Chinese AI video generation has overtaken the US: Here’s why ByteDance and Kuaishou are winning

Chinese AI video generation has overtaken the US: Here’s why ByteDance and Kuaishou are winning

For a long time, the rivalry associated with AI technology was believed to be an exclusively American issue. OpenAI, Google, Anthropic – these are only a few names among those who have been actively conducting research on the topic and developing innovations, whose offices are situated in Silicon Valley. Nevertheless, there is one area within AI technology development that has seen the tables turned. Concerning the development of AI generated videos, China is not following behind. On the contrary, China is ahead.

Digit.in Survey
✅ Thank you for completing the survey!

Also read: What refrigerator capacity do you actually need?

Seedance 2.0 from ByteDance and Kling from Kuaishou now occupy the first places in the ratings of independent video models, outperforming their Western counterparts in terms of realism, accuracy, and usability. Representatives of companies working with these innovations pay attention to the issue specifically, saying that American AI models are unable to generate good videos.

How did this happen?

That unique data advantage that cannot be replicated by anyone else. In reality, it is never about who can build a better model. It has always been a game of data – and China held a natural advantage.

ByteDance controls TikTok, while Kuaishou manages one of the largest short-video social networks. They have an extensive database consisting of not only high-quality videos but behavioral data too. Every like, swipe, or sharing adds value to their data. While text data can be collected at scale by web scraping, it is virtually impossible for any other competitor to collect as much video data from scratch. The only exception to this might be Google’s Veo 3 built on YouTube videos, albeit with restrictions in the type of content used.

Also read: Best home projector under Rs 20,000 in 2026

OpenAI’s Sora model, on the other hand, was shut down back in March of this year because of prohibitively expensive compute requirements for video models.

Looser guardrails, lower prices

However, apart from the statistics, there are two other very practical reasons why Chinese services are outperforming competitors among creators – they are cheaper, and they comply with more requests. Apparently, the US services tend to deny all requests for using their software free of charge without providing any reasoning, arguing that it contradicts their Terms of Service agreement.

Practical effects are already present. According to one source, a retailer made one purchase for 100,000 artificial intelligence videos – an amount that was considered impossible just a year ago. Advertising, e-commerce, and entertainment industries are some of the fields that have been impacted by this change.

While the US continues leading in terms of language models and programming languages, the picture becomes entirely different in regards to video content.

Also read: Tata-ASML deal: Why it matters for India’s semiconductor ambitions 

Vyom Ramani

Vyom Ramani

A journalist with a soft spot for tech, games, and things that go beep. While waiting for a delayed metro or rebooting his brain, you’ll find him solving Rubik’s Cubes, bingeing F1, or hunting for the next great snack. View Full Profile