Alibaba has released Wan3.0, the newest version of its AI video generator, and it can now make clips up to 30 seconds long, twice what the previous model could produce. That alone is a meaningful jump, but the more interesting change is what you can feed into it.
Wan3.0 accepts text prompts, images, video, and audio, and it can also take web pages, PDFs, and PowerPoint slides as raw material and turn them into video. In practice, that means a product sheet, a slide deck, or even a webpage could become a short video without anyone touching a camera or a video editor.
The pricing is transparent and public. A 30-second clip in standard quality runs about 3 dollars, and a sharper, high-definition version costs around 6 dollars, billed by the second through Alibaba's cloud service or its API. Compared to hiring a video production team, that is close to free, which is exactly the point.
The timing is not a coincidence. Alibaba just raised 10 billion dollars through a share sale in Hong Kong, the largest follow-on offering that market has ever seen, and said the proceeds would go into AI infrastructure. Days before that, the company told investors its quarterly profit had fallen 75 percent from a year earlier because of how much it is spending on AI, even though revenue tied to AI products grew sharply. Wan3.0 is part of how Alibaba plans to turn that spending into paying customers.
It also lands in the middle of a genuinely brutal race. ByteDance, Tencent, Kuaishou, and Alibaba are all pushing AI video tools at the same time, cutting prices and adding features to win over advertisers, filmmakers, and social media creators. Kuaishou's Kling has already turned this into real revenue, and rivals are matching its pricing to keep up. That competition is good news for anyone buying these tools, because it means video generation keeps getting cheaper and better at the same time.
The business case goes beyond marketing clips. A company could turn a training manual into a short instructional video, or convert a product catalog into demo footage, without a production budget. Robotics and self-driving car companies are also a target market, since realistic generated video can help train systems to recognize objects and situations without filming everything in the real world.
The catch is quality control. AI video still struggles with faces, hands, and on-screen text staying consistent from one frame to the next, and Alibaba's claim that Wan3.0 fixes this has not been independently tested yet. Anyone using this for a public-facing video should still review the output closely before publishing it, the same way you would check a first draft written by a new employee.