SenseTime Technology responded to the news of the ongoing new round of organizational restructuring and layoffs by 'Science and Technology Innovation Board Daily', actively promoting strategic transformation, focusing on key business and strategic growth areas of "large equipment-large model-application", and carrying out corresponding organizational and talent structure optimization and adjustment to better meet business development needs. Currently, SenseTime's social recruitment and campus recruitment are proceeding normally, with overall business development steady.

It is reported that on October 22, SenseTime issued a 10th anniversary letter to all employees. In the letter, SenseTime's chairman and CEO Xu Li mentioned that at the organizational level, around strategic and core resources, a more centralized and efficient organizational structure will be built to promote the concentration and intensive investment of resources, accelerating the youthfulness of organizational and management processes.

Following the issuance of the all-employee letter, SenseTime initiated organizational adjustments and layoffs on the same day, focusing on retaining business related to daily new large models and large equipment, while adjusting other business lines including security, autonomous driving, medical, and more.

According to the financial report, SenseTime Technology has undergone multiple personnel adjustments in the past two years. The number of employees at the end of December 2021, 2022, and 2023 was 6114, 5098, and 4531 respectively, showing a decreasing trend year by year.

However, the number of employees increased in the first half of 2024. As of the end of June 2024, SenseTime Technology had a total of 4672 employees.

Currently, the craze for large models has driven the rise of the computing power industry. SenseTime Technology has been establishing the Lingang Smart Computing Center since 2020, which is also an important component of SenseTime's large equipment.

In the all-employee letter, Xu Li compared traditional AI 1.0 with generative large model AI 2.0, stating that the key difference lies in the change in AI cost structure.

In the traditional AI 1.0 era, the main cost of model production comes from the investment of research and development personnel. In the era of generative large model AI 2.0, the cost of model production mainly lies in the investment of computing resources. As the scale law is verified in large language models, multimodal models, video generation models, and slow thinking inference processes, the cost of producing and using large models can directly equate to the consumption of computing resources.

"In short, to popularize and commercialize generative large model AI, efforts need to be made to reduce the production and usage costs of large models. This inevitably requires the combination of large models to iterate and optimize computing power, while also iterating the design and application of large models based on the characteristics of computing resources."

Therefore, Xu Li believes that in the field of generative large model AI, SenseTime's core global strategy is to seamlessly integrate large equipment for computing operations (SCO), large models and applications (CNI), with the application driving the model, and the model driving the optimization of computing power. He stated that SenseTime has established a triad strategy of "large equipment-large model-application", aiming to enhance the efficiency of computing resource utilization through orders of magnitude optimization, to serve customers well. "We are committed to becoming the most knowledgeable large model service provider in terms of computing power, as well as the most knowledgeable computing power service provider in terms of large models."

Xu Li also mentioned that in the traditional AI field, the core capabilities of visual perception and multimodal models will be fully utilized. Resources will be concentrated, directions will be clarified, and through a set of research and development investments, both domestic and international markets will be served.

According to the semi-annual report, in the first half of 2024, SenseTime achieved a total revenue of 1.739 billion yuan, a year-on-year growth of 21.4%. Among the three major businesses, generative AI revenue was 1.051 billion yuan, a year-on-year growth of 255.7%, increasing its proportion to 60.4% of total revenue, becoming the company's leading business.

Although there has been some revenue growth, SenseTime has still not been profitable in nearly three years since its listing. In the first half of this year, the net loss attributable to shareholders was 2.477 billion yuan, a year-on-year narrowing of 21.2%. For SenseTime, in the competition in the large model field, it not only faces challenges from internet giants, but also from unicorn companies backed by giants.

Renowned economist Pan Helin believes that the main business of SenseTime is still AI, but AI must be integrated with application scenarios to generate profits. At this stage, SenseTime is struggling to find a breakthrough, so there are considerations to reduce labor costs through architectural adjustments. Currently, the AI industry is indeed facing a profit dilemma. With the beginning of the era of diverse models, the competition among AI companies is intensifying, and many AI companies have not found a profitable breakthrough, thus being forced to make long-term plans through personnel adjustments.