
China’s artificial intelligence race has no finish line. deep seek, moonshot AI, alibaba And even home appliance companies xiaomi In recent weeks, companies have announced new models and are vying for a spot on the leaderboard.
From native AI startups to platform giants, companies in all sectors face increasing pressure to innovate, grow their user bases, and find ways to generate revenue. At the same time, huge R&D costs must be managed as computing power and hardware spending increases.
SenseTime, one of China’s earliest AI companies, pivoted to stay relevant in the generative AI era. Long known for facial and image recognition, the company is now developing multimodal systems that can combine text, audio, and visual data.
SenseTime, founded in Hong Kong in 2014, faces U.S. sanctions over allegations related to surveillance of Muslim minorities in Xinjiang, which the company denies.
The latest model, SenseNova U1, integrates language and visual processing into one system, increasing speed and efficiency by eliminating the need to convert between different modes.
SenseTime is betting on cost efficiency as a competitive advantage. Lin Dahua, co-founder and chief scientist, said the company was inspired by DeepSeek’s approach to delivering high-performance models under financial and technical constraints.
Lin Dahua, co-founder and chief scientist, at SenseTime’s offices in Hong Kong’s Science Park.
CNBC
ChatGPT Images 2.0, OpenAI’s artificial intelligence tool that generates images from text prompts, produces “exquisite and beautiful” results, but SenseNova U1 costs one-tenth the cost, Lin said.
“A top-of-the-line model that can handle most tasks may not be needed in many cases,” Lin told CNBC. “There is still a gap between it and international frontier models like OpenAI’s GPT Image 2 and[Gemini’s]Nano Banana, but it costs much less and is very efficient.”
There is limited overlap between the US and China AI markets, so the real competition could be closer to home.
ByteDance’s AI video model, Seedance, initially raised competitive concerns, Lin said. SenseTime has since integrated some of that functionality into its short video tool Seko, allowing users to combine Seedance’s background generation with their own audio capabilities.
More than model racing
Technology is only half the battle; business models are becoming increasingly important. OpenAI’s reported failure to meet revenue and user goals suggests risks for Chinese and U.S. players alike, Jefferies said in an April 28 note, according to the Wall Street Journal.
Pure AI model companies face a tough equation: low customer loyalty, limited differentiation, a crowded field and high training costs, Jeffries said.
In contrast, large internet platforms have stronger cash flows, access to user data and established customer bases to sell AI applications, the bank added.
In China, the following platform companies alibaba, tencent Baisan Lin, senior equity advisor at UBP, said both companies could leverage their core businesses to subsidize AI development and strengthen existing businesses.
Sensetime booth at the World Artificial Intelligence Conference 2021 held in Shanghai, China on July 7, 2021.
Photo | Future Publishing | Getty Images
“They are clearly in a better position than standalone companies that have been losing money,” Lin added, noting that even big companies like Alibaba and Kuaishou are also seeing heavy spending on AI weighing on their profits.
To stand apart, sense time Lin said they combined large-scale AI models, applications and infrastructure to improve quality of service while reducing cost per use. Many of the company’s products are targeted at business customers, who often demand higher quality service and are less likely to switch providers even if they pay more.
sense time The company narrowed its net loss by 58.6% last year and reported positive EBITDA in the second half for the first time since going public in 2021, and investors will continue to monitor developments going forward. Lin said the company’s AI costs are “manageable” and is primarily focused on improving model efficiency.
shares of sense time It rose 2% in afternoon trading Wednesday.
Price to Win, or Win to Price
Pricing strategies vary by sector.
Some companies, including DeepSeek, have recently lowered prices or offered discounts to attract users. In addition, the following Chipu This shows that we are promoting the commercialization of advanced models.
Alibaba’s cloud unit Baidu As the demand for AI computing power skyrockets, prices have also increased. ByteDance is planning a subscription service for certain features of its popular AI chatbot Doubao.
“Price competition may serve a strategic function in short-term promotions, but long-term sustainability depends on differentiated value,” Lin said.
Analysts say some AI companies may be following a familiar strategy of hemorrhaging cash to gain market share and then monetize by raising prices, given China’s huge market.
“The use of AI is so expensive that we can’t keep subsidizing it,” said UBP’s Lin.
“Either you can paint a picture of the huge future usage and demand and convince investors that short-term losses are acceptable, or you need to start monetizing much sooner.”
Bet on the world beyond Washington
Facing U.S. export and investment restrictions, SenseTime has focused its international expansion on markets such as Southeast Asia, North Asia, the Middle East and, most recently, Brazil.
Although the US and Israel’s war against Iran has caused short-term disruption, affecting flights and traffic, Lin said the company’s long-term strategy in the region remains unchanged.
Cost efficiency and practicality are equally important in overseas markets.
“Often the reason behind repeat purchases is not because the technology is particularly advanced, but because we offer the best service at a competitive price,” Lin said.
