Sensetime-W (0020.HK) recently released its full-year 2025 results report. The data shows that the company's total revenue reached RMB 5.015 billion during the reporting period, a year-on-year increase of 32.9%, hitting a record high. The annual net loss was RMB 1.782 billion, narrowing by 58.6% year-on-year. More notably, the EBITDA for the second half of 2025 reached RMB 376 million, achieving its first positive turn since listing.
Looking beyond the impressive headline figures, after business adjustments and strategic transformation, Sensetime's revenue growth has shown a reliance on its generative AI business. The innovative business units are still far from becoming a "second growth curve." Against the backdrop of AI expansion, rising computing power costs and depreciation have put pressure on gross margins. Behind the significant reduction in losses lies the company's divestment of persistently cash-burning operations, while its core businesses remain in a loss-making state. Since 2018, Sensetime has consistently failed to achieve an annual profit.
More critically, the valuation logic for the AI sector has shifted. Compared to traditional AI companies, the capital market has turned its favor towards new, more imaginative general-purpose large model players. While former Sensetime Vice President Yan Junjie led MiniMax to a market capitalization of HKD 300 billion, Sensetime, once the leader among the "AI Four Dragons," has long fallen out of the industry's top tier, with its market capitalization now less than a quarter of MiniMax's.
Generative AI has become the dominant revenue driver. The most prominent part of Sensetime's 2025 annual report is undoubtedly its generative AI business. The financial report indicates that revenue from the generative AI business increased from RMB 2.404 billion in 2024 to RMB 3.630 billion in 2025, a growth of 51.0%. This business now accounts for 72.4% of total revenue, up from 63.7% in 2024, making it Sensetime's primary revenue source.
This growth is primarily driven by the continued explosion in demand for generative AI model training, fine-tuning, and inference, as well as integrated industrial solutions promoting the joint commercialization of computing platforms, models, and applications.
Behind the generative AI boom is Sensetime's restructuring of its core narrative. In 2023, the company reorganized its previous four business lines—Smart Business, Smart Living, Smart Auto, and Smart City—into three new segments: Generative AI, Traditional AI, and Smart Auto. By the end of 2024, this was adjusted again to Generative AI, Smart Auto, and Visual AI. In the first half of 2025, Sensetime changed its revenue reporting segments to Generative AI, Visual AI, and X-Innovation Business.
Benefiting from its forward-looking layout of "AI infrastructure - large models - applications," Sensetime's generative AI business revenue has successively broken through the RMB 10 billion, 20 billion, and 30 billion marks over the past three years, with its revenue share increasing by 37.6 percentage points.
Beneath the growth story of Sensetime's transformation from a computer vision leader to a generative AI service provider, there are underlying concerns. Compared to the explosive growth rates of 199.9% and 103.1% in 2023 and 2024 respectively, the growth rate of the generative AI business, while still high, has shown signs of deceleration.
Furthermore, while scale is increasing, it does not seem to have translated into "economies of scale." In 2025, Sensetime's cost of sales increased by 37.4% year-on-year to RMB 2.958 billion, with AIDC computing power operational costs surging by 163.5%. Amid the ongoing expansion of the generative AI business, server depreciation, hardware procurement, and cloud service expenses remain high, leading to a further decline in the company's overall gross margin to 41.0%.
As generative AI leaps to become the core growth engine, Sensetime also faces the test of whether its business structure is imbalanced.
In the first half of 2025, Sensetime's Visual AI segment absorbed the smart cockpit business from the former Smart Auto unit. Meanwhile, under the "1+X" strategy, capital-intensive and long-cycle businesses like Smart Auto, Smart Healthcare, Home Robots, and Smart Retail were consolidated into the X-Innovation Business segment.
As a leading company in China's computer vision field, Sensetime's foundational Visual AI business was once its largest revenue source. However, since 2023, revenue from this segment has declined significantly, dropping 39.5% to RMB 1.112 billion in 2024. After the reporting change in 2025, Visual AI revenue was RMB 1.047 billion, a slight increase of 3.4% year-on-year.
In fact, To-G businesses, primarily focused on smart city and security projects, have long faced challenges such as project-based delivery, long payment cycles, and saturated demand. Sensetime's proactive scaling back and shift in focus for this segment also indirectly suggests that the Visual AI business has limited growth potential and is unlikely to become a second growth curve, instead focusing on its profitability. As stated in the financial report, the Visual AI business is transitioning from a technology investment phase to a scale harvest phase, becoming a solid pillar driving the group's revenue growth and cash flow improvement.
In contrast, the X-Innovation Business, representing future development directions, has shrunk. In 2025, revenue from this segment fell 5.9% year-on-year to RMB 302 million, with its revenue share dropping from 8.6% to 6.0%. Sensetime attributed this decline to the spin-off of its smart driving business in August 2025.
The deeper issue is that while sectors like smart healthcare, home robotics, and smart retail are considered promising, they are still in the investment phase, with some having unclear commercialization paths. They are neither capable of rapid scaling nor of generating stable profits in the short term.
Sensetime revealed in its report that under its "Flagship (Group) + Speedboats (Ecosystem)" collaborative system, ecosystem companies incubated by the group have made smooth progress in primary market financing, receiving high recognition from external capital, including internet giants, top venture capital firms, and industrial funds. However, the report did not disclose specific financial data or project progress for these "X" innovation businesses.
The "significant reduction in losses" is a notable highlight in Sensetime's financial report. It was reported that Sensetime executives frequently mentioned this key point during the post-earnings conference call.
In terms of data, Sensetime's net loss for 2025 was RMB 1.782 billion, narrowing by 58.6% year-on-year. The adjusted net loss was RMB 1.956 billion, narrowing by 54.3%. To highlight performance improvement, Sensetime emphasized that the second-half 2025 EBITDA was RMB 376 million, achieving its first positive reading since listing.
In fact, in the first half of 2025, Sensetime had already spun off the loss-making smart driving and world model businesses from its "Jueying" Smart Auto unit, contributing to the overall net loss reduction. Additionally, the AI GPU chip business has been excluded from the consolidated statements since the beginning of 2025, and the terminal chip company has been deconsolidated since the second half of the year. According to Sensetime's plan, the X-Innovation businesses will ultimately be spun off from the group's consolidated financial statements.
Reflected in the 2025 financial report, Sensetime recorded an "other net gains" item exceeding RMB 1.9 billion, mainly comprising a gain of RMB 1.313 billion from the disposal of subsidiaries and associates, and a fair value gain of RMB 646 million on financial assets measured at fair value through profit or loss.
For comparison, this figure was RMB 539 million in 2024 and RMB 641 million in the first half of 2025.
Furthermore, to achieve the loss reduction, Sensetime implemented comprehensive control over its "three expense" items. Sales expenses, already down 20% year-on-year in 2024, decreased a further 13.1% to RMB 569 million. Administrative expenses also decreased by 16.2% year-on-year to RMB 1.226 billion.
More critically, R&D expenses decreased by 8.6% year-on-year to RMB 3.775 billion. A detailed look shows this reduction was mainly due to decreased employee benefits expenses, partially offset by increases in server operations and cloud service fees. In 2024 and the first half of 2025, Sensetime's R&D expenses increased by 19.2% and 12% respectively, primarily driven by depreciation and amortization related to generative AI business investments, along with server operations and cloud service fees.
Disclosed employee numbers corroborate this. As of the end of 2025, Sensetime had 2,472 employees, a reduction of 1,284 compared to 3,756 at the end of 2024. Extending the timeline, the company had over 5,000 employees at the end of 2022, showing a逐年下降 trend.
In summary, excluding the RMB 1.276 billion contribution from asset sales in the second half of the year and the impact of reduced R&D spending, Sensetime's core businesses remain in a state of significant loss.
Looking over a longer period, since its establishment over a decade ago, the pressure to achieve profitability has persistently been a sword of Damocles hanging over Sensetime.
Historical financial data shows that from 2019 to 2024, Sensetime's annual losses were RMB 4.968 billion, RMB 12.158 billion, RMB 17.177 billion, RMB 6.093 billion, RMB 6.495 billion, and RMB 4.307 billion respectively. Adding the 2025 loss of RMB 1.782 billion, Sensetime's cumulative losses over the past seven years have reached a staggering figure of approximately RMB 53 billion.
On the difficult path towards a profitability inflection point, Sensetime continues to rely on external funding. The 2025 annual report shows that as of December 31, 2025, the net proceeds of HKD 2.787 billion from the December 2024 placement, and the net proceeds of HKD 2.498 billion from the July 2025 placement, have been fully utilized. The net proceeds of HKD 3.146 billion from the December 2025 placement are expected to be fully used by the end of 2026.
Most of the funds are designated to support the development of Sensetime's core AI businesses, including AI infrastructure construction, generative AI R&D, large model product deployment, and the integration and application of AI in innovative fields.
Contrasting with the persistent losses and need for external funding is the simultaneous "blood loss" in the capital market for Sensetime. As of March 26, Sensetime's share price was HKD 1.85, with a total market capitalization of less than HKD 75 billion.
This stands in stark contrast to its peak at the end of 2021 when it listed—its IPO price was HKD 3.85, and its share price surged rapidly after listing, with gains once exceeding 22%, and its market cap briefly surpassing HKD 150 billion.
In a dramatic turn of events, just before Sensetime's IPO at the end of 2021, then Vice President Yan Junjie chose to leave. He founded an AI company called MiniMax in early 2022.
Earlier this year, Yan Junjie's MiniMax listed on the Hong Kong Stock Exchange, with its share price doubling on the first day, pushing its market cap above HKD 100 billion. In the following two months, it continued to attract strong capital interest, and its market capitalization has now reached HKD 300 billion. Contrasting with its high valuation, this new AI large model powerhouse reported full-year 2025 revenue of $79.038 million, a year-on-year increase of 158.9%, but incurred a massive annual loss of $1.872 billion, an increase of 302.3% year-on-year.
Behind the valuation gap lies the capital market's fundamentally different pricing logic for new versus old AI companies. Compared to Sensetime's "loss reduction" narrative, companies like MiniMax and Zhipu, despite having larger loss scales, command higher premiums from investors due to their combined advantages in AI application business models, high growth expectations, and scarcity value. Investors are more willing to pay for such "growth stories."
Comments