Kuaishou’s Kling AI platform generates videos from text and still images.
nuphoto | nuphoto | Getty Images
Beijing – China’s video-rich entertainment world is gaining corporate data. We are currently increasing the number of artificial intelligence tools that make money to generate ads and film clips.
Tiktok Parent Bytedance holds the first and third spots of the top-rank text-to-video generation AI model of research firms’ artificial analysis that began in the past two months. Google holds the second and fourth spots, while Kling AI from Beijing-based short video app Kuaishou ranks fifth.
Wei Xiong, China Internet Analyst at UBS Securities, said:
“We believe AI video generation has the potential to rebuild the content industry,” she said. “By increasing production efficiency, lowering barriers to creation and unlocking new monetization models.”
With such AI tools, users can upload a single image or multiple images and tell the AI to generate video clips based on them. Other tools allow users to enter text, from which AI generates video clips.
More than 20,000 companies, from advertisers to film animators, are already using Kling AI to generate videos, the Beijing-based company claimed at the World AI Conference in Shanghai this week. The latest version of Kling 2.1 allows you to automatically add relevant sound effects to match the AI-generated video.
It’s not just Chinese users.
“Whether it’s user scale or commercial revenue, it’s the majority overseas,” Kling AI’s operations manager Zeng Yushen told CNBC, Mandarin’s CNBC, translated by CNBC. She said the company plans to increase support for the tool in places such as Japan, Korea and Europe.
“This is something we’ve observed, and the big models of AI are becoming increasingly globalized,” she said. “People don’t seem to care which country’s product is.”
Kuaishou claims that Kling Ai made 150 million yuan ($20.83 million) revenues during the first three months of the year, with daily advertising spending of 30 million yuan during that period. The company has not yet announced when it will release its second quarter results. Zeng refused to share the model training costs for Kling AI.
While reducing production costs means a “slight” market, UBS’Xiong says, “current model features remain constrained by clip length, motion consistency and controllability.”
Chinese video AI companies also face competition with the US, surpassing the Trump administration’s restrictions on China’s access to the sophisticated semiconductors required to train AI models.
Amazon and Google have launched tools to generate videos from images and text. This release comes almost a year after Microsoft-backed Openai launched its video-generating model SORA to ChatGPT subscribers in December, making its capabilities revealed in February 2024.
However, Kling AI was already open to the public in June 2024. Users subscribe and purchase credits to generate videos.
Vidu, a rival tool for Beijing-based startup Shengshu, launched to global users about 12 months ago and said it expects annual revenue of $20 million in annual revenue based on user subscription fees around March this year.
“Chinese companies tend to first try to identify commercial ‘problems’… The regions where companies pay for services are a challenge for AI applications,” said Paul Triolo, China’s partner and senior vice president of advisory firm DGA-Albright Stonebridge Group.
He pointed out how Chinese startup 3DStyle uses generated AI to design new clothing styles and integrate them with automated manufacturing, internet-connected.
Although U.S. companies also apply AI to certain industries, Triolo said, Chinese companies face a very competitive environment and are often able to integrate AI more quickly as they can be recruited from a “very qualified” local base of software engineers.
“AI as a filmmaker”
Chinese e-commerce giant Alibaba is continuing its trends this week by releasing the latest version of its video-generating AI model called WAN2.2. The company claimed that the open source model allows users to control lighting, time, color tone, camera angle, frame size, configuration and focal length.
Open source allows users to download the model for free and customize the product if not commercialized. Since opening the “WAN” model series in February, Alibaba claimed that the model has been downloaded more than 5.4 million times from the Hugging Face Platform, and a similar model has been downloaded in China, known as Modesscope.
“The era of AI in films is over. We’re entering the era of AI as filmmakers,” said Winston MA, adjunct professor at the NYU School of Law. He pointed out that China’s 1.4 billion population gave local businesses “a huge” amount of video viewing data.
“Just Tiktok took the global market by storm with short videos in the mobile internet age, Chinese AI companies could successfully lead the generative AI revolution in visual digital entertainment.
Avatars and games
Chinese companies are building AI tools, not just generating videos.
Last week, Baidu announced that the latest AI-powered digital technology, which won $7.65 million sales during a June interactive live streaming session, will be released in October for use in a wider industry.
In 3D visualization, Tencent has released the Hunyuan World Model for creating digital panoramic images of scenes generated from text and visual prompts. Visual uses a “mesh” file format that gamer developers can use to edit specific parts of an image.
“Beyond support [Tencent’s] Daniel Ahmad, Director of Research and Insights at NIKO Partners, said:
Weekly analysis and insights from Asia’s biggest economy in your inbox
Subscribe now
Niko has discovered that over half of Chinese game development studios are already using AI to generate content and reduce development time and costs.
However, game development reflects the broader challenges of using AI at scale to generate video and graphics.
“While there’s a lot of interest in AI, we’ve already seen the backlash from games that have poorly implemented technology.”
