Hong Kong-listed artificial intelligence company SenseTime recorded its first profit since going public, helped by higher generative AI revenue and gains on investments, while its core operations remained loss-making on an adjusted basis.
SenseTime reported revenue of RMB2.91 billion ($407 million) for the six months ended June 30, 2026, up 23.4 percent from a year earlier. Gross profit rose 32.9 percent to RMB1.21 billion, lifting its gross margin by 2.9 percentage points to 41.4 percent, according to the company’s interim results filed with the Hong Kong Stock Exchange.
Investment gains support the bottom line
Profit for the period reached RMB617.3 million, reversing a RMB1.49 billion loss in the first half of 2025. SenseTime said this was its first profit under International Financial Reporting Standards since its listing.
The positive result included a significant increase in the fair value of companies that SenseTime incubated or invested in. Its non-IFRS adjusted net loss was RMB385.9 million, although that figure narrowed 67.3 percent year on year. Adjusted earnings before interest, taxes, depreciation and amortization turned positive at RMB384.9 million, compared with a RMB540.3 million loss a year earlier.
The difference between reported profit and the adjusted loss remains an important qualification. It indicates that valuation gains, alongside improvements in the underlying business, played a substantial role in the headline result. The filing did not treat the adjusted measure as a substitute for the statutory accounts, and investors will still need to watch whether operating improvements can be sustained without similar valuation movements.
Generative AI drives most revenue
Generative AI revenue increased 28.2 percent to RMB2.33 billion and accounted for 79.9 percent of the group total. Computer vision revenue rose 13.9 percent to RMB496.8 million, while overseas revenue increased 127 percent from the previous year.
SenseTime introduced a recurring-revenue measure in the results. It reported RMB1.14 billion from existing contracts with recurring characteristics, more than double the RMB510.1 million recorded a year earlier. Recurring revenue represented 39.3 percent of the total, up from 21.6 percent.
The company said it served more than 1,000 enterprise customers during the period across sectors including education, government, marketing, logistics, banking and insurance. More than 100 were added during the first half. Its computer vision operations covered more than 20 countries and regions and had served more than 4,500 customers cumulatively, with returning customers contributing 67 percent of revenue in that segment.
Compute and agent services expand
SenseTime is organizing its AI business around multimodal models, a compute platform it calls the Token Factory, and an orchestration layer for enterprise and consumer agents. The company said the total compute capacity under its operation had reached 48,000 petaflops.
Its SenseCore platform handled an average of more than 2.4 trillion tokens per day in July, about 22 times the volume recorded a year earlier. Beyond supporting SenseTime’s own SenseNova models, the platform provided services to four external foundation-model providers, according to the filing.
The company said joint optimization of models and infrastructure had cut the path from research validation to large-scale deployment to one week. It also reported an 8 percent reduction in average unit electricity cost through coordinated compute and energy management, with savings exceeding RMB12 million during the period. These efficiency figures were not independently benchmarked in the results.
SenseTime also said more than 45 million people had used its Kapi consumer-agent products. Its Raccoon office agent has been deployed with companies including Lenovo, Ping An Technology, JD.com and China’s three large telecommunications operators.
Many operating and product-use figures in the announcement are company-reported and have not been independently audited. The interim financial statements were prepared under IAS 34 and reviewed by the company’s auditor and audit committee.
Image credit: https://www.sensetime.com/en, CC BY-SA 4.0, Link