SenseTime's Generative AI Revenue Nears 80% of Total, Achieves First-Ever Half-Year Profit
nashnova research
SenseTime posted RMB 617 million in net profit for H1 2026 — its first half-year profit since listing in Hong Kong — with generative AI driving nearly 80% of revenue and recurring income doubling, marking a real inflection point for pure-play AI commercialization.
What does this "first-ever profit" actually mean?
SenseTime (商湯科技, 00020.HK) reported H1 net profit of RMB 617.3 million — the first half-year profit under IFRS since its 2021 Hong Kong IPO.
Revenue rose 23.4% year-on-year to RMB 2.91 billion.
This means → SenseTime has crossed from "burning cash for scale" into "making money from core operations" — almost unprecedented among China's pure-play AI companies.
Where did the money come from? Why can generative AI carry 80% of revenue?
The generative AI business — services that use large models to produce text, images, and code for clients — brought in RMB 2.33 billion, nearly 80% of group revenue.
The company disclosed recurring revenue for the first time: RMB 1.14 billion, up 124.4% year-on-year, roughly 40% of total revenue.
This means → clients are not just buying once — they are renewing. The revenue mix is shifting from one-off projects to ongoing subscriptions, making earnings far more predictable.
What does management mean by "three independent loops"?
CEO Xu Li said SenseTime's core capabilities split into three pillars: models, compute factories, and an AI-agent orchestration system — each now a "self-sustaining commercial loop."
In plain terms = each pillar can generate revenue on its own; the business is not propped up by a single leg.
CFO Wang Zheng added: "Among pure-play AI companies, we are one of the very few whose overall trajectory is clearly heading toward profitability."
Peers are still losing money — why did SenseTime break out first?
For comparison: MiniMax posted a H1 net loss of USD 358 million; Zhipu AI lost RMB 2.07 billion. Both achieved triple-digit revenue growth but remain deep in cash-burn mode.
SenseTime management attributes the gap to a strategic choice: stop chasing model scale for its own sake, and instead focus on helping enterprise clients finish concrete production tasks — plus serving individual entrepreneurs.
This reflects a deeper signal: in China's AI industry, "can you make money" is replacing "how big is your model" as the market's core yardstick for valuing a company.
What to watch next?
The key question is singular: can this profit model be replicated into H2 and beyond?
Recurring revenue at 40% of the total is a solid start, but still some distance from "highly predictable." If that ratio keeps climbing, the valuation framework shifts from "AI concept stock" to "SaaS-style subscription company."
This means → in the next earnings report, recurring-revenue growth and client retention rates will matter more than the headline revenue number.
市场有风险，内容仅供研究参考，不构成投资建议。