Alibaba Cloud has signaled a decisive shift in its artificial intelligence strategy, transforming its Bailian 'Model Studio' into an open ecosystem that hosts its fiercest competitors. At the latest Alibaba Cloud Summit, the tech giant announced that Zhipu AI’s GLM-5.1 and other leading third-party models will now be accessible through its infrastructure. This move mirrors the 'supermarket' approach pioneered by global cloud leaders, positioning Alibaba as the indispensable plumbing of China’s generative AI landscape.
While continuing to develop its proprietary Qwen series, the launch of the Qwen 3.7-Max flagship model underscores Alibaba’s dual-track ambition. By providing both the premier home-grown model and the platform for rivals, Alibaba is hedging its bets in a volatile market. The goal is no longer just to build the best Large Language Model (LLM), but to ensure that regardless of which model developers choose, they are running it on Alibaba’s compute.
The hardware layer of this strategy was also fortified with the debut of the 'Zhenwu M890' AI chip. This custom silicon is designed to optimize the performance of the Qwen 3.7-Max and its peers, suggesting a move toward vertical integration that could lower costs for enterprise clients. As the domestic price war for AI tokens intensifies, the ability to offer superior performance-to-cost ratios through specialized hardware will be a critical differentiator against rivals like ByteDance and Huawei.
This consolidation of software, hardware, and platform services marks a new chapter in China’s 'War of a Hundred Models.' By integrating Zhipu AI—one of China’s elite 'AI Tigers'—Alibaba is acknowledging that the future of AI is multi-model. For international observers, this evolution demonstrates that the Chinese market is moving away from fragmented experimentation toward a more mature, infrastructure-centric ecosystem led by established cloud titans.