"Alliance of AI stragglers" seeks a change of life

Nowadays, AI applications are like a Korean idol group where newcomers come out in large numbers-a new version is launched before the face is recognized, and then there is a crackling campaign. As a result, there is always a feeling that the work and strength are not worthy of the traffic. As for those AI predecessors who debuted ten years ago, most of them faced problems such as technical route risks, transformation difficulties, and lack of support from financial owners, leaving one or more echelons behind. These two types of companies can have a common name: the AI Stragbehind Alliance. Take Shangtang, a former industry leader, as an example. When it went public in Hong Kong in 2021, Gui was the world's most valued AI unicorn company, but its share price has been falling deeply since 2022. At present, its market value has evaporated by more than HK$300 billion compared with its peak. Shangtang Stock Price Performance (since 2022) The exit of its founding team is even more regrettable. Mr. Tang Xiaoou, who has strong scientific research strength and is known as the pioneer of Face Recognition technology, passed away suddenly in 2023. Co-founder Xu Bing also announced his resignation as executive director and secretary of the board of directors before the Dragon Boat Festival this year, and was appointed as the head of the AI chip business. Shangtang emphasized in the announcement that this job adjustment is part of the company's overall talent deployment and business focus. To some extent, it can also be understood that the core technologies of the AI 1.0 era cannot be used to lead the 2.0 era. In order to successfully transform, Shangtang showed its determination to start a business again. There are still many AI pioneers like Shang Tang who are working hard to move towards the 2.0 era. Of course, this also includes some back waves who have become front waves at an extremely fast speed. transformation Perhaps the most prominent event in the entire AI 1.0 era was in 2016, when AlphaGo defeated South Korea's top chess player Lee Se-Seok. But compared to the deep learning technology behind AlphaGo, contemporary China AI startups are mainly betting on computer vision technology (CV). The so-called computer vision (CV) technology mainly refers to Face Recognition and image detection technology. This track has given birth to the famous AI Four Little Dragons, namely Shang Tang, Yun Cong, Kuang Shi and Yitu. The reason why computer vision technology became the earliest sub-track to achieve breakthroughs in the AI 1.0 era is the result of the combined effect of technology maturity, clarity of application scenarios, data and hardware support, and market demand. Most of the founding teams of companies such as Shangtang and Guangshi come from top academic institutions such as the Chinese University of Hong Kong and Tsinghua University. They are the kind that can really write cutting-edge academic papers. The concentrated emergence of scientists and entrepreneurs has greatly shortened the transformation cycle of technology from laboratory to industrialization, creating conditions for the rapid growth of enterprises. Eleven years ago, the Face++ Face Recognition system launched by Obstacle Technology in 2014 had an accuracy rate of 99.5%, far exceeding traditional methods. This technological breakthrough directly promoted the paying propensity in security, finance and other industries, forming the commercial foundation for the early implementation of artificial intelligence technology. China's huge application scenarios provide computer vision companies with unique data advantages. Shangtang cooperates with the government to obtain traffic monitoring data, while Yuncong relies on bank customers to accumulate financial scenario samples to further upgrade its core technology. The AI Four Little Dragons developed very rapidly at that time and became big stars in the capital market. But all this is based on two most important assumptions: technological leadership will inevitably translate into commercial advantage, and government support can continue to create demand. After 2020, both assumptions were overturned. In terms of policy, China's artificial intelligence strategy will shift to application implementation after 2020. The shift in policy focus has made it difficult for a business model that relies on a single policy support to be sustainable. This change in national-level strategy will largely determine the investment decisions of investors. Simply put, investors are no longer willing to provide unlimited bullets to the AI Four Little Dragons, allowing them to continue to burn money on research and development and ignore losses. In the later period, the market began to use a magnifying glass to examine the financial situation of the four AI dragons, and hoped that their products could generate actual business and cash flow in various industries. Taking Shangtang as an example, although its revenue of 3.772 billion yuan in 2024 achieved a year-on-year growth of 10.8%, it is still 19.7% lower than the historical peak of 4.7 billion yuan in 2021. The net loss in 2024 will be 4.278 billion yuan, equivalent to burning 11.72 million yuan per day. Although the loss will be 33.7% narrower than the 6.43 billion yuan in 2023, the absolute value still exceeds the annual revenue. Under financial pressure, human resource costs have eventually become an unavoidable problem for the four AI dragons: Shangtang's total employee count has been reduced from 6113 to 4672 since 2021, and non-core businesses have been abolished; Cloud has implemented a 20% salary cut for all employees since last year. With the move, the core technical team has left this year; Yitu Technology's layoffs have reached an astonishing 70%. After ChatGPT opened the AI 2.0 era with completely different technical routes, technological transformation, especially proving to the market that it has the ability to enter the 2.0 era, has become the most important task for the four AI dragons. changing fate Compared with the AI 1.0 era, the most remarkable feature of the 2.0 era is the rise of large-scale pre-training models. These models are pre-trained on massive data to master general language and image understanding capabilities, and then can be adapted to various specific tasks through fine-tuning. The emergence of ChatGPT marks a fundamental transformation in artificial intelligence technology. Artificial intelligence in the 2.0 era seems to be the artificial intelligence that social progress really needs. The fundamental differences in technical architecture between AI 1.0 and AI 2.0 directly lead to the huge differences in application scenarios and business values between the two. Liang Wenfeng changed the landscape of China's artificial intelligence industry Take the fourth paradigm as an example. This company developed a machine learning platform ten years ago and is now transforming into an Agent service provider. This kind of business iteration itself can explain the changes in industry trends. The so-called AI Agent refers to an agent that can actively sense the environment, plan goals, and use tools to perform tasks.