TNW→ original

Alibaba releases Wan3.0 video model days after raising $10 billion for AI

Alibaba has launched Wan3.0, the next generation of its video generation model, capable of producing clips up to 30 seconds long. The release came just days after the company raised $10.2 billion in a record share sale in Hong Kong, and it is directing all of that money to AI. Chinese reviewers praise the model for its stability and realistic imagery.

AI-processed from TNW; edited by Hamidun News
Alibaba releases Wan3.0 video model days after raising $10 billion for AI
Source: TNW. Collage: Hamidun News.
◐ Listen to article

Alibaba has launched Wan3.0, a new version of its video generation model that can create clips up to 30 seconds long. The release came just days after the company raised about $10.2 billion in a Hong Kong share offering earmarked entirely for funding its AI business, The Next Web reports.

Money and model, hand in hand

The Next Web says the sequence of events is no coincidence: over the past two weeks, Alibaba has been trying to convince investors that its AI capital spending is turning into products, and Wan3.0 is the first tangible piece of evidence. The offering was a record for Hong Kong: according to SCMP, the company issued new shares worth HK$80 billion, and the entire amount is going toward global AI expansion.

  • Wan3.0 generates clips up to 30 seconds long, compared with the typical 5-10 seconds for most models
  • The $10.2 billion offering is the largest share sale in the history of the Hong Kong exchange
  • All proceeds are directed to AI: data centers, models, and global infrastructure

What Wan3.0 can do

Wan is Alibaba's open family of video models, and earlier versions were downloaded millions of times in the open-source community. The third generation bets on length and coherence: 30-second clips move the model out of the "social media clip generator" category and into a tool for advertising, product previews, and short-form storytelling.

Chinese trade publication Jiqizhixin relays the first industry assessments of three qualities of the model: motion stability, realism, and a "feel for texture" in the imagery. Coherence across long scenes is exactly where most video models stumble: holding a character, lighting, and the physics of a shot steady for 30 seconds is an order of magnitude harder than doing so in a five-second clip.

What this means

The video model race has entered a phase where infrastructure wins, not demos: OpenAI with Sora, Google with Veo, and Alibaba with Wan are no longer competing on the beauty of individual clips, but on length, controllability, and the cost of generation. Alibaba has two systemic advantages here: its own data centers and an open distribution model that makes Wan the default for developers unwilling to pay for Western APIs. The fresh $10 billion turns this strategy into a question of execution rather than resources.

Frequently asked questions

How does Wan3.0 differ from Sora and Veo?

The main difference is openness: earlier versions of Wan were released with open weights and became the standard for local video generation. Length is another factor: 30-second clips sit at the upper end of what commercial video models can do today.

Why did Alibaba raise $10.2 billion?

The company has officially stated that the entire offering will go to its AI business: computing infrastructure, model development, and the international expansion of its cloud business.

ZK
Hamidun News
AI news without noise. Daily editorial selection from 50+ sources. A product by Zhemal Khamidun, Head of AI at Alpina Digital.
✦ Unified AI Workspace

Access Frontier LLMs & Multimodal AI in One Workspace

Discover NomadNet AI (nomadnet.ai) — unified access to frontier language models, research synthesis, and multimodal workflows.

Try NomadNet AI →Launch NomadNet Agent
What do you think?
Loading comments…