BEIJING — In the intensifying race for artificial intelligence supremacy in China, SenseTime, a pioneering but sanctioned tech firm, is positioning cost-efficient models as a viable path to victory, even as it navigates U.S. restrictions and fierce domestic competition.
The company's latest innovation, SenseNova U1, represents a shift toward multimodal AI systems that process text, audio, and visual data seamlessly, aiming to deliver high performance at a fraction of the cost of Western counterparts. Founded in Hong Kong in 2014, SenseTime has long been recognized for its advancements in facial and image recognition technology. However, U.S. sanctions imposed since 2019 have accused the company of aiding surveillance of Muslim minorities in Xinjiang, allegations that SenseTime has firmly denied. Despite these challenges, the firm reported narrowing its net loss by 58.6% last year and achieving positive EBITDA in the second half of the year for the first time since its 2021 listing on the Hong Kong stock exchange.
SenseTime's co-founder and chief scientist, Lin Dahua, emphasized the strategic focus on efficiency during an interview with CNBC. "While ChatGPT Images 2.0, an artificial intelligence tool from OpenAI that generates images from text prompts, produces 'exquisite and beautiful' results, SenseNova U1 costs ten times less," Lin said. He added, "You may not need the top model in many cases when it can handle most tasks. There is still a gap between us and the international frontier models like OpenAI's GPT Image 2 and (Gemini's) Nano Banana, but our cost is much lower – it's very efficient."
This approach draws inspiration from other Chinese AI players like DeepSeek, which has gained attention for producing competitive models despite financial and technological constraints. China's AI landscape has seen a flurry of activity in recent weeks, with startups such as DeepSeek and Moonshot AI, alongside giants like Alibaba and even consumer electronics maker Xiaomi, releasing new models and vying for spots on global leaderboards. The pressure is mounting as companies grapple with soaring research and development expenses, coupled with the high costs of computing power and specialized hardware.
Domestically, the competition is particularly cutthroat, with limited overlap between U.S. and Chinese AI markets due to geopolitical tensions. Lin highlighted ByteDance's AI video model, Seedance, as an initial competitive threat. SenseTime responded by integrating elements of Seedance's background generation capabilities into its own short-video tool, Seko, which now combines those features with proprietary audio functions. This adaptation underscores the rapid pace of innovation and collaboration—or imitation—in China's AI sector.
Beyond technology, business models are emerging as a critical battleground. According to a report in The Wall Street Journal, OpenAI missed its revenue and user growth targets last year, a cautionary tale for AI firms on both sides of the Pacific. Jefferies analysts, in an April 28 note, described the challenges for pure-play AI companies: low customer loyalty, limited product differentiation, a saturated market, and exorbitant training costs. "Pure-play AI model companies face a tough equation," the note stated.
In contrast, large internet platforms hold significant advantages. Vey-Sern Ling, a senior equity advisor at Union Bancaire Privée (UBP), explained that companies like Alibaba, Tencent, and ByteDance can leverage their robust cash flows, vast user data troves, and established customer bases to subsidize AI development and integrate it into core operations. "They are obviously in a better position than the standalone ones, which continue to be loss-making," Ling said in comments to CNBC. He noted, however, that even these giants are feeling the strain, with heavy AI investments contributing to profit pressures at firms like Alibaba and short-video platform Kuaishou.
SenseTime is attempting to carve out its niche by bundling large AI models with applications and infrastructure services, which Lin says improves service quality while reducing per-use costs. The company primarily targets enterprise clients, who prioritize reliability and are more willing to pay premiums while showing lower churn rates. "Many of its products target enterprise clients, who often demand higher-quality services, are willing to pay more and less likely to switch providers," Lin explained.
Financially, SenseTime's trajectory offers some optimism for investors. The company's AI-related costs remain "manageable," according to Lin, with a strong emphasis on enhancing model efficiency to control expenses. This comes amid a broader pricing dynamic in China's AI market, where some players are slashing costs to lure users. DeepSeek, for instance, has recently introduced discounts, while Zhipu AI has raised prices to push commercialization of advanced features. Meanwhile, the cloud divisions of Alibaba and Baidu have increased rates in response to booming demand for AI computing resources, and ByteDance is developing a subscription model for premium functionalities in its popular chatbot, Doubao.
Lin cautioned against relying solely on price wars. "Price wars might serve a strategic function in short-term promotions, but sustainability in the long run depends on differentiated value," he told CNBC. Analysts echo this sentiment, drawing parallels to other Chinese tech sectors where initial subsidies build market share before eventual price hikes enable profitability. "They cannot keep subsidizing the usage of AI because it's very expensive," Ling from UBP remarked. He suggested that AI firms must either "paint a picture of huge future usage and demand and help investors understand that near-term losses are acceptable" or accelerate monetization efforts.
U.S. export controls and investment bans have forced SenseTime to redirect its international ambitions away from Western markets. Instead, the company is expanding into Southeast Asia, North Asia, the Middle East, and more recently, Brazil. These regions offer opportunities where cost-efficiency and practical applications resonate strongly. However, geopolitical events have posed hurdles; Lin noted short-term disruptions from the ongoing U.S.-Israeli conflict with Iran, including flight cancellations and reduced interactions, though the firm's long-term strategy in the Middle East remains intact.
"Oftentimes, the reason behind repeat buying is not about the technology being particularly advanced, but providing the best service at a competitive price," Lin said, highlighting a universal appeal for SenseTime's model abroad. This focus on value over cutting-edge novelty could prove advantageous in emerging markets where budgets are tighter and infrastructure less advanced.
The broader implications of China's AI push extend beyond corporate rivalries. As the sector matures, questions about sustainability loom large. With platform giants subsidizing losses and startups racing to differentiate, the path to profitability remains uncertain. SenseTime's progress— from sanction-hit pariah to efficiency-focused contender—illustrates the resilience of Chinese innovation under constraints. Investors will be watching closely as the company reports its next earnings, particularly how it balances expansion with fiscal discipline.
Looking ahead, the AI race in China shows no signs of slowing. As more models flood the market and pricing strategies evolve, the emphasis on cost control and enterprise solutions may define the winners. For SenseTime, proving that cheaper can indeed compete with the best will be key to its survival and growth in this high-stakes arena.