--- title: "Iluvatar CoreX general-purpose GPU revenue increased by 149.6% year-on-year, with the inference series surging by 238.2%." type: "Topics" locale: "en" url: "https://longbridge.com/en/topics/39725538.md" description: "Iluvatar CoreX's general-purpose GPU revenue increased by 149.6% year-on-year, with the inference series surging by 238.2%. Recently, Iluvatar CoreX, a leading Chinese provider of general-purpose GPU products and AI computing power solutions, announced its full-year 2025 results. The data shows that in 2025, the company achieved revenue of RMB 1,033.6 million, a year-on-year increase of 91.6%; gross profit was RMB 558.0 million, up 110.5% year-on-year; adjusted net loss was RMB 437.7 million..." datetime: "2026-04-06T07:35:31.000Z" locales: - [en](https://longbridge.com/en/topics/39725538.md) - [zh-CN](https://longbridge.com/zh-CN/topics/39725538.md) - [zh-HK](https://longbridge.com/zh-HK/topics/39725538.md) author: "[亚太商讯那些事儿](https://longbridge.com/en/profiles/18018177.md)" --- > Supported Languages: [简体中文](https://longbridge.com/zh-CN/topics/39725538.md) | [繁體中文](https://longbridge.com/zh-HK/topics/39725538.md) # Iluvatar CoreX general-purpose GPU revenue increased by 149.6% year-on-year, with the inference series surging by 238.2%. Iluvatar CoreX General GPU Revenue Up 149.6% YoY, Inference Series Surges 238.2% Recently, Iluvatar CoreX, a leading Chinese provider of general GPU products and AI computing power solutions, announced its full-year 2025 performance. Data shows that in 2025, the company achieved revenue of RMB 1,033.6 million, a year-on-year increase of 91.6%; gross profit was RMB 558.0 million, up 110.5% YoY; adjusted net loss was RMB 437.7 million, narrowing by 32.1% YoY; gross margin improved from 49.1% in the same period last year to 54.0%, indicating a steady profitability trend. In terms of business structure, revenue from the company's general GPU products surged 149.6% from RMB 369.6 million in 2024 to RMB 922.6 million in 2025, accounting for 89.3% of total revenue; revenue from AI computing power solutions decreased by 42.2% from RMB 166.2 million in 2024 to RMB 96.1 million in 2025; other income and gains increased significantly from RMB 45.0 million in 2024 to RMB 97.0 million in 2025. Notably, the gross margin of AI computing power solutions rose from 31.7% in 2024 to 41.5% in 2025, mainly due to a shift in customer demand towards configurations less reliant on externally procured components, leading to a corresponding decrease in the proportion of third-party procured components in the company's solutions. Based on its self-developed Tiangang and Zhikai series chips, the company provides customers with full-stack solutions ranging from single servers to intelligent computing center clusters. During the reporting period, the company generated revenue of RMB 96.1 million, accounting for 9.3% of total revenue for the same year. As the artificial intelligence industry transitions from technological breakthroughs to large-scale implementation, the demand for AI computing power has also evolved from a sole focus on hardware parameters to a comprehensive competition encompassing the full-stack capabilities of hardware, software, and services. Against this backdrop, the company has made multiple advancements in technological innovation, commercialization, and ecosystem development, further consolidating its position in the domestic general GPU and AI computing power solutions market. At the R&D level, the company adheres to the principle of independent development, following the rhythm of "mass-producing one generation, designing the next, and pre-researching another," consistently implementing a hardware-software co-design philosophy, continuously advancing product iteration, and striving to provide high-quality computing power support for customers. It is reported that the company has completed in-depth adaptation for dozens of mainstream open-source and commercial large models domestically and internationally. Relying on its full-stack self-developed software stack, the company possesses forward-looking adaptation capabilities that enable native support for new model structures and new operators on Day0 of their release, seamlessly matching the rapid iteration pace of large model technology. Simultaneously, the company launched a specialized large model acceleration toolkit with lossless quantization technology at its core, achieving a leap in performance across the entire process while fully preserving model accuracy, improving long-text processing efficiency by over 50%, increasing system computing power utilization by over 60%, and reducing cluster communication overhead by over 30%. The related technology has successfully supported production-level cluster deployments for leading large model manufacturers, with chip computing power utilization reaching industry-leading levels. In terms of business expansion and customer base, the company continues to grow. As of December 31, 2025, the company has served over 340 customers from various industries, with its products and solutions accumulating over 1,000 deployments and applications in important industries such as financial services, healthcare, and transportation. Meanwhile, the number of projects deployed by the company for major industry customers, including internet, large AI models, scientific research, finance, healthcare, education, and transportation, continues to increase. Looking at its development trajectory in recent years, the company has maintained a sustained and rapid growth trend. Data shows that the company's revenue steadily increased from RMB 189.4 million in 2022 to RMB 289.0 million in 2023, and further grew to RMB 539.5 million in 2024. This growth is primarily attributed to the continuous iteration of the company's products and solutions, the maturing of its commercialization capabilities, and the ongoing expansion of its high-quality customer base. Regarding software ecosystem development, during the reporting period, the company released a new generation of software development platform, natively compatible with mainstream GPU programming models in the industry, providing full-stack software support for the release of computing power from its self-developed GPGPU chips and their large-scale commercial use; after deep optimization, the platform achieved an improvement in code migration efficiency of over 80%, supports "minute-level" smooth migration of operators, and its compiled code performance leads the industry. At the same time, the company established a full-scenario self-developed core acceleration library system, with performance generally improving by over 20% compared to the previous version, deeply adapting to the full range of cutting-edge large model application scenarios, effectively improving chip execution efficiency, significantly lowering the barrier to entry for customers, and comprehensively enhancing the platform's engineering implementation capabilities. Currently, the company's software ecosystem has entered a new stage of deep performance optimization and full-scenario adaptation. Looking ahead, the company will continue to integrate global resources, accelerate technological iteration, and promote the transformation of technological breakthroughs into computing capabilities for a wider range of industries. ### Related Stocks - [ILUVATAR COREX (09903.HK)](https://longbridge.com/en/quote/09903.HK.md)