The competition in China’s artificial intelligence sector is more intense than ever. Companies like DeepSeek, Moonshot AI, Alibaba, and even Xiaomi have recently released new AI models, all vying for top positions in the market.
From small AI startups to massive tech giants, every company faces immense pressure to innovate, grow their user base, and find ways to make money. They also grapple with the high costs of research and development, as well as the increasing expense of computing power and specialized hardware.
SenseTime, one of China’s older AI firms, has shifted its focus to stay relevant in the age of generative AI. Previously known for facial and image recognition, the company now builds “multimodal” systems that can process text, audio, and visual information together.
Founded in Hong Kong in 2014, SenseTime has faced challenges, including U.S. sanctions due to allegations of surveillance of Muslim minorities in Xinjiang, which the company denies. Their newest model, SenseNova U1, combines language and vision processing into one system. This makes it faster and more efficient because it doesn’t need to translate between different types of data.
SenseTime believes that being cost-effective will give them an advantage. According to co-founder and chief scientist Lin Dahua, the company learned from DeepSeek’s strategy of creating powerful models despite financial and technical limitations. Lin pointed out that while OpenAI’s ChatGPT Images 2.0 generates “exquisite and beautiful” results, SenseNova U1 costs ten times less.
“You often don’t need the very best model when a slightly less powerful one can handle most tasks,” Lin explained. “We still have some catching up to do with international leaders like OpenAI’s GPT Image 2 and Gemini’s Nano Banana, but our costs are much lower – it’s very efficient.” With limited overlap between the U.S. and Chinese AI markets, the main competition for these companies is often right at home.
Lin initially saw ByteDance’s AI video model, Seedance, as a major competitor. However, SenseTime has since integrated some of Seedance’s features into its own short-video tool, Seko, allowing it to combine Seedance’s background generation with its own audio functions.
