HONG KONG — Alibaba has introduced new artificial intelligence chip technologies and outlined plans for more powerful models ahead of a high-profile meeting between Chinese and U.S. leaders, where AI competition is expected to be a central topic.
The announcement comes as prominent U.S. AI figures, including Anthropic CEO Dario Amodei, warn that China’s rapid AI advancement threatens the United States and urge a broader slowdown in the technology’s development.
Chinese President Xi Jinping is scheduled to visit Washington on Wednesday for a state summit with U.S. President Donald Trump, where discussions on AI, trade, and tariffs are anticipated.
During its annual flagship conference in Hangzhou, Alibaba introduced a new AI chip and detailed its roadmap for the next generation of AI models.
CEO Eddie Wu stated that the newly launched Zhenwu V900 chip is currently the most powerful AI chip in China, offering triple the performance of its predecessor, the Zhenwu M890.
These chips are essential for training frontier AI models and powering “inference”—the computational processes a trained AI uses to generate user responses. Alibaba utilizes its Zhenwu chips in its data centers to provide computing resources for both the company and its cloud clients.
Alibaba also revealed plans to train a new AI model with five to ten trillion parameters, a metric reflecting an AI’s learning capacity. This compares to its current Qwen3.8-Max model, the most advanced in the Qwen series, which features 2.4 trillion parameters.
For context, Chinese AI firm Moonshot claims its Kimi K3 model, released in July, is the world’s largest open model with 2.8 trillion parameters.
Additionally, Alibaba emphasized its strategy to expand the data centers underpinning its cloud computing operations, citing exponentially growing demand for AI computing power. The company projects it will surpass 20 gigawatts of computing capacity by 2032.
To put this in perspective, SpaceX currently operates roughly 1.4 gigawatts of AI computing capacity and aims to exceed 10 gigawatts by 2027.
Analysts suggest that recent Chinese advancements in AI are providing Beijing with increased leverage in its upcoming negotiations with Washington, as China intensifies its push for technological self-reliance despite U.S.-led export restrictions that have blocked access to advanced AI chips and semiconductor manufacturing equipment.
Furthermore, open Chinese models—often more cost-effective than proprietary closed models from leading U.S. AI labs—are gaining traction globally, including within the United States.
Chinese tech giant Huawei also unveiled new chip technologies last week as it seeks to compete with global industry leaders like Nvidia.
CEO Wu declared that Alibaba is “mobilizing every resource” to fulfill customer demand for AI.
He added that global supply chain shortages across the AI data center sector are currently restricting the pace at which the company can scale its compute infrastructure.
Wu also drew a parallel between AI growth and the Industrial Revolution, asserting that “machine intelligence” will rapidly surpass human cognitive capacity.
While acknowledging that “machine intelligence today is not a substitute for human intelligence,” he predicted that machines will eventually produce over 1,000 times more “thinking” than all of humanity combined, a significant leap from the less than 3% currently generated.