Alibaba Takes On Gemini With 30-Second AI Video

Alibaba's new Wan 3.0 is emerging as a serious competitor to Google's video-generation stack, though the two currently play to different strengths. Wan 3.0's standout differentiator is document-to-video: it can transform PDFs, PowerPoints, Word files, spreadsheets, web pages, text, images, audio, and existing video into a single coherent video up to 30 seconds long, with native audio-video generation, 1080p output, and consistency across characters, objects, voices, and scenes throughout the sequence.

Google, by contrast, offers a broader creative ecosystem spanning Gemini Omni, Veo 3.1, Flow, YouTube, and Google Vids. Gemini Omni enables conversational video generation and editing from text, images, and existing footage, while Veo 3.1 delivers strong reference-image consistency, vertical 9:16 output, and upscaling to 4K. Google also embeds SynthID watermarking for AI-content identification — a provenance feature Alibaba currently lacks.

The most disruptive aspect of Wan 3.0 isn't video generation itself, but its ability to convert business documents directly into relatively long, coherent audiovisual content. An enterprise could feed the model a 20-page product presentation or marketing report — Alibaba supports inputs up to 50 pages and 100MB — and receive a short promotional or explanatory video in return, making it particularly compelling for enterprise marketing, training, and product demonstrations.

Google currently holds the edge in ecosystem maturity, creator workflows, 4K capability, YouTube integration, and content provenance, while Alibaba has moved aggressively toward "anything-to-video" and document-to-video automation.

The larger shift is significant: AI video generation is moving from prompt-to-video toward information-to-video, with the next competitive battleground centered on which platform can most effectively turn enterprise knowledge into compelling multimedia content.