Alibaba Rolls Out Wan3.0 AI Video Model

Alibaba officially rolled out its latest AI video generation model, Wan3.0, on Monday with enhanced capabilities, a day after the Chinese internet giant launched a $10 billion share placement to fund rising AI spending.

The new model can generate 30-second videos from documents, spreadsheets, slides and web pages, Alibaba Cloud said in a post on WeChat. The company said Wan3.0 has been used in short drama and film production, advertising and marketing, tourism promotion and music video creation since a public beta version launched on August 6.

Alibaba's Tongyi Lab, the unit behind the Wan model line, said Wan3.0 unifies capabilities that previously required separate models under its predecessor, Wan2.7, into a single system, and supports resolution tiers up to 1080p. The model is available through Alibaba Cloud's Model Studio platform and Qwen Cloud, the company's AI-native cloud marketplace, though access remains limited to invited accounts and no open-weight version has been released.

The release comes a day after Alibaba announced the share sale, which Reuters described as the largest-ever primary follow-on offering by a Hong Kong-listed company, as the company ramps up spending in the intensifying global AI race. Alibaba priced the placement at HK$80 billion, or roughly $10.2 billion, through 710 million newly issued shares at HK$112.70 each, according to a regulatory filing, with the transaction expected to close on August 26. "The Equity Placement is being undertaken to extend the Company's global AI leadership," Alibaba said in the filing, adding that it intends to use all of the net proceeds to expand its AI infrastructure.

Alibaba reported a 75% plunge in quarterly earnings from a year earlier last week, due to soaring AI-related capital expenditure, according to Reuters, even as quarterly revenue rose to nearly $40 billion, up 9% year-over-year, driven in part by AI demand and the popularity of its open-source Qwen models.

The video-generation release also lands amid intensifying competition among Chinese AI labs on generative video specifically. ByteDance's Seedance model reached comparable 30-second clip lengths earlier this year, making native long-form generation a increasingly common baseline capability among leading Chinese video models rather than a standalone differentiator for Alibaba.