisition campaign.

$17.8 “Try Before You Buy”

Subsequent Monthly Payments

Installation is free, but cloud servers and tokens cost money.

The specific product combination strategy is quite aggressive: Baidu Cloud’s intelligent lightweight application server originally priced at $130/month is now $9.9 for the first month; the basic computing Qianfan Coding Plan is now $7.9 for 15,000 conversation quotas, valid for one month, totaling $17.8.

For developers accustomed to hourly billing with traditional cloud service pricing in the hundreds, $17.8 for a complete development environment for a month is almost equivalent to “try before you buy.”

At the event, a product manager for Baidu Cloud’s lightweight application server used a vivid metaphor to explain their product logic: “The cloud server is the body of the lobster, while Baidu Intelligent Cloud’s Qianfan Coding Plan is the brain of the lobster.”

Technical staff on-site demonstrated the complete technical chain: OpenClaw was pre-deployed on the Baidu Intelligent Cloud server, and users could access it directly through local mapping after purchase. This means developers can obtain a “ready-to-use” AI development environment without complex environment configuration.

The installation process has been simplified to the extreme: After scanning the code to purchase the service package ($9.9 server + $7.9 Coding Plan), technical staff assist in completing the cloud deployment of OpenClaw, after which users can configure port mapping locally or access it directly through a browser. According to a NetEase editor’s experience, the entire installation process can be completed in under ten minutes.

For developers looking to quickly validate AI application prototypes, this “concierge-level” service indeed lowers the trial-and-error costs. However, it is important to note that the renewal price will be key to user retention.

The event did not explicitly disclose the renewal standards, and staff hinted that next month’s activity prices would be announced later. This may be a backup plan for Baidu Cloud.

The Defensive Battle “Caught Up” by Tencent

“It might be that Tencent caught us off guard.”

A Baidu employee’s half-joking remark to a NetEase editor revealed the industry background of this event. Not long ago, Tencent Cloud stirred up a wave of “shearing wool” enthusiasm in the developer community with its aggressive low-price strategy and integration of AI development toolchains.

Baidu’s move clearly has a strong defensive nature. By bundling cloud servers with the AI development platform (Qianfan), Baidu aims to establish user awareness in the emerging battlefield of “AI native application development”—after all, in the price war at the IaaS layer, companies have already become fiercely competitive, and the real differentiation lies in the integration of AI capabilities.

Delivering cloud computing services through “ground promotion” is rare in the industry. Baidu chose to hold the event in its own technology park, targeting employees from neighboring Tencent, Lenovo, NetEase, Sina, and various tech companies, which itself carries a strong implication of “precision targeting.”

This may be the effect Baidu wants—first attracting early adopters with extreme low prices, then retaining users through product experience. However, the essence of cloud computing is long-term service, and the sweet period of $16.8 lasts only one month. Once the trial period ends, if renewal prices return to normal levels (hundreds), developers accustomed to low prices are likely to “abandon ship.”

The image shows the OpenClaw interface after being installed by technical staff on-site. Source: NetEase editor’s on-site photography.

This means Baidu may face large-scale resource idling and data migration costs. More critically, if too many “shearing wool” participants are attracted, it could lead to resource squeezes for actual AI developers who need computing power.

Additionally, data security issues remain difficult to guarantee. The on-site deployment model uses local port mapping for direct access to the cloud, designed for “ready-to-use.” This design, while lowering the usage threshold, also means that there is an unstrictly secured channel established between the user’s local network environment and the Baidu Intelligent Cloud server. For developers handling core business code or sensitive data, this is akin to opening a temporary backdoor in the firewall.

In the afternoon, the booths in front of the K1 building began to dismantle. People returned to their respective workstations with their newly “installed” lobsters (servers + development environment), preparing to start a month of trial use.

This event is like a metaphor: in the current landscape where AI large models are reshaping the cloud computing pattern, giants are competing for developers and users in increasingly down-to-earth ways. When the price of cloud servers is driven down to cheaper than a cup of milk tea, and technical deployment is simplified to a ten-minute process, no one knows whether these “lobsters” will grow into dragons or become a stagnant pool a month later.