CoinW Research Institute
On September 14, at the close, Suiryan Technology reported 424.26 yuan, up 6.87% from the closing price on the first day, with a total transaction volume of 2.899 billion yuan and a turnover rate of 39.49%, estimating an average transaction price of about 410.08 yuan. Although the transaction amount decreased by 48.68% compared to the first day, the closing price has now surpassed both the first and second day's average transaction prices, indicating an improvement in short-term support. However, whether such a rebound can be sustained still needs to be assessed in conjunction with industry changes, new product deliveries, and subsequent profit changes.
Volume Reduction Increases Short-term Support, Transaction Focus Still Needs to Shift Up
On September 14, Suiryan Technology opened low at 370 yuan, then gradually rebounded, ultimately closing at 424.26 yuan, close to the day's highest price of 427.94 yuan. Shares purchased on the first day can already be sold; however, after the price dropped in the morning, it regained the previous closing price of 397 yuan and further recovered to about 415 yuan, the first day's average transaction price, indicating that there is significant support at a low price, alleviating the previous lack of support for high-priced transactions on the first day.
However, looking at the overall transaction distribution for the day, the price focus has not entirely shifted upwards. The average transaction price on September 14 was about 410.08 yuan, still approximately 1.2% lower than the first day's average transaction price, indicating that a considerable amount of trading occurred at lower levels that day. The rising closing price reflects increased buying pressure towards the end of trading, but whether it forms a more stable price foundation still depends on whether the average transaction price can be maintained during subsequent declines.
A decrease in transaction volume also needs to be understood cautiously; a volume rise could indicate a weakening of selling pressure, but it cannot be directly equated to the disappearance of sell pressure. Suiryan Technology had about 17.9 million freely tradable shares at the beginning of its listing, accounting for only 4.16% of the total share capital. When the tradable chips are few, slight changes in buying and selling forces can magnify price volatility. If the transaction volume gradually returns to normal levels, the stock price can maintain the first day's average transaction price and continue to drive an increase in daily average transaction prices, then the short-term shares can be considered further stabilized.
Individual Stocks Rise Against the Trend, but Domestic Computing Power Sector Has Not Formed Resonance
On September 14, Suiryan Technology rose 6.87%, while Cambricon remained flat, Moore Threads dropped 3.31%, and MuXi shares fell 5.71%. The Sci-Tech Innovation 50 Index decreased by 1.62%. This set of data indicates that on that day, capital was primarily reassessing Suiryan's own trading price and did not form a generalized buying pressure covering the domestic computing power sector.
This divergence is not surprising; Suiryan is closer to the computing power manufacturers in Tencent's AI supply chain, with its revenue and orders highly correlated to Tencent's procurement pace; the client structures, product lines, and commercialization stages of Cambricon, Moore Threads, and MuXi are not the same. Although Suiryan's current valuation is at a high level, it is not yet high enough to reshape the entire sector's pricing system. Therefore, the fact that peers did not follow suit does not mean the market is denying Suiryan, but only indicates that this round of rises is mainly driven by its own tradable shares, Tencent's orders, and changes in the chip structure during the initial listing.
From an industry perspective, the growth in AI inference demand still provides a business basis for Suiryan. According to the China Telecom Research Institute, it is expected that with increased application of AI intelligent agents, the proportion of inference computing power in overall computing power demand will continue to rise. From 2023 to 2025, more than 80% of Suiryan's income from accelerated cards and modules comes from inference products, which is in line with the current business direction and changes in industry demand. However, growth in industry demand will not automatically translate into company profits; inference customers are more concerned with equipment procurement costs, deployment efficiency, software compatibility, and actual usage costs. Whether Suiryan can benefit from industry expansion ultimately depends on product performance, supply capabilities, and whether customers are willing to continue purchasing.
Based on the latest closing price of 424.26 yuan, Suiryan Technology's total market capitalization is about 182.581 billion yuan, equivalent to about 184 times the expected revenue for 2025. Due to the company’s rapid revenue growth in 2026, solely using the previous year's revenue for calculation will exaggerate this multiple, but such a high valuation still requires the company to maintain rapid growth and gradually reduce losses. In the first half of 2026, Suiryan achieved revenue of 1.12 billion yuan, surpassing that of the whole year of 2025, but the net loss attributable to the parent for the same period was still 632 million yuan. The market's focus moving forward will shift from whether revenue can grow to whether revenue growth can lead to a reduction in losses.
Frontier Model Development Slows Down, Can Tencent's Orders and L600 Continue to Drive Growth?
Suiryan’s business foundation primarily lies in inference, and changes in the rhythm of frontier model development first affect the computing power input on the training end; thus, the two are not directly correlated. For Suiryan, it is more crucial to pay attention to whether relevant discussions further impact AI capital expenditures and customers’ overall computing power budgets, and indirectly convey to the company’s subsequent orders via procurement plans from clients like Tencent.
Recently, Anthropic CEO Dario Amodei suggested that the pace of enhancing frontier model capabilities should be appropriately slowed, which was publicly supported by OpenAI CEO Sam Altman and xAI head Elon Musk. As frontier model development mainly affects the computing power input on the training side, these statements will not directly change Tencent's procurement from Suiryan in the short term but may trigger new concerns in the market about AI capital expenditures peaking or slowing down. If cloud vendors reduce budgets for data center expansions, training cluster builds, and AI chip procurement, the negative impact may initially suppress the valuations of the AI hardware supply chain, then gradually transmit through customer budgets, procurement plans, and supplier orders, ultimately affecting Suiryan. On the other hand, if some investments shift from large-scale training towards deployment of existing models, cost optimization, and application implementation, it may continue to support the demand for inference computing power. Therefore, for Suiryan, what truly needs observation is not whether industry leaders support slowing model development, but whether overall AI capital expenditures from Tencent and other clients are declining, whether the proportion of inference budgets is increasing, and whether related procurement can be timely transformed into deliveries and revenues.
This transmission chain primarily falls on Tencent, which, according to end customer estimates for 2025, attributes 83.79% of Suiryan's revenue to Tencent-related sales. The company indicated during its IPO roadshow that it has signed Tencent orders effective until 2027, providing some visibility for subsequent delivery and revenue recognition. However, concentrated orders also mean that the company's performance is more susceptible to impacts from a single client's budget and procurement pace. If Tencent directs more resources towards inference deployment and application expansion, whether Suiryan can continue to benefit will depend on its share in Tencent's procurement and the delivery progress; if the procurement pace slows, revenue growth will also be directly affected.
Additionally, extending growth from Tencent to more clients hinges on whether the L600 can transform testing demand into bulk procurement. As of the day the listing announcement was signed, this fourth-generation training-inference integrated module had not yet entered large-scale delivery. The company disclosed that the L600 has passed hardware and model matching from three potential internet clients and is advancing through gray testing, with large-scale delivery expected in 2027. Transitioning from testing to revenue requires order confirmation, mass production, acceptance, and payment; the gross margin after mass production will determine whether new income can truly improve profitability.
Surrounding this transmission chain, CoinW has launched related contracts that can serve as supplementary observation. Tencent (TENCENT) can be used to observe expectations for core customers' AI investments, NVIDIA (NVDA) and AMD can be used to observe global AI chip demand and competition, and TSMC (TSM) provides a reference for advanced manufacturing and packaging conditions. These targets help assess the external environment, but how many orders Suiryan can ultimately obtain, how much can be delivered, and how much profit can be improved still depend on the company's own disclosures.
In summary, Suiryan Technology standing above the average transaction price of the first day with a transaction amount significantly lower than the first day indicates an improvement in short-term support. The heated discussion on slowing frontier models does not imply that computing power demand will immediately decline. If industry investments shift more towards the deployment of existing models, cost optimization, and scalable applications, Suiryan's business structure, mainly based on inference products, may benefit. However, whether this potential change can translate into actual growth for the company still depends on whether Tencent's orders can continue to be delivered, whether new clients can transition from testing to bulk procurement, and whether L600 can drive improvements in gross margins and narrow down losses after mass production. Subsequent simultaneous increases in the transaction focus and improvements in orders, deliveries, and profitability will provide more solid operational support for the current valuation; if only the price rebounds, it should still be viewed as a short-term correction during the initial listing phase.
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。
