Alibaba Group introduced its most advanced artificial intelligence (AI) chip to date on Tuesday, as well as ambitious plans to train a massive next-generation AI model.
The blitz reflects China’s accelerating push toward technological self-reliance, coming just days ahead of a closely watched summit between Chinese and U.S. leaders.
Speaking at Alibaba Cloud’s annual flagship conference in Hangzhou, company CEO Eddie Wu introduced Zhenwu V900, calling it the “most powerful AI chip in China today.” Designed to handle both frontier model training and inference, the new processor delivers three times the performance of its predecessor, the Zhenwu M890, which was released just four months ago.
A single cluster built on the V900 can link up to 500,000 cards to operate as a unified system, according to the company. Mass production and commercial release are scheduled for the first quarter of 2027.
Additionally, Alibaba signaled major software ambitions with plans to train a new AI model featuring between 5 trillion and 10 trillion parameters — a key measure of an AI system’s processing capacity.
By comparison, Alibaba’s current flagship model, Qwen 3.8-Max, holds 2.4 trillion parameters, while Beijing-based startup Moonshot AI’s Kimi K3 stands at 2.8 trillion. The proposed scale would move Alibaba significantly closer to matching the world’s most advanced frontier models.
To power these initiatives, Alibaba aims to expand its data center capacity to over 20 gigawatts by 2032, addressing what Wu described as “exponentially” rising demand for AI computing. To fund its broader AI expansion, the firm recently raised $10.2 billion through a Hong Kong share offering and committed more than $53 billion over three years to integrate AI into its core operations. Following Tuesday’s presentation, Alibaba’s Hong Kong-listed shares rose as much as 5.1%.
(Separately, Xiaomi has unveiled MiMo-V2.6-Pro, the latest iteration of its flagship AI model lineup, which has claimed the title of the world’s top-performing open-weight language model. According to third-party benchmarking firm Artificial Analysis, MiMo-V2.6-Pro secured the highest score among open-weight systems on its Intelligence Index, logging a benchmark score of 46.)
Stephanie Walter, practice leader for AI Stack & Enterprise Application Development at HyperFRAME Research, sees Alibaba’s moves as part of an AI sovereignty strategy.
“Alibaba is building domestic options across chips, models, cloud infrastructure, and data centers so China’s AI development is less vulnerable to foreign suppliers and U.S. export restrictions,” Walter said. “Alibaba still needs to show that these investments translate into reliable, scalable AI services. A larger chip or model does not necessarily deliver better performance, lower costs, or more useful applications. But the direction is clear: the global AI stack is becoming more regional, with China building an alternative ecosystem around the technology and capacity available to it.”
The technological showcase comes at a critical geopolitical juncture. Chinese President Xi Jinping is set to arrive in Washington, D.C., on Wednesday for a state visit and summit with President Donald Trump, where AI competition, trade, and tariffs will take center stage.
Washington has maintained strict export controls to block Chinese firms from purchasing high-end chips from vendors like NVIDIA Corp., as well as critical semiconductor manufacturing equipment. Despite those barriers, Chinese technology companies — including Alibaba and rival Huawei — are rapidly advancing domestic alternatives to narrow the technology gap.
Analysts note that China’s progress in open-source AI models and native hardware grants Beijing added leverage in trade negotiations, even as supply chain constraints and foundry limits pose ongoing challenges.
Highlighting the long-term stakes, Wu likened the rise of AI to the Industrial Revolution, predicting that machine intelligence will eventually outpace human cognitive capacity by a factor of 1,000. He noted that Alibaba is currently “mobilizing every resource” to meet surging client demand and overcome global supply chain bottlenecks.
