In 2025, generative artificial intelligence transitioned from a "battle of a hundred models" to comprehensive penetration across application layers.
With the conclusion of the A-share annual reporting season, AI application companies, represented by Kingsoft Office, iFlytek, BlueFocus, and Kunlun Tech, have delivered their 2025 report cards, showing differentiated but directionally similar results. Revenue scale continued to expand, with AI capabilities deeply embedded into core products. Metrics for users and API calls grew rapidly. However, intensive research and development investments continued to compress short-term profits, indicating the industry is still in a critical stage of transitioning from technological breakthroughs to commercial validation.
According to Wind data, companies related to the A-share AI application theme generally achieved revenue growth in 2025, but profitability diverged significantly. The scale effects of leading companies coexisted with the breakout pressures faced by small and medium-sized enterprises, leading to a further increase in industry concentration.
Thirty-five companies saw net profit attributable to shareholders increase by over 50%.
Financially, the primary driver of revenue growth for AI application companies in 2025 clearly came from the restructuring and penetration of AI capabilities into existing products.
Looking at the overall sector performance, the financial results of AI application-related companies varied markedly. Among 116 companies in the AI application theme, 28 achieved year-on-year operating revenue growth exceeding 30%. These included Cambricon, Hongbo Co., Ltd., and Asiainfo Security, primarily concentrated in sub-sectors like AI computing chips, servers, marketing, and gaming.
Thirty-five companies reported net profit attributable to shareholders growing by more than 50%. Specifically, Shengyi Technology's net profit surged by 273.52%, Runze Technology increased by 182.07%, Perfect World rose by 156.76%, and BlueFocus grew by 177.29%.
Conversely, 40 companies experienced a decline in operating revenue. Among them, XuanYa International, Montnets Cloud Tech, and Talkweb Information saw declines exceeding 20%, reflecting an uneven landscape within the industry and significant transformation pressures for some traditional businesses.
Kingsoft Office reported approximately 5.929 billion yuan in operating revenue for 2025, a 15.78% year-on-year increase. Net profit attributable to shareholders was about 1.836 billion yuan, up 11.63%. The growth was largely driven by AI-powered office solutions. During the reporting period, the monthly active users of WPS AI in China surpassed 80.13 million, a 307% increase year-on-year. The daily token calls generated by WPS AI services exceeded 200 billion, growing more than twelvefold.
Kingsoft Office explicitly stated in its annual report that it has initiated exploration into a native office paradigm centered around AI agents. By leveraging task understanding, workflow orchestration, and multi-model scheduling mechanisms, the company is redefining office interaction models, pushing office software to evolve from functional tools into intelligent agents. Furthermore, WPS Lingxi has been upgraded to an "All-in-One AI Office Partner" and introduced three specialized agents for presentations, spreadsheets, and documents, precisely tailored to high-frequency workplace scenarios, achieving a leap from "understanding intent" to "actually getting work done."
iFlytek demonstrated a different approach, building upon its large model foundation to amplify commercialization capabilities through its platform and industry-specific applications. In 2025, iFlytek achieved operating revenue of approximately 27.105 billion yuan, a 16.12% increase year-on-year. Net profit attributable to shareholders was about 839 million yuan, rising 49.85%.
iFlytek mentioned in its annual report that its AI open platform offered 920 AI capabilities and solutions in 2025. The community of AI developers surpassed 10 million, with large model developers reaching 2.29 million, marking a 124% annual growth rate. Revenue from the AI platform and authorization services reached 1.252 billion yuan, within which revenue from large model APIs and MaaS platform services was 385 million yuan, a 263% increase.
BlueFocus, a representative enterprise in AI marketing, reported operating revenue of 68.693 billion yuan for 2025, a 12.99% year-on-year increase. Net profit attributable to shareholders was 225 million yuan. Its overseas business revenue reached 56.496 billion yuan, accounting for over 82% of total revenue and growing 16.89% year-on-year.
Kunlun Tech's model aligns more closely with an "AI content platform + application portfolio." In 2025, the company achieved operating revenue of 8.198 billion yuan, a significant 44.78% increase year-on-year. However, it reported a net loss attributable to shareholders of 1.593 billion yuan. Its short-form video and AI short-form video platform reached a monthly gross transaction volume nearing $36 million, with Annual Recurring Revenue exceeding $400 million. Revenue from its overseas AI Agent browser, Opera, reached $615 million, a 27.92% increase.
In its 2025 annual report, Kunlun Tech proposed that AGI is no longer a philosophical question but an economic one. With enhancements in individual and organizational productivity, the goal of AGI has achieved an "epic-level convergence."
Intensive R&D and marketing investments remain the norm.
Despite noticeable revenue growth, the profitability and investment structure of AI application companies indicate the industry is still in its early stages. Financial data also reveals significant divergence within the AI sector.
Among the 116 theme companies, 34 reported net losses attributable to shareholders in 2025, accounting for approximately 30% of the total. Companies like CloudMinds, Chinese All, Tom猫, and DeepGlint reported losses exceeding 100 million yuan.
On one hand, some upstream companies in the AI industry chain demonstrated strong profit elasticity.
Shengyi Technology reported 2025 revenue of 19.292 billion yuan, up 79.77% year-on-year. Net profit attributable to shareholders was 4.312 billion yuan, surging 273.52%, making it one of the companies with the highest profit growth within the AI application theme.
Sheng Yi Technology reported revenue of 28.431 billion yuan for 2025, a 39.45% increase. Net profit attributable to shareholders was 3.334 billion yuan, up 91.75%.
These companies are primarily in upstream AI computing hardware segments like PCBs and CCLs, benefiting from the explosive demand for AI servers and showcasing the ability to monetize their position in the hardware infrastructure chain.
On the other hand, not all AI-themed companies achieved profit growth.
Transsion Holdings reported 2025 revenue of 65.591 billion yuan, down 4.55% year-on-year. Net profit attributable to shareholders was 2.581 billion yuan, a sharp decline of 53.49%, indicating challenges with both revenue and profit decreasing.
37 Interactive Entertainment reported revenue of 15.966 billion yuan, down 8.46% year-on-year. Although its net profit attributable to shareholders grew by 8.5%, overall growth remained sluggish.
From the perspective of the industry's overall investment structure, high-intensity R&D spending remains mainstream. Kingsoft Office's annual R&D expenditure reached 2.095 billion yuan, a 23.57% increase, with an R&D expense ratio of 35.34%. iFlytek's R&D investment accounted for 19.79% of its operating revenue. Kunlun Tech stated directly in its 2025 report that the primary reason for the loss was a significant increase in R&D and market investment: annual R&D expenses reached 1.676 billion yuan, up 8.6%, while sales expenses soared to 4.182 billion yuan, an 81.53% increase. The company defined these as strategic investments, noting that "these two strategic investments have increased operating costs in the short term, putting pressure on the company's current profits."
Judging from their statements, several companies share a similar assessment of the current phase.
Kingsoft Office pointed out that AI Agents are evolving from assistive tools into intelligent executors. Intelligent agents based on large model technology have achieved a paradigm shift from "conversation" to "capable of doing work," able to independently complete complex tasks like creating PPTs, data analysis, and meeting minutes summarization. Multi-Agent collaborative systems are beginning to see scaled application, accelerating the transformation of office software into AI-driven operational workflow hubs.
iFlytek believes that artificial intelligence, as a strategic digital technology leading the new wave of scientific and technological revolution and industrial transformation, is spawning countless new products, scenarios, and business formats, injecting innovative vitality into the high-quality development of the digital economy.
Kunlun Tech stated that the overall AI industry is still in a phase of rapid development, with market demand continuously growing. The company's strategic direction aligns with the industry's long-term trends. However, it noted that the pace of technological iteration in this field is extremely fast, requiring companies to continuously maintain innovation capabilities to avoid being replaced by emerging technologies.
A review of the overall performance of AI application theme stocks in 2025 clearly shows that the AI application industry has not yet entered a stage of stable profitability, and business models are still being explored. On one hand, the application layer is beginning to establish revenue scale and a user base. On the other hand, high-intensity R&D and marketing investments remain the norm, putting pressure on short-term profits. In 2025, AI application companies resembled entities in a commercial validation period: metrics like user scale, API call volume, and gross transaction volume grew rapidly on one end, while high R&D expense ratios and significant profit volatility were evident on the other.
During this stage, differentiation among companies is becoming apparent. Some are leveraging AI to transform existing products, others are amplifying capabilities through platform ecosystems, and some are seeking monetization paths through content and application portfolios. Regardless of the path, the common prerequisite remains sustained investment.
Comments