The computing power sector is warming up. What trading signals should Suiruan Technology look for next?

CN
1 hour ago

CoinW Research Institute

As of the market close on September 15, Suiruan Technology reported 451.99 yuan, up 6.54% from September 14, and an accumulated increase of 13.85% from the closing price of 397 yuan on its first trading day. On that day, the overall rebound of the computing power chip concept saw Suiruan Technology rise more than 10% at one point, with domestic computing power stocks such as Cambrian, ChipON, Muxi, and Moer Thread also showing strength, resulting in a noticeable improvement in market sentiment compared to the previous trading day.

Based on a closing price of 451.99 yuan, Suiruan Technology's total market capitalization is approximately 194.51 billion yuan, about 3.18 times its issuance market value of 61.187 billion yuan, equivalent to about 196.5 times its projected operating revenue of 990 million yuan for 2025. After a rapid increase in valuation, the market needs to verify two clues simultaneously: one is whether Tencent's orders, L600 deliveries, and the expansion of non-Tencent clients can continue to drive revenue growth; the other is whether recent discussions in the global AI industry regarding the development pace of cutting-edge models will change the computing power investment structure of large tech companies. If capital shifts more from large-scale training to inference, application deployment, and data center efficiency improvement, Suiruan's product structure, primarily focused on inference, may receive new support; if discussions evolve towards a contraction in the overall AI budget, a high customer concentration will amplify order and valuation fluctuations.

Bullish Logic: Tencent Orders and Inference Demand Remaining Valuation Support

The bullish logic is primarily based on revenue growth and the visibility of Tencent orders. In the first half of 2026, Suiruan achieved operating revenue of 1.12 billion yuan, a year-on-year increase of 279.08%, which has already exceeded the total for the entire year of 2025, which is 990 million yuan; the company expects operating revenue for the first three quarters of 2026 to be between 2.3 billion and 3 billion yuan, requiring about 1.18 billion to 1.88 billion yuan to be contributed in the third quarter. Tencent remains the most crucial source of demand, with Tencent-related sales accounting for 83.79% of the company’s revenue according to terminal customer metrics in 2025. The company indicated during its IPO roadshow that it has signed Tencent orders that can be executed until 2027. If the third quarter revenue is close to the higher end of the management's expected range and Tencent's procurement, delivery, and acceptance pace remains stable, the market's expectations for high growth will receive more direct financial data verification.

The second layer of support comes from Suiruan’s product structure focused on inference; in 2025, training and training-inference integrated products accounted for only 1.15% of AI accelerator card and module revenue, with about 98.85% primarily coming from inference products. Therefore, existing revenue has relatively limited direct dependence on the scale of frontier model training and relies more on ongoing calls and application deployments after the model goes live. As scenarios such as search, advertising recommendations, office, enterprise services, and intelligent agents expand, inference demand is expected to form an independent growth driver; at the same time, the fourth-generation training-inference integrated module L600 has completed hardware and model matching for three potential internet clients and entered gray testing. If it can subsequently shift to bulk procurement, Suiruan's growth sources will have the opportunity to expand from Tencent and inference business to more clients and a more complete product line.

Bearish Logic: Company Profitability and Customers Await Validation

The core basis for bears is that valuation has clearly advanced ahead of profit realization. Based on the closing price of 451.99 yuan on September 15, the company’s total market capitalization is approximately 194.51 billion yuan, with a static price-to-sales ratio of about 196.5 times, expanding from approximately 61.187 billion yuan at issuance to 3.18 times. In the first half of 2026, the company's gross margin was 30.37%, with a net loss attributable to the parent company of 632 million yuan, and the net cash flow generated from operating activities was negative 878 million yuan. Rapid revenue expansion has not yet led to synchronized improvements in profit and cash flow. At the current valuation level, the market not only requires revenue to continue to grow rapidly but also expects scale effects, product structure optimization, and cost dilution to gradually narrow losses. Therefore, any revenue, gross margin, or delivery pace below expectations may amplify valuation fluctuations.

At the same time, customer concentration and the commercialization pace of new products further increase this sensitivity. In 2025, the revenue share of the top five customers accounted for 96.89%, with Tencent-related sales accounting for 83.79%. Although signed orders have increased short-term revenue visibility, Tencent's capital expenditure, procurement share, product acceptance, and pricing changes will directly impact the company's performance. Meanwhile, L600 is still in the stage of transitioning from technology validation to large-scale commercialization, and the fourth-generation product has not yet entered large-scale production and delivery. From gray testing to stable income, it still needs to complete order confirmations, production ramp-up, software adaptation, customer acceptance, and cost dilution.

Therefore, whether the current bearish logic holds true will ultimately depend on subsequent operating data validation. If actual revenue in the third quarter approaches the upper limit of the management's expected range, with improvements in gross margin and a significant reduction in net loss attributable to the parent company, and if internet clients other than Tencent begin to place bulk orders and form repeat purchases, market concerns about customer concentration and profitability paths will significantly diminish. Conversely, if revenue growth continues to rely mainly on concentrated delivery from Tencent, and if L600 remains stuck in gray testing or small-scale stages, with non-Tencent customers unable to form large-scale procurement, then a market valuation close to 200 billion yuan will still primarily depend on support from future growth expectations. Changes in AI budgets, fluctuations in order rhythms, or declines in sector valuations could lead to larger price adjustments.

AI Capital Expenditure Structure Facing Adjustment, Can Suiruan Benefit from the Growth of Inference Demand?

Notably, attention should also be paid to new variables in the global AI industry, stemming from leading companies’ re-discussions on the pace of cutting-edge model development. Anthropic CEO Dario Amodei suggested that the speed of enhancing cutting-edge model capabilities should be appropriately slowed down, allowing more time for safety assessments and governance mechanisms, and OpenAI and xAI subsequently also expressed support for pursuing cutting-edge capabilities more cautiously. Meanwhile, during a call at the All-In Summit attended by Nvidia CEO Jensen Huang, former US President Trump publicly opposed slowing down the entire industry due to extreme AI risk concerns. Huang acknowledged that safety issues need to be seriously addressed but opposed converting unfounded extreme risk predictions directly into reasons for restricting industry development.

The truly noteworthy aspect of this discussion is its potential to change the structure of newly added computing power investments rather than just affecting training demand. If frontier model companies slow down single capability leaps while continuing to expand the deployment of existing models in search, advertising, office, intelligent agents, and enterprise services, part of capital expenditure may shift from pursuing larger-scale training to inference calls, efficiency optimization, and data center construction. Since about 98.85% of Suiruan's revenue from AI accelerator cards and modules comes from inference products in 2025, this change provides some directional support for it; however, if safety discussions, financing costs, and investment return pressures further prompt large tech companies to reduce overall AI budgets, the inference side would also struggle to completely detach from overall capital expenditure changes. With Tencent-related sales still accounting for 83.79%, whether the company can benefit ultimately depends on whether Tencent continues to increase investments in AI infrastructure and whether new applications can convert into sustained inference calls and hardware repurchases.

Currently, CoinW has launched multiple stock contracts related to the AI industry chain, which can serve as a reference for observing changes in cloud vendors' capital expenditures, computing power demand, and data center construction. On the demand side, it can be combined with MSFTUSDT and GOOGLUSDT to observe expectations for investment in cloud computing, large models, and enterprise AI; computing power and semiconductor aspects can be observed through AVGOUSDT, TSMUSDT, MUUSDT, and SKHYNIXUSDT for AI chips, advanced manufacturing, and HBM demand, while infrastructure can be assessed with DELLUSDT and VRTUSDT to determine whether AI server and data center power supply and cooling construction continue to expand. If related cloud vendor metrics weaken while the semiconductor and infrastructure chains remain resilient, the market may be more inclined to trade on shifts in computing power investment towards inference deployment and data center efficiency improvements; if the demand side and hardware side weaken in unison, it will be necessary to remain vigilant about a potential decline in overall AI investment expectations. For Suiruan, these external metrics are more for observing the overall prosperity change of the AI industry chain, and the final judgment should still combine Tencent orders, expansion of non-Tencent clients, large-scale delivery of L600, and the ability of revenue growth to drive improvements in gross margin and shrink losses.

免责声明:本文章仅代表作者个人观点,不代表本平台的立场和观点。本文章仅供信息分享,不构成对任何人的任何投资建议。用户与作者之间的任何争议,与本平台无关。如网页中刊载的文章或图片涉及侵权,请提供相关的权利证明和身份证明发送邮件到support@aicoin.com,本平台相关工作人员将会进行核查。

Share To
APP

X

Telegram

Facebook

Reddit

CopyLink