By Venture capital is flooding into China’s emerging video generation startups, with investors injecting hundreds of millions of yuan, banking on their potential to compete in the artificial intelligence landscape, particularly against OpenAI’s groundbreaking text-to-video model, Sora.
This week witnessed significant milestones for two leading Chinese generative AI startups, as they announced the successful closure of substantial funding rounds, each exceeding 100 million yuan ($14 million), signaling their intentions to challenge Sora in the near future.
One of these startups, Beijing-based Shengshu Technology, disclosed securing several hundred million yuan in its recent fundraising endeavor. The investment, spearheaded by Qiming Venture Partners, in collaboration with Delta Capital, Zhipu AI, and Baidu Ventures, among others, will bolster Shengshu’s efforts in multimodal large model research, product innovation, and market expansion.
Established in March 2023 by scientists from Tsinghua University Institute for Artificial Intelligence, Shengshu focuses on developing multimodal large models encompassing images, 3D, and video. Their product range includes image-to-text generation, joint image-text generation, image-text rewriting, and transforming static images into dynamic, multi-angle viewable content.
On Monday, AIsphere, another prominent generative AI startup, announced raising over 100 million yuan in a funding round led by venture firm Fortune Capital. Founded in April 2023 by Wang Changhu, former head of ByteDance AI Lab, AIsphere introduced PixVerse, an AI video creation platform supporting text-to-video generation, which garnered over 1 million visits in February, according to Similarweb.
Both Shengshu and AIsphere leverage diffusion transformer architecture in their video generation models, akin to Sora’s approach. AIsphere’s founder, Wang, expressed determination to surpass Sora within three to six months, citing Sora’s validation of diffusion transformers’ effectiveness in scaling as video models.
Shengshu’s CEO, Tang Jiayu, aims to challenge Sora’s current iteration by year-end, asserting that China’s domestic models, compared to GPT-4, are rapidly narrowing the gap with Sora. The emergence of Sora has shifted investor focus towards multimodal investments, intensifying support for ventures like Shengshu and AIsphere in their pursuit of AI innovation.
