IFLYTEK(002230) 东方证券·2024-05-04 13:06Investment Rating - Buy (Maintained) [1] Core Views - The company released its 2023 annual report and 2024 Q1 report, with 2023 revenue reaching 19.65 billion yuan (+4.41% YoY) and net profit attributable to the parent company of 657 million yuan (+17.12% YoY) [14] - The company's increased investment in core technology and AI model development has impacted short-term profitability but laid a solid foundation for long-term competitive advantages [14] - The Spark cognitive model has been widely applied across multiple industries, including education, healthcare, automotive, office, and industrial sectors, driving the development of the developer ecosystem [14] - The company continues to invest in AI models, with significant improvements in capabilities such as logical reasoning, language understanding, and text generation, approaching the level of GPT-4 Turbo [14] Financial Performance - Revenue for 2023 was 19.65 billion yuan, with a YoY growth of 4.4% [8] - Net profit attributable to the parent company for 2023 was 657 million yuan, with a YoY growth of 17.1% [10] - Gross margin for 2023 was 42.7%, with a net margin of 3.3% [11][12] - Revenue is expected to grow to 31.27 billion yuan by 2026, with a CAGR of 14.1% from 2023 to 2026 [8][9] Valuation and Target Price - Target price is set at 50.21 yuan, based on a PS valuation method with a 5x multiple for 2024 [14] - The current stock price as of April 30, 2024, is 45.11 yuan [1] Industry and Market Performance - The company operates in the computer industry in China [3] - The stock's 52-week high and low prices are 81.88 yuan and 37.4 yuan, respectively [2] - The A-share market capitalization is 104.447 billion yuan [2] Key Financial Metrics - EPS for 2023 was 0.28 yuan, expected to grow to 0.49 yuan by 2026 [11] - ROE for 2023 was 3.9%, expected to increase to 5.8% by 2026 [12] - The company's P/E ratio for 2023 was 158.9, expected to decrease to 91.9 by 2026 [13] Strategic Focus - The company is focusing on AI model development, particularly the Spark cognitive model, which has been applied across various industries [14] - The company has partnered with Huawei to develop a domestic computing platform capable of supporting trillion-parameter model training [14] - The Spark V3.5 model, trained on the domestic platform, has shown significant improvements in multiple capabilities, including logical reasoning and language understanding [14]