Lead-in: Despite being in the same AI supply chain, some companies saw single-quarter net profit grow by more than sevenfold, while others declined by over 50%.
Recently, multiple A-share technology companies disclosed their third-quarter reports.
Kinwong Electronic reported a Q3 attributable net profit of CNY 1.014 billion, up 239.67% year-on-year; Shannon Chip Technology reported a Q3 attributable net profit of CNY 1.699 billion, up 743.02% year-on-year; SigmaStar Technology saw its Q3 attributable net profit rise 752.76% year-on-year. On the other side, Unigroup Guoxin Microelectronics saw its Q3 attributable net profit drop 56.63% year-on-year.
On one hand, profits are accelerating; on the other, performance is under clear pressure. Between these two sets of figures lies not just a growth rate gap of hundreds of percentage points.
Growing demand for AI servers is driving volume increases in high-performance PCBs (Printed Circuit Boards), storage, and certain chip products. However, as demand transmits to different segments, the results vary: some companies improve profitability through product upgrades, others benefit from rising storage prices combined with shipment expansion, while still others remain affected by changes in sales volumes and business structure.
It is also worth noting that the companies with the fastest profit growth do not necessarily have the easiest operating models. Storage distributors require significant capital for procurement and inventory buildup, computing power service providers need continuous equipment investment, and PCB manufacturers face challenges such as capacity expansion, accounts receivable, and increased inventory.
The AI market is entering a more specific phase: the market is beginning to ask where demand actually lands in terms of products, whether profits can be sustained, and how much cost enterprises must incur to achieve growth.
Two Fundamentally Different Growth Paths
Let's start with PCBs.
AI servers impose higher requirements on circuit boards. As data transmission rates increase, servers need to handle more complex signal transmission and connectivity issues, raising the technical standards for high-performance PCBs. For manufacturers capable of producing these products, opportunities arising from industry upgrades are reflected not only in increased orders but also in changes to product mix and profitability.
Kinwong Electronics' Q3 report serves as an example.
In Q3 2026, the company achieved revenue of CNY 6.621 billion, up 66.06% year-on-year; attributable net profit reached CNY 1.014 billion, up 239.67% year-on-year. For the first three quarters, revenue totaled CNY 15.233 billion, up 37.45% year-on-year; attributable net profit was CNY 1.616 billion, up 70.42% year-on-year.
The fact that single-quarter net profit growth far exceeded revenue growth indicates a marked acceleration in profit release.
The company stated in its financial report that performance growth primarily stemmed from increased revenue from high-performance PCB products in high-speed communications and AI data infrastructure, a higher proportion of value-added products, and price adjustments passed down to customers based on upstream material cost changes, alongside optimized order structures.
Here lies a key distinction: increased AI demand is merely the starting point. Whether a company can secure a better product mix and profit margins determines the elasticity of its performance.
However, Kinwong Electronics' growth must also be viewed within the context of its complete financial statements. In the first three quarters, net cash flow from operating activities was CNY 1.218 billion, down 20.27% year-on-year; both accounts receivable and inventory increased significantly compared to the end of the previous year.
This does not imply operational problems. Order growth often accompanies inventory buildup and capital occupation, but subsequent observation of accounts receivable recovery, inventory turnover, and new capacity utilization is necessary to determine if high growth can translate into sustained cash returns.
The growth logic of the storage supply chain differs from that of PCBs.
Shannon Chip Technology's Q3 report, disclosed on the evening of October 26, showed that for the first three quarters, the company achieved revenue of CNY 106.426 billion, up 303.14% year-on-year; attributable net profit reached CNY 5.340 billion, up 1,386.13% year-on-year. In Q3 alone, revenue hit CNY 46.221 billion, up 398.27% year-on-year; attributable net profit was CNY 1.699 billion, up 743.02% year-on-year.
The core highlight of this report is the resonance formed by rising storage product prices and expanded shipment volumes.
Shannon Chip Technology explicitly stated in its report that the substantial increase in revenue during the first three quarters was mainly driven by the combined effect of year-on-year rises in product selling prices and business scale expansion. Price hikes plus shipment volume boosts drove up sales scale; when revenue growth exceeds cost growth, profits gain greater room for expansion.
However, Shannon Chip Technology's business model requires separate understanding. The company focuses primarily on storage chip distribution, with minimal self-developed business volume. One cannot directly interpret its revenue exceeding CNY 100 billion as income from chip design or manufacturing, nor simply use revenue scale to gauge its technological content.
For distribution businesses, product prices, procurement costs, transaction volumes, and capital turnover all impact final profits. Industry price hike cycles can amplify profit elasticity but may also increase the capital required for procurement and inventory.
Therefore, the high growth of Shannon Chip Technology and Kinwong Electronics actually stems from two different industrial logics: the former is driven by the joint push of storage prices and transaction volumes, while the latter reflects improved profitability from growing demand for high-performance PCBs, price adjustments, and optimized order structures.
This is an easily overlooked aspect when observing AI hardware. Even though both companies benefit from AI infrastructure construction, their revenue composition, profit margins, and capital needs may be entirely different.
Chip Design: Ice and Fire Duality
If we shift our gaze from PCBs and storage to chip design, performance divergence is equally evident.
Montage Technology benefits from server memory architecture upgrades. According to its earnings preannouncement disclosed on October 21, the company expects Q3 revenue of approximately CNY 2.079 billion, up about 46% year-on-year; expected attributable net profit ranges from CNY 1.2 billion to CNY 1.3 billion, representing a year-on-year increase of 153.6% to 174.7%.
Specifically, the interconnect chip product line is expected to generate sales revenue of approximately CNY 2.047 billion in Q3, up 49.3% year-on-year, with gross margins estimated at 68% to 68.5%.
The company stated that applications like AI inference and Agents drive server computing power demand, while DDR5 memory architecture iterations further boost demand for memory modules and memory interconnect chips. In Q3, shipments of DDR5 memory interface and module companion chips increased, and the proportion of high-generation RCD chips in shipments also rose.
Compared to relying solely on product price increases, this growth path warrants closer attention to changes brought about by product iteration. Improvements in server performance place higher demands on memory bandwidth, data transmission, and interconnect capabilities. Companies that keep pace with technology upgrades and increase the share of high-generation products have the opportunity to capture greater profit elasticity from demand growth.
Another source of growth comes from edge-side AI.
SigmaStar Technology primarily designs, develops, and sells edge-side AI SoCs (System on Chips), IoT (Internet of Things) SoCs, and automotive electronics SoCs. The market it faces differs from data center chips, with products applicable to smart camera devices, smart home systems, automotive electronics, and other edge computing scenarios.
The company achieved Q3 revenue of CNY 1.917 billion, up 151.24% year-on-year; attributable net profit was CNY 0.701 billion, up 752.76% year-on-year. For the first three quarters, revenue totaled CNY 4.575 billion, up 111.21% year-on-year; attributable net profit reached CNY 1.564 billion, up 673.71% year-on-year.
The company disclosed that during the reporting period, product shipment volumes and average selling prices rose simultaneously. This means that sales volume growth and product price changes jointly drove performance.
SigmaStar Technology's performance highlights a noteworthy direction: growth opportunities in AI hardware exist not only in large data centers but also in smart terminals and edge devices, which may generate new chip demand. However, at the company level, it remains to be seen whether customer demand, product shipments, and subsequent orders can continue.
Looking at Unigroup Guoxin Microelectronics, the situation is markedly different.
The company reported Q3 revenue of CNY 1.414 billion, down 23.85% year-on-year; attributable net profit was CNY 0.248 billion, down 56.63% year-on-year. For the first three quarters, revenue totaled CNY 4.858 billion, down 0.95% year-on-year; attributable net profit reached CNY 1.045 billion, down 17.23% year-on-year.
In its Q3 report, the company explained that IC business sales volumes decreased year-on-year during the reporting period, and changes in product structure led to lower gross margins and reduced gross profit amounts.
These figures should not be simply interpreted as AI development impacting Unigroup Guoxin Microelectronics. They more directly illustrate that the semiconductor industry is not a monolith growing in sync. Differences in product applications, customer demand, and business cycles across companies mean that even if the entire AI industry continues to expand, it does not guarantee that every chip company will benefit simultaneously.
From PCBs to storage, then to memory interconnects and edge-side AI chips, the common trend revealed by Q3 reports is that industrial demand is transmitting to various segments, but the actual profit captured by enterprises ultimately depends on product competitiveness, pricing power, and business structure.
Financial Truth Behind the Computing Power Frenzy
On the flip side of high growth is the cost enterprises pay for that growth.
Shannon Chip Technology's net profit grew 1,386.13% year-on-year in the first three quarters, but as of the end of September, its inventory reached CNY 6.905 billion, up 165.74% from the end of the previous year; prepaid expenses hit CNY 9.461 billion, up 1,427.10%; short-term borrowings stood at CNY 14.111 billion, up 582.88%.
The company explained that these changes were related to rising storage product prices, expanded business scale, and increased funding needs for procurement.
This is a typical characteristic of distribution businesses: to seize the upcycle, enterprises must procure early, build inventory, and maintain supply chain operations. While price hikes can drive profit growth, they may also increase inventory costs and capital requirements.
Another set of data from Shannon Chip Technology is noteworthy. In the first three quarters, net cash flow from operating activities reached CNY 5.438 billion, compared to a negative value in the same period last year. Improved operating cash flow has supported business expansion.
Thus, this report reflects both the opportunities brought by the booming storage industry and raises new questions: As product prices, procurement costs, and customer demand change, can the company maintain reasonable inventory turnover and capital efficiency?
Computing power service providers face a different kind of capital pressure.
Xiechuang Data achieved revenue of CNY 19.610 billion in the first three quarters, up 135.39% year-on-year; attributable net profit reached CNY 3.062 billion, up 338.77% year-on-year. In Q3, revenue was CNY 7.087 billion, up 109.27% year-on-year; attributable net profit was CNY 1.224 billion, up 360.33% year-on-year.
The company stated that performance growth mainly came from the expansion of intelligent computing power products and services, driven by the upcycle in the storage industry and increased volume in server remanufacturing. The gradual completion of delivery, customer acceptance, and transition to billing and operation phases for previously deployed computing clusters and comprehensive service projects also boosted service revenue.
However, the characteristic of computing power business is that revenue growth often requires upfront equipment and capital investment.
As of the end of September, Xiechuang Data's total assets reached CNY 53.105 billion, up 123.15% from the end of the previous year; fixed assets increased from CNY 9.283 billion to CNY 23.822 billion, and long-term borrowings reached CNY 11.529 billion.
This does not mean asset expansion itself is a risk. Computing power services require real equipment investment; the key lies in whether equipment is fully utilized, whether projects can continuously generate revenue, and whether operating returns can cover depreciation and financing costs.
If computing power demand continues to grow, prior investments may form the foundation for expanding business scale; if customer demand, project acceptance, or equipment utilization falls short of expectations, high investments will put pressure on operations.
Therefore, evaluating computing power service providers cannot rely solely on revenue and net profit growth rates. It is necessary to further judge how much revenue new assets can generate and how much cash return those revenues ultimately form.
PCB companies face similar issues, though the pressure manifests more in accounts receivable, inventory, and capacity expansion.
Kinwong Electronics' net profit grew 70.42% year-on-year in the first three quarters, but net cash flow from operating activities dropped 20.27% year-on-year. Accounts receivable increased from CNY 4.696 billion at the end of the previous year to CNY 7.296 billion, and inventory rose from CNY 2.516 billion to CNY 4.752 billion.
For manufacturing companies experiencing order growth, changes in these line items are not uncommon. Increased orders require more raw materials and production inputs, and expanded sales may lead to higher accounts receivable. What truly matters is whether these changes are consistently supported by revenue and collections.
When profit growth begins to outpace revenue growth, investors naturally focus on whether profitability has improved; when inventory, accounts receivable, and asset investments rise simultaneously, investors need to confirm whether profit growth has translated into operating cash flow.
This is not to deny high growth, but to understand high growth within the complete chain of enterprise operations.
Epilogue
The Q3 reports of AI hardware have provided a clearer observational framework.
First, demand growth must land on specific products. High-performance PCBs, storage products, memory interconnect chips, and edge-side AI SoCs are all related to the AI industry's development, but their customers, technical requirements, and business models differ. Judging whether a company benefits requires looking at sales volumes, prices, and revenue contributions of specific products.
Second, profit growth needs sustainability. Rising storage product prices can bring obvious performance elasticity, but price cycles may change; high-performance product upgrades can improve profitability but require continuous investment in R&D and manufacturing capabilities. Compared to single-quarter high growth rates, trends in gross margins and non-GAAP net profits help better distinguish whether growth stems from cyclical dividends or enhanced product competitiveness.
Third, cash flow and asset investments determine the cost of growth. Distribution companies need procurement and inventory; manufacturing companies need capacity construction; computing power service providers need equipment investment. Different models can create profits, but capital needs and operational risks vary. Whether enterprises can improve capital turnover efficiency, control financing costs, and enable new assets to form sustained revenue will impact long-term returns.
Finally, performance and valuation must be viewed separately.
A company's rapid performance growth does not automatically mean its stock has equivalent upside potential. Stock prices reflect not only realized profits but also market expectations for future growth. When expectations are high, even if a company continues to grow, the stock price may face pressure if the growth rate falls below the market's previous imagination.
This is why sectors like optical modules, PCBs, storage, and chip design need to be studied separately. They may share the industrial background of growing AI demand, yet possess different profit structures, capital needs, and competitive landscapes.
The Q3 reports neither prove that the AI hardware rally has ended nor confirm that all related companies will sustain high growth. Instead, they reveal another fact: as industrial demand gradually materializes, operational differences between enterprises are becoming more concrete.
Kinwong Electronics' accelerated profit release, Shannon Chip Technology's storage business volume surge, Montage Technology's benefit from memory interconnect product upgrades, SigmaStar Technology's edge-side chip shipment growth, and Xiechuang Data's computing power asset expansion respectively present different growth paths within the AI supply chain.
In the next stage, what is truly worth studying will be whether these growth patterns can continue, and whether enterprises can convert growth into more stable profits and cash flows.