ByteDance’s new model sparks stock rally as China’s AI video battle escalates

TikTok parent ByteDance’s pre-release of its latest artificial intelligence video model, Seedance 2.0, over the weekend has sent ripples across the tech and video production sectors as competition among Chinese tech titans’ AI apps intensifies. The model – still in beta and available to select users of Jimeng AI, ByteDance’s AI video platform – delivered enhanced lifelike video outputs that blurred the boundary between AI content and reality, offered smoother camera movement and improved visual consistency, according to early adopters. The buzz has sent stock prices rallying across some Chinese media and AI app firms, with analysts hopeful that AI could be further harnessed to benefit traditional filmmaking and content creation. Shares in studio company Huace Media and game developer Perfect World rose around 7 per cent and 10 per cent on Monday, while publishing house COL Group hit its 20 per cent daily price ceiling. Seedance 2.0 also featured multimodal input, supporting text, image, audio and video-based content generation, making it a handy yet powerful tool for creating video clips that would traditionally have required professional video production and editing skills. It has drawn widespread attention and early praise, with testers saying they were amazed at its ability to generate lifelike characters as well as provide improved and precise control over editing parts of the video content. “With its reality enhancements, I feel it’s very hard to tell whether a video is generated by AI,” said Wang Lei, a programmer from China’s southern Guangdong province. Wang described using simple text prompts to produce a 10-second clip charting human history from prehistoric times to the modern era, praising the result as “smooth in storytelling with cinematic grandeur”. He credited the vast video data resources available through Douyin – China’s largest short-video app and TikTok’s Chinese counterpart – with helping ByteDance train the model. Some Chinese tech entrepreneurs are raving about it. Feng Ji, founder and CEO of Game Science, the Hangzhou-based developer of Chinese blockbuster game title Black Myth: Wukong, said it was the “best video generation model on the planet”. Feng predicted that AI video advancements would lower the barrier for anyone to create professional-looking content at low cost, bringing seismic change to traditional organisations and workflows in the video content sector. He cautioned, however, that such progress could also lead to the spread of deepfakes – images, video and audio depicting real people – and raise trust issues. Users have shared creations ranging from TV commercials to film clips with characters replaced by their own faces. Analysts at Kaiyuan Securities wrote in a research note on Sunday that Seedance 2.0 could help AI manga, short drama and film sectors gain momentum in commercialisation. “By giving users ‘director-level’ precise content control, [Seedance 2.0] is expected to be widely adopted first in AI manga and short drama areas, bringing down costs and improving efficiency to benefit companies with large holdings of intellectual property or traffic,” according to the note. A video production expert in Beijing, who requested anonymity as he was not authorised to speak with the media, said the model had now reached a point with the potential to disrupt the traditional industry. He added that with AI rapidly becoming versatile in audiovisual language and special effects – which could be a gain for the industry – what remained for creators was to focus on creativity and content expression. Chinese tech firms are competing head to head in the AI video area. ByteDance’s advance came days after rival Kuaishou Technology unveiled its Kling 3.0 model series, including Kling Video 3.0, which also supports multimodal input and output across text, image, audio and video content formats with a streamlined workflow aimed at industry use.