KATHMANDU: Chinese technology giant Alibaba has launched Wan3.0, its latest artificial intelligence video-generation model, just days after raising approximately $10.2 billion through a Hong Kong share placement. The company has earmarked the entire proceeds of the share sale for artificial intelligence investment.
The timing is deliberate. Alibaba has spent the past two weeks seeking to convince investors that its rapidly rising capital expenditure is translating into tangible AI products and services.
Wan3.0 had already been available in public beta since early August through Alibaba Cloud’s Model Studio and Qwen Cloud platforms. Monday’s launch therefore represents an expansion of access rather than the model’s first public appearance.
The launch comes as Chinese AI companies strengthen their position in video generation, a segment that has received less attention in the West compared with the competition around AI chatbots.
Wan3.0 can generate videos of up to 30 seconds in a single pass, twice the 15-second limit of its predecessor, Wan2.7. It supports resolutions of up to 1080p.
Alibaba says the model can maintain consistency in characters, objects, spatial composition and motion graphics throughout a video. It can also generate faces with synchronised micro-expressions and produce voice output in multiple languages.
One of its more distinctive capabilities is the range of material it can accept as input. In addition to text, images, audio and video, Wan3.0 can process web pages and documents, including PDF and PowerPoint files. Users can therefore provide a presentation or other source document and generate a video that follows its structure and content.
That positions Wan3.0 as a potential business tool for marketing departments, advertisers and corporate communications teams, rather than solely as a creative product for filmmakers. Alibaba also identifies short dramas, social media content, autonomous-vehicle simulation and robotics training among potential applications.
Pricing is another part of the company’s competitive strategy. Wan3.0 costs $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p. At the highest resolution, producing one minute of video costs about $12.
However, price alone may not determine its commercial position. Wan3.0 has yet to undergo independent benchmarking, meaning its performance claims remain based largely on Alibaba’s own assessments.
Access is also being rolled out gradually. Beta access is available through Model Studio and Qwen Cloud on an application basis, while full application programming interface access is being introduced progressively. Alibaba is also preparing a consumer-facing website based on a members-only model.
The company has also changed its approach to openness. Alibaba previously built much of its reputation in AI video generation by releasing parts of the Wan series as open-source models. Wan2.2, launched in July 2025, remains its latest major video model released in that manner. Wan3.0’s model weights are closed.
The commercial rationale behind Alibaba’s AI spending is becoming increasingly visible in its financial results. Cloud and AI revenue reached 48.44 billion yuan in the June quarter, up 45 percent year on year. AI model services alone have generated more than 16 billion yuan in annualised recurring revenue, according to the company.
The cost of that expansion is equally significant. Quarterly net profit fell 75 percent as capital expenditure rose 75 percent to 67.68 billion yuan. Alibaba committed 380 billion yuan to AI infrastructure over three years in early 2025 and has reportedly been considering increasing that commitment to 480 billion yuan.
Whether Wan3.0 can justify that level of investment remains uncertain. Video-generation models are expensive to operate, customers can switch providers based on pricing, and independent benchmarks have yet to establish where Alibaba’s latest model stands against its rivals.
Still, the launch sends a clear signal: Alibaba is seeking to turn its massive AI infrastructure spending into commercially scalable products-and video generation is becoming an important part of that strategy.
