Alibaba’s Wan3.0 AI Video Model: 30-Second Generation, Document Inputs and the Bigger AI Bet

Alibaba officially rolled out Wan3.0 on August 24, 2026, turning an AI video model that had been in public beta since August 6 into a visible part of the company’s broader AI product strategy. The timing was significant: Alibaba had just arranged an HK$80 billion ($10.2 billion) share placement for AI development, while rising infrastructure spending was weighing heavily on near-term earnings. 51

What Wan3.0 adds

The headline upgrade is longer single-pass generation. Wan3.0 can create a video of up to 30 seconds in one generation, compared with Wan2.7’s reported 15-second limit. Alibaba also highlights “reality-grade” rendering and an expanded Omni-Reference system. 530

Instead of relying only on conventional prompts and media references, users can provide:

• Text
• Images
• Audio
• Video
• Documents
• Spreadsheets
• Presentation slides
• Webpages

That input range gives Wan3.0 a more practical document-to-video workflow. A product team, marketer, or creator can start with source material such as a slide deck, spreadsheet, PDF, or webpage rather than manually converting every detail into a text prompt. The available launch material confirms the document and webpage inputs, while the broader file-format details vary across third-party descriptions. 530

How the model was used during beta

Wan3.0 entered public beta on August 6 through Alibaba Cloud Model Studio and Qwen Cloud. 30 Alibaba said that, during the beta period, the model had been used for short dramas and films, advertising and marketing, tourism promotion, and music-video creation. 5

Those categories suggest Alibaba is positioning Wan3.0 for both creative production and business communication. However, the available reporting does not provide a verified usage total, revenue figure, or independently confirmed list of commercial customers. The reported applications should therefore be treated as Alibaba’s account of early use, not as a measured adoption benchmark.

Why the launch was tied to Alibaba’s fundraising

The Wan3.0 rollout followed Alibaba’s HK$80 billion share placement, worth about $10.2 billion. The company sold 710 million shares at HK$112.70 each, and the proceeds were earmarked for artificial-intelligence development and broader “full-stack” AI capabilities. The transaction was described as the largest primary follow-on offering by a Hong Kong-listed company. 2310

The financing helps explain why a model launch attracted attention beyond the AI industry. Wan3.0 is a concrete product that illustrates what Alibaba is trying to build with a much larger AI infrastructure and services budget. The share sale, by contrast, shows the scale of capital the company believes is necessary to compete in AI.

The earnings trade-off behind the AI push

Alibaba’s AI expansion is already carrying a material short-term cost. The company reported a 75% year-over-year decline in quarterly net profit, while cloud revenue rose 45% and capital expenditure increased 75%, according to Reuters. 1

That makes Wan3.0 part of a larger investment test rather than an isolated feature release. Alibaba is spending heavily on infrastructure, cloud capacity, and AI models with the expectation that those products will support future growth. The immediate results show the tension clearly: AI-related activity is strengthening the company’s strategic focus and cloud business, but the required spending is pressuring profitability.

What to watch next

The most important question is not simply whether Wan3.0 can produce a 30-second clip. It is whether Alibaba can turn document-to-video generation and other multimodal capabilities into repeatable demand from creators, businesses, and cloud customers.

For now, the evidence supports a measured conclusion: Wan3.0 expands Alibaba’s video-generation offering and had reported early use across several production categories, but public information does not yet establish the scale or commercial durability of that adoption. Its launch is best understood as an early product signal within Alibaba’s far larger—and expensive—bet on AI and cloud computing.