US artificial intelligence (AI) video generator start-up Pika Labs has become the latest sensation in China’s tech community, after its co-founder Demi Guo was identified as a native of Hangzhou, capital of eastern Zhejiang province, and daughter of former Sunyard Technology chairman Guo Huaqiang.
While Silicon Valley-based Pika Labs, co-founded in April by Guo and her Stanford University classmate Meng Chenlin, is situated on the other side of the Pacific, its sudden fame has set off a rush by Chinese investors to buy the stock of information technology services firm Sunyard, where the elder Guo remains the largest stakeholder.
The surprise surge in Sunyard’s shares, which have risen more than 21 per cent this week, prompted the company to file a statement to the Shanghai Stock Exchange to clarify that Sunyard has no business ties with, nor investments, in Pika Labs.
The Pika 1.0 video-generating AI model, launched on Tuesday, is designed to generate and edit videos in a wide range of formats – from cartoons to realistic cinematic content – based on simple text prompts such as “Elon Musk in a spacesuit, 3D animation”.
“Our vision for Pika is to enable everyone to be the director of their own stories and to bring out the creator in each of us,” the company said in a post announcing the release of the model.
The start-up said it has so far raised US$55 million in seed and Series A funding rounds from some of the biggest names in Silicon Valley, including Nat Friedman, former CEO of open-source community GitHub, and Daniel Gross, former partner of incubator Y Combinator.
In a recent interview with Chinese media Overseas Unicorn, Guo attributed Pika Labs’ popularity in an increasingly crowded AI market to the start-up’s technological strength.
The company currently has only four full-time employees, including the co-founding duo, who had been PhD candidates at Stanford studying computer graphics, natural language processing and generative AI, before they dropped out earlier this year to start Pika.
Guo said in the interview that one of the main reasons she quit school was to focus her attention on the quickly-developing technology underpinning video-generation models, which she expected to advance as fast as image-generation models like Midjourney and Stable Diffusion.
“If you compare AI-generated videos from last year with those generated this March and those from the past one or two months, you’ll see that video-generation models are developing really fast,” Guo was quoted as saying.
The company is looking to launch a commercialised version of its video generator next year that can create short clips good enough to be used in films, Guo said, adding that her team is striving to become the most technologically advanced player in the AI video-generation sector.