In the third quarter of 2025, Iflytek Co., Ltd. reported an operating revenue of approximately 6.078 billion RMB, representing a year-on-year increase of about 10.02%. The net profit attributable to shareholders reached 172 million RMB, marking a substantial year-on-year growth of approximately 202.40%. The company's financial report emphasized its commitment to a "self-controllable" strategy in core AI technologies, including algorithms, computing power, and data.
**Building Fundamental Capabilities: Three-Pillar Approach of Algorithm, Computing Power, and Data** Iflytek highlighted its adherence to a self-controllable core technology foundation amid the complexities of international technological competition, ensuring that elements like algorithms, computing power, and data remain within its control. Specifically, in terms of computing power, the company unveiled a new version of its large model based on "national computing power" training. The report noted that on July 25, the upgraded version of the "Xunfei Spark X1," trained using wholly domestic computing resources, was launched. Although this version's parameter scale is an order of magnitude smaller than similar models in the industry, it achieved performance parity with advanced overseas models (e.g., OpenAI o3) in critical areas such as mathematical capabilities, translation, reasoning, and text generation. Furthermore, in terms of data and language support, the company mentioned that its base multi-language capabilities now extend support to over 130 languages. These initiatives demonstrate Iflytek's focus not only on "speeding up model performance" but also on emphasizing a "self-controllable" foundational ecosystem, which is significant for China's AI industry in seeking technological sovereignty amidst the international landscape.
**Product and Project Implementation: Transformation from Technical Platform to Business** With a solid foundation in place, Iflytek has transitioned into a phase focused on promoting large-scale AI applications through productization and projectization. The report disclosed key data indicating that in the first three quarters of 2025, the company ranked first in both quantity and value related to large model projects within the industry. In the third quarter, the contract value for large model projects amounted to approximately 545 million RMB, which is about 1.88 times the total of the second to fifth-ranked companies. The development of the developer ecosystem is also accelerating: as of the first three quarters, over 1.22 million new developers have joined the "Xunfei Spark" ecosystem, including 690,000 new large model developers. This indicates that Iflytek is making tangible progress in its "business and ecological" pathways while transforming its AI capabilities towards industry applications and ecosystem expansion.
**Internationalization and Multi-Language Strategy: A New Growth Point for AI Going Global** Notably, Iflytek has positioned "multi-language and international expansion" as a key component of its AI strategy. The financial report and related media coverage highlight that the "Xunfei Spark" large model base supports over 130 languages. In terms of international outreach, the company made its presence felt in the ASEAN market: on September 17, Iflytek introduced a multi-language large model base specifically developed for the ten ASEAN countries at the 22nd China-ASEAN Expo, built on entirely domestic hardware and software. At the same time, its consumer hardware products (such as smart office tools and translators) have performed exceptionally well overseas. The report disclosed that overseas consumer hardware revenue has tripled year-on-year. This strategy reflects that Iflytek is not only applying its AI capabilities domestically but is also proactively opening up international markets, using "language and intelligence" as its global entry point.
**Financial and Strategic Investments: Ensuring Deep Dives into AI** In advancing its AI initiatives, Iflytek is increasing its investment efforts. The report indicated that in the first half of the year, the company's R&D expenditure was 2.392 billion RMB, growing approximately 9.17% year-on-year, with R&D expenses accounting for 21.92% of total revenue. The company plans to issue no more than 100 million A-shares to specific targets, raising up to 4 billion RMB, of which 2.4 billion RMB is intended for computing power platform construction, 800 million RMB for "Spark Education large models and typical products," and the remaining 800 million RMB for working capital. These measures demonstrate the company's commitment to providing clear financial support for the three key aspects of "model upgrades, computing power support, and ecological construction." Future improvements in iteration capabilities, scalable application capacities, and internationalization will rely on such investments.
**Challenges and Points for Observation** Despite significant progress, several observation points remain from the report and external commentary: in terms of profitability, the company experienced strong net profit growth in the third quarter, yet faced losses in preceding periods. During the first three quarters, the cumulative net profit attributable to shareholders was a loss of approximately 67 million RMB. While the company emphasizes business transformation, government subsidies should not be overlooked in its profit growth. External media noted that government support is a critical contributing factor to the "202% net profit surge." Furthermore, in the transition from technology to application, "scalable monetization" capacity remains key. Although the large model foundation has been established, there is still potential for deep integration and maturation of business models in industries such as education, healthcare, and manufacturing.
**Towards a Development Model of Base, Ecosystem, and Global AI** Overall, Iflytek's development in the AI sector can be summarized in three phases: 1. **Foundation Building**: Based on algorithms, computing power, and data, emphasizing self-control and benchmarking against global leaders. 2. **Productization and Scene-based Implementation**: Transitioning technology to commercial application through large model project contracts and developer ecosystem construction. 3. **International Expansion**: Entering international markets through multi-language capabilities and hardware products while seeking new growth opportunities.
In the context of accelerated global AI competition and rising domestic demands, Iflytek has seized the threefold opportunities of "models, computing power, and internationalization." If it can steadily advance in monetization capabilities, continuous R&D iterations, deep international market engagement, and comprehensive industry coverage, its position in the AI sector is expected to further solidify.
Comments