SenseTime Posts First Half-Year Profit as China's AI Giants Burn Cash

For years, the question hanging over China's artificial intelligence sector has been simple: can anyone actually make money? SenseTime, the country's most prominent pure-play AI company, has just offered an answer that its cash-burning rivals cannot yet match.

The Hong Kong-listed firm reported a net profit of 617.3 million yuan (US$92 million) for the first half of 2026 — its first-ever profit for a half-year period under International Financial Reporting Standards since its 2021 listing. Revenue climbed 23.4 per cent year on year to 2.91 billion yuan. The numbers landed last week and were detailed by CEO Xu Li and chief financial officer Wang Zheng in conversations with the South China Morning Post. What makes the result striking is the contrast with peers. MiniMax and Zhipu AI, also known as Z.ai, both posted triple-digit revenue growth for the same period — and net losses of US$358 million and 2.07 billion yuan (US$308 million) respectively. Growth, in other words, has not been the hard part. Profitability has.

SenseTime's route to the black runs through generative AI, which brought in 2.33 billion yuan, or nearly 80 per cent of total group sales. The company also disclosed recurring revenue for the first time: 1.14 billion yuan, up 124.4 per cent year on year and representing almost 40 per cent of sales. That disclosure matters. Recurring revenue is the metric that separates a software business from a project shop, and it suggests SenseTime is building something stickier than one-off model licensing deals. Executives describe the company's core capabilities — models, a token factory, and an agent-harness system — as each forming what Xu called an independent commercial closed loop. Wang's framing was blunter: among pure-play AI companies, SenseTime is one of the very few whose overall business trend is clearly moving toward profitability.

The strategy behind the numbers is a deliberate refusal to chase model size for its own sake. Instead of competing on parameter counts, SenseTime has focused on helping clients complete enterprise tasks, and on serving a segment that China's tech giants have largely overlooked: one-person firms and solo entrepreneurs. This is a distinctly Chinese story. The country has a vast population of freelancers, micro-businesses and independent operators who need productivity tools but cannot afford enterprise contracts. SenseTime is betting that monetising them — alongside larger enterprise clients — creates a more durable base than the subsidy-fuelled land grabs that have defined much of China's AI race.

For international investors watching Asia's technology capital, the signal is twofold. First, the era of measuring Chinese AI companies purely by revenue growth or model benchmarks may be giving way to a harder test: unit economics. Second, SenseTime's pivot suggests the winners in this sector may not be the firms with the largest models, but those that find paying customers fastest. The company's Hong Kong listing gives global funds a rare liquid proxy for Chinese AI, and a profitable half-year changes the conversation in boardrooms from survival to scaling.

None of this means the race is over. MiniMax and Zhipu AI are still growing faster on the top line, and their losses reflect heavy investment in frontier research that could yet pay off. SenseTime itself must prove that one profitable half-year is a trend rather than a quarter's timing. But for a sector that has consumed billions in capital with little to show in earnings, the company has done something its peers have not: it has shown that a Chinese AI firm can make money. The next question is whether anyone else can follow.