SenseTime Says Cost-Efficient AI Models Can Outlast Bigger Rivals
CNBC reported Tuesday that Hong Kong-listed artificial intelligence company SenseTime believes cost-efficient AI models can capture meaningful market share. Co-founder Lin Dahua made the case directly to CNBC, arguing that buyers rarely need the most powerful tool available.
SenseTime’s Cost-Efficiency Bet
SenseTime’s newest release, SenseNova U1, merges language and vision processing into one unified system. Removing the translation step between data types speeds up performance and cuts running costs. Lin told CNBC the model operates at roughly one-tenth the price of OpenAI’s image-generation product, even while acknowledging a quality gap with frontier Western systems. His argument is pragmatic: most enterprise tasks do not require top-tier output, so price becomes the deciding factor.
The company has drawn strategic inspiration from DeepSeek, which built competitive models under tight financial and hardware constraints. SenseTime is applying a similar philosophy to its own product lineup as it targets corporate clients rather than consumers. Enterprise buyers tend to prioritise reliability and price over novelty, and they churn at lower rates than individual users.
A Crowded Field With No Clear Finish Line
China’s AI landscape has grown sharply more competitive. In recent weeks alone, Alibaba, Xiaomi, Moonshot AI and ByteDance have all pushed out new or updated models, compressing the window any single player has to gain ground. ByteDance’s video model Seedance initially posed a challenge for SenseTime’s own short-video offering. SenseTime responded by integrating Seedance’s background-generation capabilities with its own audio tools inside its Seko platform.
The Platform Advantage Problem
Investment bank Jefferies flagged a structural issue facing standalone AI firms in an April note. Pure-play model companies carry high training costs, thin customer loyalty and limited ways to differentiate. Platform giants such as Alibaba, Tencent and ByteDance carry a built-in advantage: their core businesses generate cash that subsidises AI development while also supplying proprietary user data. Senior equity advisor Vey-Sern Ling at UBP told CNBC those platform players are in a stronger position than loss-making independents. Even so, heavy AI investment has weighed on profits across the sector, including at larger platform operators.
SenseTime’s Background and Sanctions Status
Founded in Hong Kong in 2014, SenseTime built its early reputation on facial and image recognition technology. The United States placed the company on its sanctions list over allegations tied to surveillance of ethnic minorities in Xinjiang, allegations SenseTime has denied. Despite those restrictions, the company is pressing ahead with international expansion, including plans in the Middle East that remain unchanged. SenseTime narrowed its annual net loss by more than 58% last year, a sign the cost-cutting strategy is producing early results.
Read Next: What DeepSeek’s Efficiency Push Means for the Global AI Market