SenseTime H1 report shows AI-related growth

Although the overall revenue for the first half of the year remained relatively stable compared to the same period last year, the surge in sales costs led to a considerable drop in gross profit margin.

Photo from CFP

Photo from CFP

By YU Hao

 

SenseTime's H1 report showed total revenue of 1.43 billion yuan (US$200 million), a 1.3 percent year-on-year increase. 

The company's gross profit in H1 was 650 million yuan, down by 30.6 percent. The net losses of 3.1 billion yuan, or 2.4 billion yuan after adjustments, were both less than at this time last year. SenseTime's cash flow remains negative, with 1.8 billion yuan flooding out of the company, slightly less than last year.

Much of SenseTime's hopes rely on Large Language Models (LLMs) but lag behind most competitors. LLMs have attracted phone makers as customers for SenseTime's AI sensors and AI ISP chips. However, despite the increase in revenue, profitability has been dragged down by the smart city and smart car businesses.

Although the overall revenue for the first half of the year remained relatively stable compared to the same period last year, the surge in sales costs led to a considerable drop in gross profit margin. Notably, hardware costs and subcontracting services saw a year-on-year increase of 54.1 percent, while the depreciation of the intelligent computing center grew by 75 percent.

Investment in the intelligent computing center has yielded positive outcomes, with SenseCore AI's GPU count now elevated to 30,000 units, resulting in a decrease in research and development costs. 

In terms of foundation model iteration, the new "InternLM-123B" model has increased its parameter count to 123 billion, and it is projected that "SenseChat" will be upgraded to version 3.0 by September.