AIsphere Secures Major Series A1 Funding, Aiming to Surpass Sora Within Six Months

“AIsphere”, a Chinese AI start-up in large video model development, has recently secured a Series A1 funding round in the hundreds of millions, with Dawn Capital as the exclusive investor and Guangyuan Capital as the financial advisor. The funding will be utilized for technical R&D, team expansion, and other areas.

The company gained significant attention following the release of OpenAI’s Sora in February 2024, which sparked a new wave in the large video model industry. AIsphere, founded by former ByteDance visual technology head Wang Changhu in April 2023, aims to build top-tier AI video generation models and applications for industries such as marketing, advertising, and gaming.

Despite the industry buzz around Sora, founder Wang Changhu believes there is still significant market space to explore. He asserts that Sora’s technological development is between GPT-2 and GPT-3 levels, leaving room for growth. He identified three dimensions to evaluate video model generation quality: accuracy, consistency, and information richness.

AIsphere’s AI video generation product, PixVerse, is now operational in overseas markets. The product supports text-to-video and image-to-video features, achieving a 4K resolution for generated video content by October 2023. PixVerse has attracted users from various industries, with over a million monthly visits as per

The company’s strategy is to focus on the Consumer-end market, aiming to enhance the entire AI video generation value chain. This involves improving the efficiency of content generation, review, distribution, and commercialization, and ultimately achieving a closed-loop AI Native video product application.

Currently, AIsphere maintains a team of over 20 individuals, a modest increase from the previous year. The team’s expertise, drawn from experiences with top platforms like Douyin and TikTok, is seen as a core competitive advantage. The team also demonstrates strong execution capabilities, with weekly model iterations and simultaneous optimization of multiple versions. They have been early adopters of the Transformer technology to replace the Unet in the Diffusion Model, leading to gradual improvements in model performance.

The team embodies a typical engineering culture, characterized by flat management and a collective pursuit and belief in technology. Looking ahead, AIsphere aims to create the next generation of end-to-end Super AI-Native video platforms, products, and applications, aspiring to become the leading multimodal AI company in the domestic market.

Looking ahead, AIsphere aims to match or surpass industry benchmarks like OpenAI’s Sora over the next 3 to 6 months, while continuing to innovate and expand its product offerings for both domestic and international markets.

SEE ALSO: Baidu’s Robin Li: Ernie Bot 4.0 Has Surpassed GPT-4 in Chinese Language Processing