CoinW Research Institute
On September 14, Suiryan Technology closed at 424.26 yuan, up 6.87% from the first day's closing price, with a daily transaction volume of 2.899 billion yuan and a turnover rate of 39.49%, estimating the average transaction price at about 410.08 yuan. Although the transaction amount decreased by 48.68% compared to the first day, the closing price has already surpassed both the first and second day's average transaction prices, indicating an improvement in short-term support. However, whether such a rebound can continue still needs to be assessed further in light of industry changes, new product deliveries, and subsequent profitability.
Volume Reduction Rebounds Improve Short-term Support, Transaction Focus Still Awaits Upward Shift
On September 14, Suiryan Technology opened low at 370 yuan, then gradually rebounded, eventually closing at 424.26 yuan, close to the day's high of 427.94 yuan. Shares purchased on the first day can now be sold, but after a morning decline, the stock price returned to above the previous closing price of 397 yuan and further reclaimed the first day's average transaction price of about 415 yuan, indicating a noticeable support force at lower levels, which alleviates the lack of support at high transaction levels on the first day.
However, based on the overall transaction distribution, the price focus has not completely risen. The average transaction price on September 14 was approximately 410.08 yuan, still about 1.2% lower than the first day's average transaction price, indicating that a considerable portion of transactions occurred at lower levels that day. The rise in closing price reflects an increase in late trading buying, but whether this establishes a more stable price foundation still depends on whether it can continue to hold the average transaction price during subsequent declines.
The decline in transaction volume also needs to be interpreted cautiously; a volume increase can indicate a weakening of selling pressure but does not directly equate to the disappearance of selling pressure. At the time of Suiryan Technology's initial public offering, approximately 17.9 million shares were freely traded, accounting for only 4.16% of the total shares. When there are fewer circulating shares, slight changes in buying and selling force can amplify stock price fluctuations. If transaction volumes gradually return to normal levels afterward, and the stock price can maintain the first day's average transaction price while driving the average transaction price higher daily, then short-term shares can be considered further stabilized.
Individual Stocks Rebound Against the Trend, But Domestic Computing Sector Has Yet to Form a Resonance
On September 14, Suiryan Technology rose 6.87%, while during the same period, Cambricon remained flat, Moore Threads fell 3.31%, and Muxi shares dropped 5.71%, with the Sci-tech Innovation 50 Index down 1.62%. These numbers indicate that funds were primarily reassessing Suiryan's own trading price that day, without forming a widespread buying trend across the domestic computing sector.
This differentiation is not surprising, as Suiryan is closer to being a computing supplier within Tencent's AI supply chain, and its revenue and orders are highly correlated with Tencent's procurement timing; Cambricon, Moore Threads, and Muxi have different customer structures, product lines, and stages of commercialization. Although Suiryan's current valuation is relatively high, it hasn't reached a level sufficient to reshape the pricing system of the entire sector. Therefore, the lack of follow-up rises from peers does not mean the market denies Suiryan; it merely indicates that this round of growth is primarily driven by its own circulating shares, Tencent's orders, and changes to shares during the initial public offering.
In terms of industry direction, the growth in AI inference demand continues to provide Suiryan with a business foundation. According to the China Telecom Institute's estimates for this weekend, with the increase in AI agent applications, the proportion of inference computing power in overall computing demand is expected to continue to rise. From 2023 to 2025, over 80% of Suiryan's revenue from accelerated cards and modules is projected to come from inference products, which aligns well with the changes in industry demand for the company's existing business direction. However, the growth in industry demand will not automatically translate into company profits, as inference customers are more concerned about equipment procurement costs, deployment efficiency, software compatibility, and actual usage costs. Whether Suiryan can benefit from industry expansion ultimately hinges on product performance, supply capability, 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 approximately 182.581 billion yuan, equivalent to about 184 times its expected revenue for 2025. Given the company's rapid revenue growth projected for 2026, simply using the previous year's revenue for calculations would exaggerate this multiple. However, such a high valuation still requires the company to maintain fast growth and gradually reduce losses. In the first half of 2026, Suiryan achieved revenue of 1.12 billion yuan, already exceeding the whole year's revenue for 2025, but the net loss attributable to the parent company during the same period was still 632 million yuan. The focus of the market's attention will shift from whether revenue can grow to whether revenue growth can lead to reduced losses.
Cutting-edge Model Development Slows, Can Tencent's Orders and L600 Sustain Growth?
Suiryan's business foundations mainly lie in inference, while changes in the rhythm of cutting-edge model development primarily affect the computing power investments on the training side, so the two are not directly correlated. For Suiryan, what is more important is whether relevant discussions further affect AI capital expenditures and the overall computing budget of customers, and how this indirectly transmits to the company's subsequent orders through clients like Tencent.
At the same time recently, Anthropic CEO Dario Amodei has suggested that the pace of advancing cutting-edge model capabilities should be appropriately slowed down, and OpenAI CEO Sam Altman and xAI head Elon Musk later publicly expressed support. Since cutting-edge model development primarily impacts the computational investment on the training side, these statements will not directly change Tencent's procurement from Suiryan in the short term; however, they might trigger new concerns in the market about AI capital expenditures peaking or slowing down. If cloud vendors reduce their budgets for data center expansion, training cluster construction, and AI chip procurement as a result, the negative impact might initially suppress the valuations of the AI hardware supply chain, then gradually transmit along customer budgets, procurement plans, and supplier orders, ultimately affecting Suiryan. On the other hand, if some expenditures shift from large-scale training to deploying existing model inference, cost optimization, and practical application, it might continue to support demand for inference computing power. Therefore, for Suiryan, the focus should not be on whether industry leaders support slowing model development, but on whether Tencent and other customers' overall AI capital expenditures decline, whether the share of inference budgets increases, and whether related procurements can timely translate into deliveries and revenues.
This transmission chain first falls on Tencent, as of 2025, in terms of end customer metrics, Tencent-related sales account for 83.79% of Suiryan's revenue. The company has stated during its IPO roadshow that it has signed Tencent orders that can be executed until 2027, providing some visibility for subsequent deliveries and revenue recognition. However, this concentration of orders also means that the company's performance is more susceptible to the budget and procurement rhythm of a single customer. If Tencent invests more resources in deployment and application expansion, whether Suiryan can continue to benefit will depend also on its share in Tencent's procurements and the progress of deliveries; if the procurement rhythm slows, revenue growth will be directly affected.
At the same time, extending growth beyond Tencent to more customers will hinge on whether the L600 can convert testing demand into bulk procurement. As of the date of signing the listing announcement, this fourth-generation integrated training inference module has not yet entered large-scale delivery. The company disclosed that the L600 has passed hardware and model matching with three potential internet clients and is currently advancing to gray testing, with large-scale deliveries expected in 2027. Transitioning from testing to revenue requires confirmation of orders, mass production, acceptance, and payment; the gross margin after mass production will determine whether the new revenue can truly improve profitability.
Regarding this transmission chain, related contracts already launched by CoinW can serve as supplementary observations. Tencent (TENCENT) can be used to observe core customer expectations for AI investments, while Nvidia (NVDA) and AMD can be used to gauge global demand and competition for AI chips, and TSMC (TSM) provides a reference for advanced manufacturing and packaging conditions. These targets can help assess the external environment, but how many orders Suiryan ultimately secures, how much delivery it achieves, and how much profit it improves will still rely on the company's own disclosures.
In summary, the fact that Suiryan Technology stood back above the first day's average transaction price under a clear decline in transaction volume the day after its listing indicates that short-term support has improved. The increasing discussions around slowing down cutting-edge model development do not imply that computing power demand will immediately decline. Should industry investments shift more towards deploying existing model inference, cost optimization, and scaled applications, Suiryan's business structure largely based on inference products may benefit. However, whether these potential changes translate into actual company growth still depends on whether Tencent's orders can sustain delivery, whether new customers can transition from testing to bulk procurement, and whether L600's mass production can drive improvements in gross margin and narrow losses. Only if an upward shift in transaction focus aligns with improvements in orders, delivery, and profitability will the current valuation obtain more solid operational support; if only a price rebound occurs, it should still be regarded as a short-term correction from the initial public offering stage.
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。
