Alibaba Group has unveiled the Zhenwu M890 AI processor and a new data center powered by 10,000 of its own chips [1].
This move signals a strategic shift toward a fully integrated AI stack. By developing its own hardware, Alibaba aims to lessen its dependence on U.S. chip technology as export controls tighten on high-end semiconductors.
The company announced the Zhenwu M890 processor in early April 2026 [2]. The chip is designed specifically to support autonomous AI agents, which are systems capable of performing complex tasks with minimal human intervention. As part of this hardware push, Alibaba also revealed a server system featuring 128 chips [3].
To deploy this technology at scale, Alibaba launched a new AI data center in southern China [4]. The facility is operated jointly with China Telecom and utilizes 10,000 home-grown chips to power its operations [1]. This infrastructure allows the company to bypass the need for foreign hardware when training and deploying its latest agentic AI models.
While the company is advancing its hardware capabilities, the financial markets have shown a mixed reaction. Following the chip announcement, Alibaba's forward price-to-earnings multiple stood at 16.7× [5].
Some reports have referred to the new agent-centric hardware as the XuanTie C950 [5], though other industry sources identify the processor as the Zhenwu M890 [1]. The discrepancy highlights the rapid pace of development within the company's semiconductor subsidiary, which includes the Alibaba DAMO Academy.
By controlling the entire pipeline, from the silicon design to the data center operations, Alibaba is attempting to insulate its AI ambitions from geopolitical volatility. The integration of the M890 chips into a massive cluster allows the company to optimize software and hardware simultaneously, potentially increasing the efficiency of autonomous agents.
“Alibaba Group has unveiled the Zhenwu M890 AI processor and a new data center powered by 10,000 of its own chips.”
Alibaba's transition to an in-house AI stack represents a critical step in China's effort to achieve technological sovereignty. By designing chips specifically for autonomous agents, the company is not just replacing U.S. hardware but is tailoring its architecture for the next generation of AI. This reduces the impact of U.S. export restrictions and creates a closed-loop ecosystem where hardware and software are co-optimized for performance.
