The startups powering Asia’s AI-generated video boom

AI-generated videos are back in the spotlight, sparked by the buzz around Google Veo 3’s May release. On October 1, OpenAI followed suit with the launch of Sora 2, the latest version of its video generation model.

Currently, the Sora mobile app is available only in the US. However, OpenAI has unlocked the Sora 2 API for developers, allowing users to access the model globally via AI video platforms such as Higgsfield and Invideo.

Meanwhile, China continues to hone its AI video generation chops, with models developed by Kuaishou (Kling), MiniMax (Hailuo), Alibaba (Wan), and ByteDance (Waver) dominating Artificial Analysis’ leaderboard as of September, both for text-to-video and image-to-video.

It’s no coincidence that these companies, except for MiniMax, operate their own video platforms. This has most likely provided each firm with a large trove of training data to continuously improve its model.

But scaling these systems isn’t cheap. The Economist estimates that, based on pricing for the first version of Sora, OpenAI spends US$1 on computing power for every video generated by its model, and each user can produce up to 100 videos daily.

Chinese companies have also started seeing the potential economics of AI-generated videos. Kuaishou, for example, reported that Kling’s annual recurring revenue had surpassed US$100 million in March, 10 months since the platform’s launch. In June, the company debuted AI-generated microdramas.

Zooming out to the broader Asia region, a number of startups have recognized the opportunity to compete in the AI video production industry. Tech in Asia’s data shows there are at least 61 companies in Asia that offer tools to create or edit videos using generative AI.

Interestingly, more startups in this space were founded before ChatGPT’s November 2022 release than after – a sign that Asia’s AI video scene didn’t emerge from the recent hype cycle.

Seventeen companies have secured over US$10 million each. Most recently, Beijing-based AIsphere, which develops AI video model and platform PixVerse, has closed a US$60 million series B round led by Alibaba. This brings the startup’s total funding to over US$115 million.

PixVerse, with 100 million registered users globally, has recently ranked 25th on US-based VC firm a16z’s list of most popular genAI mobile apps.

Then there’s Mumbai-based VideoVerse, which has recently been acquired by Sports Illustrated publisher Minute Media. VideoVerse previously raised over US$100 million for its sports-focused video creation platform.

Jaden Xie, co-founder of AIsphere, says Asia holds the edge in AI video generation because of the talent pool it has. He adds that it’s rooted in the success of homegrown video platforms like TikTok and Kuaishou.