SenseTime launches a sprint towards generative AI

Author | Liu Baodan

Editor | Huang Yu

All In AI has become a consensus among technology companies. As a representative of the previous wave of AI in China, SenseTime is also fully shifting towards the generative AI track.

On December 3rd, Xu Li, Chairman and CEO of SenseTime, released a letter to all employees announcing the establishment of a new "1+X" structure, where "1" represents the core business. SenseTime aims to create an industry-leading AI cloud, achieving seamless integration of large devices, foundational models, and AI applications. In the field of computer vision (CV), it will develop a general visual model to delve into various application scenarios.

"X" represents the ecosystem of restructured and split enterprises, including smart cars "Jueying," home robots "Yuanluobo," smart healthcare, smart retail, and more. At the same time, each ecological enterprise will establish an independent CEO responsible for business development.

This adjustment is based on a reorientation of the company's strategic direction and core areas. With the completion of the strategic restructuring, SenseTime can better respond to the opportunities and challenges brought by the AI 2.0 era and fully embrace generative AI.

During the previous wave of AI, SenseTime became one of the "Four Little Dragons of AI" due to its technological advantages in facial and visual recognition, successfully listing on the Hong Kong Stock Exchange in 2021. By the end of 2022, with the generative AI wave triggered by ChatGPT, SenseTime also began to increase its investment in AI large model technology and applications.

SenseTime's daily new large model began in late 2022, with over 10,000 GPUs' computing power resources invested. In April 2023, SenseTime launched the 1.0 version of daily new, which is one of the first large language models released by SenseTime. In July 2024, daily new was further upgraded to version 5.5, achieving comprehensive performance and real-time interaction experience comparable to GPT-4o, becoming China's first multimodal real-time interactive large model.

Until this strategic restructuring, SenseTime has built an organizational structure centered on generative AI.

In October, on the occasion of SenseTime's tenth anniversary, Xu Li stated that the company believes in and looks forward to the arrival of the era of general artificial intelligence. "Pragmatically speaking, we have taken two steps: the traditional AI 1.0 and generative large model AI 2.0."

In Xu Li's view, an important distinction between AI 1.0 and AI 2.0 lies in the change of the cost structure of AI. In the traditional AI 1.0 era, the main cost of model production was the investment in R&D personnel. In the generative large model AI 2.0 era, the cost of model production mainly lies in the investment in computing power resources.

In short, the popularization and commercialization of generative large model AI require efforts to reduce the production and usage costs of large models, which inevitably necessitates the iteration and optimization of computing power in conjunction with large models, while also needing to iterate the design and application of large models based on the characteristics of computing power resources.

Therefore, in the field of generative large model AI, SenseTime's core strategy is to achieve seamless integration of large computing devices (SCO), large models, and applications (CNI), driving models with applications and optimizing computing power with models. To this end, SenseTime has made organizational adjustments around its strategy and core resources Xu Li emphasized the strategic significance of the new structure: first, to firmly establish generative AI as the core business, accelerating profitability and stable cash flow; second, to leverage the synergy of the group's infrastructure advantages, which is more conducive for ecosystem enterprises to quickly seize market opportunities in vertical industries; third, the establishment of an independent CEO mechanism for ecosystem enterprises is more suitable for market-oriented development.

In the face of generative AI, known as the "fourth industrial revolution," SenseTime's advantages lie in years of technological accumulation, AI infrastructure, and application ecosystem. According to the financial report, in the first half of 2024, SenseTime's generative AI business revenue reached 1.051 billion yuan, a substantial year-on-year increase of 255.7%, with revenue accounting for 60.4%, becoming the company's new pillar business.

However, the challenges ahead for SenseTime should not be underestimated. AI is in a technological race, which is an unprecedented global competition. At the same time, the significant gap between the huge investments in AI and commercial revenues will still exist in the short term, representing uncertainty for the participating players.

After delving into the AI field for 10 years, SenseTime has finally welcomed the wave of generative AI, and it will set off again, striving to seize this opportunity