Enflame Technology's Cumulative Losses Exceed 5.1 Billion Yuan: Gross Margin Plummets, Related Transactions with Tencent Account for Over 70%

Port Business Observer by Shi Zifu and Wang Lu

In recent years, driven by a sustained wave of capitalization in the domestic GPU industry, well-known industry enterprise Shanghai Suiyuan Technology Co., Ltd. (hereinafter referred to as Suiyuan Technology) submitted its application to the STAR Market on January 22, with CITIC Securities serving as the sponsor.

On February 11, the Shanghai Stock Exchange website announced that Suiyuan Technology has officially entered the inquiry stage. From a fundamental perspective, despite being backed by its largest shareholder Tencent, the company’s ongoing losses remain a cause for concern.

1

Continuous losses exceeding 5.1 billion yuan, significant decline in gross margin

According to the prospectus and Tianyancha, Suiyuan Technology was established in March 2018. The company is one of China’s leading enterprises in cloud AI chips, aiming to become a “leader in general artificial intelligence infrastructure.” Over nearly eight years, the company has independently developed and iterated four generations of architectures and five cloud AI chips, building a comprehensive product system that covers AI chips, AI acceleration cards and modules, intelligent computing systems and clusters, as well as AI computing and programming software platforms.

The prospectus states that in terms of underlying hardware, the company utilizes自主指令集 (self-developed instruction set), benchmarking NVIDIA’s Tensor Core acceleration computing units and NVLink interconnect technology. It has自主架构 (independent architecture) GCU-CARE acceleration computing units and GCU–LARE high-speed inter-chip interconnection technology. These architectures are not only flexible for programming but also deeply support high-parallelism acceleration computing for large AI models. On the software platform level, the company has developed its own full-stack AI computing and programming software platform, “Yuzuan TopsRider,” which includes drivers, compilation languages and compilers, operator libraries, and toolchains, independent of NVIDIA’s CUDA ecosystem. This platform links the company’s hardware with AI applications, significantly reducing programming difficulty and migration costs for mainstream AI models, enabling the hardware to better unleash performance in application scenarios. Regarding computing clusters, during the reporting period, the company’s Qianka and Wanka intelligent computing center projects have generated revenue. The company has also jointly developed超节点 (super-node) solutions with customers and built commercially valuable Wanka high-speed interconnect clusters.

Financial data shows that from 2022 to 2024 and the first three quarters of 2025 (within the reporting period), the company’s operating revenues were 901.038 million yuan, 3.01 billion yuan, 7.22 billion yuan, and 5.40 billion yuan, respectively. After deducting non-recurring gains and losses, net profits were -1.15 billion yuan, -1.567 billion yuan, -1.503 billion yuan, and -912 million yuan, with total accumulated losses exceeding 5.1 billion yuan.

Suiyuan Technology states that the company has not yet achieved profitability during the reporting period, mainly due to: 1) cloud AI chip hardware requires advanced wafer manufacturing and packaging/testing processes, with rapid product iteration on an annual basis; software needs to build and continuously improve AI computing and programming platforms to support deep iteration of mainstream AI large models; and also requires joint R&D with supply chain partners to ensure supply stability. These factors lead to substantial R&D investments; 2) the company faces demands from major internet companies, requiring joint validation and refinement of multiple generations of products, continuous software adaptation and optimization to meet business needs, and aligning product solutions with customer commercial value before scaling deployment.

During the reporting period, the company’s revenue scale has not been fully realized, unable to cover rigid R&D costs, which restricts short-term profitability. As of the end of September 2025, the accumulated losses under consolidated and parent company accounting were -4.165 billion yuan and -1.864 billion yuan, respectively. If losses persist, it could adversely impact the company’s operations and development, and the company would be unable to distribute dividends to investors.

In other words, while revenue continues to grow rapidly, the loss trend has not slowed. When Suiyuan Technology can turn losses into profits remains a key focus of external attention.

Meanwhile, the company faces significant international competitors.

The prospectus discloses that international firms represented by NVIDIA are the global standard-setters, technology leaders, and industry dominators in AI computing power. They lead in hardware performance, software ecology, system definition, and integration capabilities, surpassing domestic manufacturers. Furthermore, these international companies leverage their first-mover advantage to establish partnerships with top wafer fabs, leading IDM storage manufacturers, major AI large model developers, and internet companies, involving capacity locking, joint R&D, and mutual investments. By 2024, international firms hold over 70% of the AI acceleration card market in China, forming a stage of quasi-monopoly competition.

“From the current capital market’s favorability, Suiyuan Technology’s listing is not difficult; the real challenge is how to compete with NVIDIA and other international giants. From another perspective, the company also needs continuous breakthroughs in core technologies, which may prolong the timeline for turning losses into profits,” analysts noted.

Additionally, it is noteworthy that Suiyuan Technology’s gross margin has experienced a significant decline during the period, far weaker than industry averages.

During the reporting period, the company’s main business gross margins were 78.07%, 22.60%, 30.59%, and 36.23%, while the industry average gross margins were 35.77%, 57.47%, 56.97%, and 62.86%. Clearly, Suiyuan Technology’s gross margin has been halved, whereas comparable industry companies have seen substantial increases.

2

Over 70% related-party transactions with Tencent, continuous negative operating cash flow

In terms of major clients, during the reporting period, sales to the top five customers accounted for 94.97%, 96.50%, 92.60%, and 96.41% of the company’s operating income, respectively. Sales to Tencent Technology (Shenzhen) (including direct sales and AVAP mode sales) accounted for 8.53%, 33.34%, 37.77%, and 71.84%. Customer concentration is high. Tencent Technology (Shenzhen) is a related party, and both direct sales and AVAP mode sales to Tencent constitute related-party transactions.

According to Tianyancha and the prospectus, Tencent Technology is currently the largest shareholder of Suiyuan Technology, holding 19.9493%. The company states that it has established a stable cooperative relationship with Tencent. As cooperation deepens, sales and revenue proportions related to Tencent have increased during the reporting period, and related-party transactions are deemed reasonable. The AI acceleration cards and modules sold to Tencent are priced lower than similar products sold to third-party non-related customers, with prices negotiated and considered fair, given Tencent’s status as a strategic long-term customer.

As of the signing date of the prospectus, the actual controllers Zhao Lidong and Zhang Yalin directly hold 17.9287% of the company’s shares, and through employee shareholding platforms Suiyuan Huizhi and Suiyuan Chongying, indirectly control 10.2070%. The two controllers together control 28.1357% of the company.

In terms of accounts receivable, the balances at each period-end were 82.6581 million yuan, 247 million yuan, 518 million yuan, and 424 million yuan. The company makes bad debt provisions based on customer credit risk characteristics, with provisions at each period-end of 1.01%, 6.45%, 8.73%, and 9.18%. The bad debt provisions amounted to 8.353 million yuan, 159.533 million yuan, 452.198 million yuan, and 389.135 million yuan, respectively. Accounts receivable accounted for 90.81%, 76.86%, 65.44%, and 71.23% of revenue.

During the same period, inventory balances were 311 million yuan, 261 million yuan, 979 million yuan, and 1.148 billion yuan. Considering the rapid iteration of cloud AI chip technology, the company prudently provisions for inventory impairment based on sales expectations. The inventory write-down losses during the period were 33.1701 million yuan, 128 million yuan, 30.2783 million yuan, and 10.5504 million yuan.

For this IPO, Suiyuan Technology plans to raise 6 billion yuan, with 1.503 billion yuan allocated to the fifth-generation AI chip series R&D and industrialization projects, 1.197 billion yuan to the sixth-generation AI chip series R&D and industrialization projects, and 3.3 billion yuan to advanced AI hardware-software collaborative innovation projects.

The company believes that the funds raised align with industry trends in AI computing infrastructure, focusing on upgrading, expanding, and deepening existing core businesses and technologies, which are closely related to its main operations and strategic goals.

Looking at current cash flow, the company’s situation is mixed. During the reporting period, cash and cash equivalents were 345 million yuan, 1.511 billion yuan, 3.522 billion yuan, and 2.734 billion yuan, accounting for 21.38%, 59.47%, 60.43%, and 55.92% of total assets. However, net cash flow from operating activities has been continuously negative, with figures of -987 million yuan, -1.209 billion yuan, -1.798 billion yuan, and -770 million yuan.

Suiyuan Technology explains that the industry features high prepayment ratios for key procurement links, long delivery cycles from raw material procurement to final product delivery, and a rapidly expanding business scale. As the company continues to grow, there is a risk that operating cash flow will remain negative for some time. If the company cannot raise funds through equity or debt financing to effectively improve cash flow, it may face operational liquidity risks. (Produced by Port Finance)