When the "hematopoietic inflection point" collides with the "valuation discount", SenseTime is brewing a storm of value reassessment.

At the end of March, the leading AI company in the Hong Kong stock market, SenseTime Technology (0020.HK), released its 2025 performance report. Hidden in this financial report is a milestone figure - the company achieved an EBITDA (earnings before interest, taxes, depreciation, and amortization) of 380 million RMB in the second half of 2025, marking its first positive turn since going public. Furthermore, SenseTime's operating cash flow also turned positive for the first time in the second half of 2025. Throughout 2025, SenseTime's cash turnover cycle significantly shortened from 228 days to 129 days. The improvement in these data indicates that SenseTime has developed a self-sustaining ability to generate cash flow. The group's revenue is also impressive. Total revenue exceeded 5 billion RMB, a year-on-year increase of 33% and setting a new historical high; the loss narrowed significantly by 58.6% compared to the previous year. Looking at the different business sectors, SenseTime's visual AI business is transitioning from a period of technological investment to a period of scalable returns. As a cornerstone business, visual AI plays a key role as a "cash cow". The financial report for 2025 shows that the CV2.0 business not only achieved revenue growth, but also made a profit for the first time, maintaining positive cash flow for two consecutive years. This is attributed to SenseTime's "universal-specialty integration" algorithm production system - using large models to understand complex scenarios and small models to efficiently execute tasks, addressing the pain points of traditional AI deployment, where computational power consumption and recognition accuracy are difficult to balance. Obviously, SenseTime, which has held the top position in the Chinese computer vision market for nine consecutive years, has now polished this business into a cornerstone that provides stable cash flow. If visual AI is SenseTime's "shield," then generative AI business is its "spear" piercing through the skies. The financial report shows that revenue from generative AI business increased by 51% year-on-year, reaching 3.63 billion RMB, accounting for over 70% of total revenue. In 2025, SenseTime continued to invest in cutting-edge technology research and fully solidified its leadership position in the global multi-modal large model field. The new series of large models improve multi-modal reasoning ceilings every day; the SenseNova-SI spatial intelligent model is the first to break the boundary between virtual and reality; the release of the new NEO architecture completes a thorough reshaping of the native multi-modal underlying logic. These breakthrough advances not only redefine the efficiency of model inference and training paradigms, but also indicate that SenseTime has taken a completely new step in the path to multi-modal integration. On the application level of large models, SenseTime has delved into various areas such as general offices, finance, intelligent marketing, content creation, with customers ranging from automotive, intelligent terminals, consumption, the internet, embodied intelligence, finance, education, healthcare, and a thousand other industries. In cooperation with Beijing Zhidemai Technology, SenseTime has helped them improve store operation efficiency by 20 times, live stream operation efficiency by 6 times, traffic investment efficiency by 5 times, and real-time monitoring efficiency by 3 times. As the solid technological foundation in the "trinity" strategy, the SenseTime large installation achieved a deep transition from a technological weakness to an industrial closed-loop in 2025. While pursuing algorithm efficiency, SenseTime has established an AI intelligent system that is the first in the world to connect the entire chain from "computing power management - IDC operation - energy storage system", using large models to predict power loads, dynamically scheduling computing tasks, and is expected to bring about a 7% saving on electricity costs. Although SenseTime has presented a financial report that can be considered the "strongest in history," the market's reaction seems relatively lagging. The current stock price creates a significant "scissors gap" with the company's record revenue, positive EBITDA, and leading global technological strength. Perhaps the underestimation stems from the market not realizing that SenseTime has reached a crossroads in the valuation system transition. Previously, the market commonly used PS valuation for unprofitable AI companies; now that EBITDA and cash flow have both turned positive, the valuation benchmark should shift towards PE (price-to-earnings ratio) and PCF (price-to-cash flow ratio) valuations for "profit-growth companies". Based on SenseTime's current situation, if it can maintain its positive trend in EBITDA in 2026 and achieve sustained improvement in operating cash flow, its valuation system will undergo a systematic reevaluation. Considering that AI companies generally operate at a loss, we can use the valuation levels of mature SaaS companies globally as a reference. The EV/Revenue multiples for mature SaaS companies typically range from 8-12 times, whereas SenseTime's multiple is currently less than 5 times, indicating a clear room for restoration. Although the market holds inertia towards the AI industry, when a company possesses a "cash cow" (profitable CV2.0), a "growth pole" (rapidly growing generative AI), and a "moat" (big devices with algorithm-electricity synergy), along with being the only one in China to achieve a closed-loop of "computing power - model - application", its value reassessment is only a matter of time. Historical experience shows that while the emotions of the capital market fluctuate in the short term, the intrinsic value of a company is long-term. Companies that can maintain financial health and continuous technological investment even in industry downturns will ultimately gain market appreciation.