What’s new: Beijing-based generative AI startup Shengshu Technology (生数科技) announced that it has completed a new funding round in less than three months, securing several hundred million yuan in fresh investment, as investors double down on their bets in the AI race.
The Pre-A round of fundraising was led by Baidu Inc. and the Beijing Artificial Intelligence Industry Investment Fund, with participation of ZGC Science City Ltd., Qiming Venture Partners and others, Shengshu said Wednesday, without giving the actual size of the fundraising.
Shengshu said the funds raised will be used to support the development of general-purpose multimodal technology, iterating and optimizing its self-developed large models, and accelerating product development and market expansion.
Shengshu unveiled its own text-to-video AI tool Vidu in April, which is taking on OpenAI’s Sora. Described as China’s “first long-duration, high-consistency and highly dynamic video generation model, Vidu can generate high-definition videos lasting up to 16 seconds.
Background: In March, Shengshu received several hundred million yuan from a group of investors including Qiming Venture Partners, along with Delta Capital, Zhipu AI and Baidu Ventures among others.
Shengshu is one of a group of Chinese startups capturing venture capital investors’ attention for their potential in the AI race to rival American tech giant OpenAI’s text-to-video model Sora.
OpenAI’s innovative model unveiled in February can generate videos up to one minute in length from user prompts, captivating audiences and sparking global interest. Previously, investors focused heavily on financing large language models. However, following Sora’s release, they have intensified support for multimodal investments.
Established in March 2023 by scientists from Tsinghua University’s Institute for Artificial Intelligence, Shengshu focuses on the research and development of multimodal large models, including images, 3D and video. Its products include image-to-text generation, joint image-text generation, image-text rewriting and transforming flat images into three-dimensional, multi-angle viewable content.
Contact reporter Han Wei (weihan@caixin.com)
