The Chinese tech giant revealed a new chip and plans for trillion-parameter AI models at its Hangzhou conference, days before Xi Jinping's Washington visit where AI competition, trade and tariffs are expected to dominate discussions.
Alibaba introduced new artificial intelligence chip technology and outlined ambitions for more advanced AI models at its flagship annual conference in Hangzhou on Tuesday (September 22), timed just days before Chinese leader Xi Jinping's state visit to Washington for talks with President Donald Trump.
CEO Eddie Wu described the company's new Zhenwu V900 chip as the most powerful AI chip currently produced in China, claiming it offers three times the performance of its predecessor, the Zhenwu M890. These chips support both the training of frontier AI systems and "inference"—the computational process behind an AI's responses to user queries—and are deployed across Alibaba's data centers to serve the company and its cloud clients.
The company also announced plans to train a future AI model with five to 10 trillion parameters, a measure often used to gauge an AI system's learning capacity. That would represent a substantial leap from its current flagship, Qwen3.8-Max, which has 2.4 trillion parameters. For comparison, Chinese firm Moonshot has said its Kimi K3 model, released in July, holds 2.8 trillion parameters, making it the largest open model globally.
Alibaba further signalled plans to scale up its cloud computing infrastructure, pointing to rapidly increasing demand for AI computing power, and said it aims to exceed 20 gigawatts of computing capacity by 2032.
Chip race intensifies amid US-China tensions
The announcement lands as American AI executives, including Anthropic CEO Dario Amodei, warn that China's AI progress poses risks to the U.S. and have called for the industry to slow its pace of development. Analysts suggest China's recent gains could strengthen Beijing's position heading into talks with Washington, even as U.S. export controls continue to restrict Chinese access to advanced chips and chipmaking equipment.
Huawei also unveiled new chip technology last week, positioning itself as a challenger to established players like Nvidia. While Chinese AI labs have historically depended heavily on Nvidia hardware for training frontier models, domestically designed chips are gaining traction as Nvidia faces restrictions on selling its most advanced processors in China.
Neil Shah, vice president at Counterpoint Research, said boosting computing capacity helps China offset limitations on chip access. Fellow analyst Parv Sharma noted that closing the broader AI gap with the U.S. will also hinge on China's progress in chip manufacturing, not just design.
Wu compared the current AI boom to the industrial revolution, predicting that machine "thinking" capacity will eventually dwarf that of humanity, even as supply chain shortages continue to constrain how quickly Alibaba can expand its infrastructure.
