China's AI virtual character market to hit $3.5b by 2029: report

China’s market for AI-powered virtual characters is expected to reach 25 billion yuan (US$3.5 billion) by 2029, according to a report from International Data Corporation shared at an AI digital human forum in Wuxi, Jiangsu province.

The sector reached 4.1 billion yuan (US$576 million) in 2024, up 85.3% year-on-year.

The forum took place alongside the China International Intelligent Communication Forum 2025.

During the event, the Yangtze River Delta AI Digital Human Industry Alliance was launched to enhance collaboration across the region.

Additional initiatives introduced include an industry-education integration community, a language service internship and training base, and a digital human application laboratory.

A new language service platform base, developed by CCTV.com, Beijing Foreign Studies University, and Jiangnan University, was also announced.

The event also showed the launch of international communication projects for Wuxi, including a new alliance, a documentary project, and an upgraded English website.

🔗 Source: China Daily

🧠 Food for thought

Implications, context, and why it matters.

International Data Corporation (IDC)’s 25 billion yuan outlook lacks detail

Scope spans digital human uses (digital humans are AI-powered virtual characters), yet IDC offers no split for 3D, enterprise versus consumer, or revenue by industry 1.
2D revenue hit 2.89 billion yuan in 2024, up 101.2% 1. The report gives no view on whether that pace beats the 85.3% rate or whether 3D or motion capture (sensor-based body tracking) or voice synthesis close the gap.
Leaders include Baidu plus SenseTime plus Xiaoice yet share splits stay vague, as does differentiation between semantic models (general-purpose language models) versus vertical large models (domain-specific AI systems) 1.
Missing cuts by use case or business model leaves vendors and enterprise buyers guessing where to place bets.

China’s mandatory AI content labels spur compliance demand

Labelling Rules require clear flags on AI-generated or edited text plus images plus audio plus video plus virtual scenes 2. Liability sits with model providers, developers, and platforms.
The Eleventh Batch filing (the latest list of algorithms registered with the Cyberspace Administration of China [CAC]) counts 211 entries 3. Deep synthesis providers with public opinion attributes or social mobilisation capabilities (features that can influence or rally the public) must register, so enterprises add provenance SDKs (software development kits for tracking content origin) and labelling tools.
Platforms for virtual characters or customer-facing digital humans need built-in detection, audit trails, and labelling APIs (application programming interfaces) for compliance, creating room for B2B (business-to-business) compliance infrastructure providers.
International SaaS (software-as-a-service) vendors targeting China must retrofit products with CAC-compliant labels, opening deals with local labelling and detection specialists who know the proposed labelling standards 4.