Alibaba and China Telecom have officially launched a major AI data center in southern China, powered by 10,000 of Alibaba's own domestically developed 'Zhenwu' chips.
This significant move didn't happen in a vacuum; it's a direct response to escalating U.S. export controls. Since 2022, the U.S. government has tightened restrictions on selling advanced AI chips and related technology to China, making it difficult for Chinese tech giants to acquire the high-performance GPUs they need. This created a powerful incentive for companies like Alibaba to accelerate their own chip development to ensure a stable supply for their growing AI ambitions.
Beijing's national strategy provided the perfect tailwind for this shift. Policies like the 'East-Data/West-Compute' plan are designed to build a national network of computing power, with specific hubs dedicated to handling massive data loads. Shaoguan, where the new data center is located, was designated as the key computing hub for the Greater Bay Area, one of China's most economically vibrant regions. This government backing provided clear direction and support for such large-scale infrastructure projects.
Alongside policy pressure, there was a strong commercial motivation. Alibaba's cloud division has been a standout performer, reporting triple-digit growth in AI-related revenue for ten consecutive quarters. With management setting a bold five-year goal of $100 billion in combined AI and cloud revenue, the business case for investing heavily in proprietary, large-scale computing infrastructure became undeniable. Owning the hardware stack gives them more control over costs and performance.
The final pieces fell into place recently. First, Alibaba's 'Zhenwu' chip proved its viability, with reports of over 100,000 units shipped and adoption by major clients. Second, a direct competitor, Huawei, launched a similar-sized cluster in nearby Shenzhen just days before, igniting a regional 'scale race.' This competitive pressure likely accelerated the timing of Alibaba's launch.
In short, the Shaoguan data center is a culmination of geopolitical pressure, national policy, commercial ambition, and technological validation. It marks a critical step in China's quest for technological self-reliance and signals a new, more intense phase of competition in the country's domestic AI cloud market.
- Glossary
- GPU (Graphics Processing Unit): A specialized electronic circuit designed to rapidly manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device. They are crucial for training large AI models.
- Compute Cluster: A group of linked computers (nodes) that work together closely so that in many respects they can be viewed as though they are a single computer. A '10,000-card cluster' refers to a cluster with 10,000 AI accelerator cards (like GPUs).
- Export Controls: Government regulations that restrict the sale and transfer of certain goods, technologies, and software to foreign countries for reasons of national security or foreign policy.