Key Highlights
- Alibaba partnered with China Telecom to activate a 10,000-chip AI computing facility in Shaoguan, Guangdong, utilizing Alibaba’s proprietary Zhenwu semiconductor technology.
- This installation represents the inaugural large-scale Zhenwu deployment within China’s Greater Bay Area, capable of training artificial intelligence models containing hundreds of billions of parameters.
- Performance metrics indicate 30% improved training and inference capabilities, with individual card throughput achieving nearly tenfold improvements over earlier systems.
- Future expansion roadmap targets scaling the facility to 100,000 chips, while smaller enterprises gain computing access through China Telecom’s service platform.
- This activation arrives following a comparable 10,000-chip Huawei Ascend 910C installation that became operational in Shenzhen during the previous month.
Alibaba (BABA) together with China Telecom have activated a 10,000-chip artificial intelligence computing installation in Shaoguan, located in Guangdong province. The entire facility operates using Alibaba’s proprietary Zhenwu AI semiconductors, created by its T-Head chip design division.
Alibaba Group Holding Limited, BABA
This activation represents the inaugural deployment of Zhenwu chips at such magnitude within the Greater Bay Area. Alibaba Cloud characterized the development as advancing China’s AI computing capabilities “from high-performance breakthroughs toward widespread industrial deployment.”
ALIBABA LAUNCHES DATA CENTER WITH 10,000 OF ITS OWN CHIPS AS CHINA RAMPS UP AI PUSH
— First Squawk (@FirstSquawk) April 8, 2026
The installation employs an advanced high-performance networking framework enabling all 10,000 chips to function as one unified supercomputer. Alibaba reports this configuration provides 30% enhanced training and inference performance, with individual card throughput increasing nearly tenfold compared to previous-generation systems.
The infrastructure supports training of models containing hundreds of billions of parameters — positioning it alongside some of the most sophisticated AI models under development worldwide.
Network latency measures 4 microseconds, which Alibaba credits to the networking framework connecting the chips. This performance metric proves critical for enterprise AI applications where rapid response times remain essential.
Beijing’s Drive Toward Indigenous AI Computing Resources
This activation aligns with broader governmental initiatives. Beijing incorporated intelligent computing infrastructure within its 15th five-year strategic plan during the previous month, while an August State Council AI directive emphasized optimized expansion of computing resources throughout China.
By the conclusion of June last year, China’s aggregate computing capacity reached 962,000 petaflops — representing 21% of global capacity, reflecting a 73% year-over-year increase, per the China Academy of Information and Communications Technology.
The Shaoguan installation has already entered service across healthcare and advanced manufacturing applications. Small and medium enterprises can obtain computing resources via China Telecom’s platform, with payment options structured per card or hourly usage.
Alibaba confirmed intentions to expand the installation from 10,000 to 100,000 chips. This expansion targets reduced operational costs alongside enhanced resource utilization.
Industry Landscape: Huawei and Domestic Semiconductor Competition
This activation follows a comparable achievement during the previous month, when China’s inaugural 10,000-card intelligent computing installation — constructed with Huawei’s Ascend 910C chips — commenced operations in Shenzhen.
That facility delivers 11,000 petaflops of computing capacity and has been integrated with a separate 3,000-petaflop installation activated in 2024. Shanghai maintains ongoing development of a 10,000-card facility through an INESA state-owned subsidiary, designed for compatibility with multiple domestic chip architectures.
While Chinese semiconductors currently lag behind Nvidia in individual performance benchmarks, Beijing’s approach emphasizes large-scale cluster architecture combined with optimized networking to narrow performance differences.
U.S. export limitations on Nvidia chips have compressed China’s domestic chip development schedule. Alibaba’s T-Head division has served as a cornerstone of this initiative, working alongside Huawei.
BABA recorded gains of 7.79% during the announcement day, with after-hours trading adding 0.82% to its Hong Kong-listed shares (728-HK).
