Author: ICVIEWS Editorial Department
Just now, Enflame Technology officially listed on the SSE STAR Market with an issue price of 142.18 RMB. As of press time, the intraday stock price reached 437 RMB, surging over 200%, with the market capitalization briefly exceeding 200 billion RMB.
What is more worth a closer look than the first-day stock price is the list of names revealed in the strategic placement roster. Tencent, as Enflame's largest customer and long-term shareholder, continued to increase its commitment in this strategic placement, subscribing to nearly 250 million RMB with a 36-month lock-up period. Xiaomi, ZTE Corporation, GigaDevice, Hua Hong Group, Shanghai INESA, Tongfu Microelectronics, and the National Social Security Fund have also joined the fray, with a 12-month lock-up. This is not merely a financial investment, but a collective bet by the upstream and downstream of the industrial chain — from wafer manufacturing to packaging and testing foundry services, and from internet customers to industrial capital. All parties have expressed their confidence in domestic cloud AI chips with real money.
Enflame Technology has allocated the 6.1 billion RMB raised into three accounts: approximately 1.5 billion RMB for the fifth-generation AI chips, about 1.2 billion RMB for the sixth-generation AI chips, and the remaining 3.3 billion RMB for the collaborative innovation project of advanced AI software and hardware. In short, the funds will be spent on two things: continuing to iterate hardware and truly building up the software ecosystem.
01. Cumulative Loss of About 4.3 Billion RMB Over Three Years, Expected to Achieve Consolidated Profitability as Early as This Year
According to the prospectus, from 2023 to 2025, Enflame Technology's operating revenues were 301 million RMB, 722 million RMB, and 990 million RMB, respectively. The net profits after deducting non-recurring gains and losses were -1.567 billion RMB, -1.503 billion RMB, and -1.197 billion RMB, respectively, indicating that profitability has not yet been achieved. Considering factors such as orders on hand, product delivery pace, employee cost budget, and R&D planning, Enflame Technology expects to achieve consolidated profitability in 2026 or 2027.
Regarding the reasons for not achieving profitability so far, Enflame Technology attributes it to two main factors. First, the R&D investment is massive. In terms of hardware for cloud AI chips, advanced wafer manufacturing and packaging and testing processes are required to iterate products rapidly on a yearly basis. On the software side, it is necessary to build and continuously improve the AI computing and programming software platform to ensure deep support for continuously iterating mainstream AI large models. At the same time, joint R&D with supply chain partners is needed to ensure supply chain stability. Second, facing the demands of internet giants, the company needs to jointly verify and refine multiple generations of products, continuously adapting and optimizing software to meet business scenario requirements, and achieving a match between product solutions and customers' commercial value before gradually scaling up deployment.
Over 96% of Enflame Technology's operating revenue comes from its main business, while other business revenue mainly stems from a small amount of raw material resale and equipment leasing. Among them, AI accelerator cards and modules, as well as intelligent computing systems and clusters, constitute the company's main sources of revenue. AI accelerator cards and modules are primarily sold to leading internet customers. With hardware performance iteration and deep software adaptation, they are deployed in an increasing number of AI business scenarios. The sales revenue has grown rapidly, and in 2025, the proportion of revenue from such products in the main business exceeded 85%.
From 2023 to 2025, the sales volumes of Enflame Technology's AI accelerator cards and modules were 21,173, 21,789, and 64,890 units, respectively. In 2025, the third-generation AI accelerator cards continued to scale up in the AI business scenarios of internet customers, and the annual sales volume further increased year-on-year compared to 2024.
From January to June 2026, Enflame Technology expects its operating revenue to show a substantial year-on-year growth trend, and it is expected to achieve the full-year revenue scale of 2025 in the first half of 2026. Due to the company's sustained R&D intensity, coupled with the impact of taxes and surcharges generated from advance receipts from customers and accrued interest expenses, the year-on-year growth rate of net profit attributable to the parent company is lower than the expected year-on-year growth rate of operating revenue for the same period.
In addition, due to the company's business expansion and strategic stockpiling, the inventory scale of Enflame Technology has increased year by year. The book balances of inventory during the reporting period were 261.429 million RMB, 979.3923 million RMB, and 1,074.2783 million RMB, respectively. Considering the rapid technological iteration of cloud AI chips, inventory depreciation losses are prudently provisioned based on the sales expectations for each period during the reporting period. During the reporting period, the amounts of inventory depreciation losses were 127.7516 million RMB, 30.2783 million RMB, and 31.6062 million RMB, respectively.
02. Four Generations of Architecture and Five Cloud AI Chips Iterated Over 8 Years
Over the 8 years since its establishment, Enflame Technology has independently developed and iterated four generations of architecture and five cloud AI chips, building a complete product system covering AI chips, AI accelerator cards and modules, intelligent computing systems and clusters, and AI computing and programming software platforms. According to IDC data and the issuer's sales volume, in 2025, the sales volume of Enflame Technology's AI accelerator cards and modules reached 66,000 units, corresponding to a market share of approximately 1.7% in China's AI accelerator card market, ranking among the top domestic AI chip manufacturers.
Cloud AI chips are the core underlying hardware of Enflame Technology's product matrix and are not sold separately to the outside world. Products of different generations are applied in the training and inference scenarios of AI models. Based on the company's cloud AI chips, the products integrate supporting memory modules, power and signal management modules, cooling components, and high-speed interfaces onto PCB substrates or OAM modules. A single AI accelerator card and module generally contains one cloud AI chip.
Intelligent computing systems and clusters are one of the main product forms contributing to revenue. The intelligent computing system brand is CloudBlazerPOD, which is a single-cabinet, single-computing-node device for AI model training and inference scenarios, constituting the smallest computing unit of the intelligent computing cluster. The CloudBlazerPOD is generally composed of 4 to 8 AI-dedicated servers and multiple network switches interconnected via high-speed cables and integrated into a single cabinet. A single AI-dedicated server is generally based on 8 of the company's AI accelerator cards or modules, equipped with hardware such as CPUs, DRAM memory, enterprise-grade solid-state drives, power and power supply modules, cooling modules, and network cards, as well as system software. A single CloudBlazerPOD typically integrates 32 to 64 AI accelerator cards and modules.
An intelligent computing cluster is a computing infrastructure at the computer room level with multi-computing nodes and multi-cabinet interconnections. Based on multiple CloudBlazerPODs, the intelligent computing cluster is further equipped with system hardware such as CPU general-purpose servers, high-speed network equipment, and independent storage servers, along with self-developed supporting system software, to achieve high-speed interconnection for clusters of a thousand cards and above. Currently, Enflame Technology only delivers products using the intelligent computing cluster model to one customer, Chengdu Gaoxin Electronics, acting as the general project integrator.
The AI computing and programming software platform is the core software platform of the product matrix. Enflame Technology provides a unified AI computing and programming software platform, TopsRider, for its self-developed AI chips, AI accelerator cards or modules, and related products, to link AI chips with AI applications. Currently, the company does not sell the AI computing and programming software platform separately to the outside world.
03. Conclusion
The numbers in Enflame Technology's prospectus are very honest: a loss of 4.3 billion RMB over three years, a steadily climbing inventory scale, and a market share of only 1.7% in the domestic AI accelerator card market in 2025. On the other hand, NVIDIA still occupies nearly 60% of the market share in China. Such a market share not only shows how much room there is for localization but also makes people realize: space does not equal share; share must be won product by product and customer by customer.
Enflame Technology is taking a pragmatic path: instead of pursuing comprehensive substitution right from the start, it dives into the real business scenarios of internet giants, relying on the hard work of hardware iteration and software adaptation to integrate its products into customers' production processes. In 2025, the sales volume of AI accelerator cards surged to 66,000 units, with over 85% of main business revenue coming from this segment. Tencent transformed from a customer to a shareholder and then to a strategic placement investor. These signals indicate that domestic chips are truly shouldering the workload on the production lines.
The "Four Little Dragons" have gathered in the capital market, turning the page to a new chapter in the story of domestic GPUs. Moore Threads, Muxi Semiconductor, Biren Technology, and Enflame Technology—these four companies have different strategies and paths, but they all reach the same destination: trying to prove one thing—the foundation of China's AI computing power cannot be held in the hands of others forever. For Enflame Technology, there are still many tough battles to face next: how long can the raised funds last? Can the fifth- and sixth-generation chips make further progress in performance and ecosystem? Can the market share of domestic chips be pushed higher before NVIDIA's next-generation products are released?
There are no ready-made answers to these questions. But at least, the stage is set, and the upstream and downstream of the industrial chain are standing together. The most difficult moment for domestic GPUs, the moment from 0 to 1, has passed. What comes next is the tough battle from 1 to 10, and this is the moment that will truly determine who can stay.