At its annual Apsara cloud conference in Hangzhou, Alibaba revealed the Zhenwu V900, billed as China’s most powerful AI processor. The chip is central to Alibaba’s goal of scaling cutting-edge AI models and expanding its global data center capacity to 20GW within six years.
- Zhenwu V900 triples AI chip performance over predecessor
- Plans for 10 trillion parameter AI models and 20GW data center capacity by 2032
- Alibaba commits $53bn to AI/cloud expansion despite profit setbacks
What happened
Alibaba introduced the Zhenwu V900 AI chip at its Apsara cloud conference, positioning it as the most powerful AI processor currently in China. This new chip delivers roughly three times the performance of the earlier Zhenwu M890, featuring 216GB memory and ultra-high inter-chip bandwidth of 1,200 GB/s. It supports a wide range of floating point precisions from FP32 down to FP4, optimized for both high precision AI model training and low precision inference tasks.
The company’s chip design division, T-Head, developed the latest processor, which will be commercially available starting in the first quarter of 2027. Alibaba also shared ambitious plans to scale data center capacity to 20GW globally by 2032, reflecting a massive infrastructure buildout to advance AI research and cloud services. Alongside hardware, Alibaba announced ongoing development of AI models scaling up to 10 trillion parameters, significantly larger than the current flagship Qwen 3.8 Max at 2.4 trillion parameters.
Why it matters
This launch solidifies Alibaba’s leadership in China’s AI chip sector amid stringent US export controls that limit access to foreign advanced AI accelerators and semiconductor manufacturing. Developing high-performance domestic chips is critical for Alibaba to maintain competitive AI training and inference capabilities and circumvent supply chain bottlenecks impacting the sector.
The Zhenwu V900’s integration capability to scale massively—up to half a million chips in a single cluster—positions Alibaba to deliver frontier-level AI workloads and power evolving agentic AI applications. Combined with aggressive AI infrastructure investments totaling over $50 billion, Alibaba aims to capture a significant share of the burgeoning AI cloud market, forecasted to generate tens of billions in annual revenue within the next few years.
What to watch next
Key developments to monitor include the commercial rollout of the Zhenwu V900 chip in early 2027 and the performance of Alibaba’s upcoming AI models, Qwen 4.5 and Qwen 5, which target up to 10 trillion parameters. Success in these areas could establish new benchmarks for AI capability and enterprise adoption across industries such as automotive, natural language processing, and embodied intelligence.
Alibaba’s progress expanding its global data center network will also be pivotal, particularly how it overcomes supply chain constraints noted by CEO Eddie Wu. Additionally, investors will watch Alibaba’s ability to accelerate AI/cloud revenue growth to reach targets of $10 billion this year and $100 billion over five years, recouping recent heavy infrastructure investments. Industry observers may further analyze how these developments position Alibaba relative to rivals amid ongoing geopolitical and technology export dynamics.