SenseTime’s AI growth surges, as China advances nationwide initiative


SenseTime has significantly narrowed its losses while achieving robust growth in its generative artificial intelligence business in the first half of 2025, as China pushes a nationwide initiative to advance the AI industry.

The Hong Kong-listed company on Thursday posted revenue of 2.36 billion yuan (US$330 million) for the six months ended June 30, marking a 35.6 per cent increase from a year earlier. Gross profit rose 18.4 per cent year on year to 907.8 million yuan, although gross margin declined to 38.5 per cent from 44.1 per cent owing to rising hardware and data centre infrastructure costs.

The company’s adjusted net loss was halved to 1.16 billion yuan, down 50 per cent from 2.33 billion yuan a year earlier.

Revenue from the generative AI division surged 72.7 per cent to 1.82 billion yuan, increasing its share of total sales to 77 per cent, up from 60.4 per cent a year earlier. SenseTime attributed this to a “tremendous growth” in demand for training, fine-tuning and inference of generative AI models.

SenseTime’s shares edged up 0.48 per cent to HK$2.09 ahead of the announcement. The stock has risen 48.2 per cent so far this year.

Xu Li, CEO of SenseTime. Photo: Handout
Xu Li, CEO of SenseTime. Photo: Handout

“The industry’s main theme in the first half of this year was multimodal [and] reasoning, combined with AI agents facilitating large-scale adoption,” said CEO Xu Li on the earnings call.

  • Related Posts

    How Hong Kong could ease its ‘positive problem’ of full metals warehouses

    Just over a year after the London Metal Exchange (LME) approved its first batch of warehouses in Hong Kong, the city now hosts 15 facilities run by seven operators, with…

    Continue reading
    Chinese 6-seat electric SUVs to stand out amid weak domestic sales, says Morgan Stanley

    Six-seat premium electric sport utility vehicles (SUVs) are emerging as an unlikely game changer in China’s automotive market, with more than a dozen new locally developed models set to challenge…

    Continue reading

    Leave a Reply

    Your email address will not be published. Required fields are marked *