Alibaba Hong Kong stocks surged over 2%, Alibaba Cloud announces price increases of up to 34%

On March 18, Alibaba’s Hong Kong stocks experienced a short-term rally in the afternoon, currently up over 2%, after initially falling more than 1% in the morning; Alibaba’s U.S. stocks rose over 2 in after-hours trading. According to news, Alibaba Cloud announced today that due to the explosive global demand for AI and rising supply chain costs, prices for Alibaba Cloud AI computing power, storage, and other products have increased by up to 34%. Among them, computing cards like Pingtouge Zhenwu 810E increased by 5%–34%, and the file storage product CPFS (Intelligent Computing Edition) rose by 30%.

Following the announcement, the concept of computing power leasing in A-shares surged again in the afternoon, led by Alibaba Cloud. DataPort (603881) hit the daily limit, with Qingyun Technology, UCloud, Hanggang Steel (600126), Hongjing Technology (301396), and Xiechuang Data (300857) quickly following suit.

According to sources cited by Caixin, another major reason for the price hikes is the “surge in token call volume.” Alibaba Cloud’s MaaS business, Bailian, achieved its highest growth rate from January to March this year. Alibaba Cloud is shifting its scarce AI computing resources toward token-related businesses.

“Wukong” Strikes OpenClaw

On March 17, at the AI DingTalk 2.0 annual product launch, DingTalk founder and CEO Chen Hang expressed concerns about the current “shrimp farming” craze. On that day, DingTalk officially launched the AI-native work platform “Wukong,” aimed at creating a “24-hour lobster army” for enterprises. This move marks Alibaba’s strategic deployment into the enterprise AI market. When the AI Agent (artificial intelligence entity) wave triggered by OpenClaw arrives, Alibaba has chosen a different approach. The former causes a surge in computing power and “token inflation,” while the latter aims for a unified output to provide measurable AI services for enterprises.

This is a key step in Alibaba’s pursuit of differentiation. On the eve of the launch, Alibaba announced the formation of the Alibaba Token Hub (ATH) business group, led personally by CEO Wu Yongming. The Qianwen Business Unit, Wukong Business Unit, and others were incorporated into this group to establish a new organization focused on “creating tokens, delivering tokens, and applying tokens.” It appears Alibaba is not simply developing a new application but intends to connect all B-end capabilities to build a complete ecosystem for the AI Agent era.

Behind these moves is Alibaba’s urgency in facing the “token inflation” of the Agent era. It is abandoning its relatively scattered exploration over the past few years and trying to concentrate its resources into a single “factory.”

ATH integrates the Tongyi Laboratory, MaaS business line, Qianwen Business Unit, the newly revealed Wukong Business Unit, and the AI Innovation Business Unit. Its core goal is summarized as: create tokens, deliver tokens, and apply tokens.

This is Wu Yongming’s third major organizational adjustment since becoming Alibaba’s CEO in September 2023, and it may also be a decisive move for Alibaba’s future over the next decade.

In an internal letter, Wu Yongming assessed that the current period is the night before the explosion of AGI, with hundreds of billions of AI Agents supporting digital work, all running on tokens generated by models.

How to understand this era? Recently, the rapid rise of “lobster” is a typical example.

This open-source project called “OpenClaw” has become popular in the developer community. It is essentially an operating system for intelligent agents, allowing developers to use a few lines of code to enable AI to autonomously call tools and perform tasks. Nvidia founder and CEO Jensen Huang mentioned this project in his GTC speech, calling it “one of the most popular open-source projects in human history.”

Although safety concerns remain controversial, its industry potential is widely recognized, and the barrier to entry for intelligent agents is rapidly lowering. An insider from a domestic large-model enterprise told 21st Century Business Herald that they observe these intelligent agents growing at hundreds, thousands, or even tens of thousands of times per day.

Behind these complex tasks is the consumption of massive amounts of tokens. “The future growth will be enormous, several times or even dozens of times more, so we expect the entire computing power calls and token consumption to increase several times or more,” said the industry insider.

Alibaba is quick to sense the temperature.

During the Spring Festival in 2026, Alibaba Cloud launched a subscription service called Coding Plan. Driven by “lobster,” demand for computing power experienced an unexpected leap, making Coding Plan one of Alibaba Cloud’s fastest-growing products in history. The originally planned “first purchase discount” had to be discontinued after two weeks.

According to IDC, by 2030, the global active AI agents will reach 2.216 billion, with annual token consumption soaring from 0.0005 Peta Tokens in 2025 to 152,000 Peta Tokens, an increase of over 300 million times.

However, the business of fueling these AI models is not easy. Jensen Huang proposed the “Token Factory Economics” at GTC, stating that future data centers will no longer be just warehouses for storing files but factories for producing tokens. Under fixed electricity costs, the entity with the highest token throughput per watt will have the lowest production costs.

China’s AI industry chain has demonstrated advantages over the past six months, giving Alibaba hope. Thanks to cheap electricity and a full supply chain, Chinese large-model vendors have priced their million-token packages at less than one-twentieth of their American counterparts. Alibaba’s Qianwen previously priced its Qwen 3.5 at just 0.8 RMB per million tokens.