Recently, GigaDevice Semiconductor Inc. (603986.SH) released its 2026 third-quarter report. For the first three quarters, revenue reached 20.297 billion yuan, up 197% year-on-year, while net profit attributable to shareholders hit 11.066 billion yuan, up 921% year-on-year.
For comparison: in all of 2025, the company's net profit was 1.648 billion yuan, meaning this year's first three quarters alone were 6.7 times that figure. Yet the share price has moved in the opposite direction. From a June peak of 846.66 yuan per share, it closed at 257.8 yuan on October 30, a drop of nearly 70% from the high.
Profit rose more than ninefold, but the share price retained only about 30% of its value. Placed side by side, these two curves look like a paradox. But financial statements answer "how much was earned," while the stock price answers "how long can this money keep being earned, and how was it earned." To address that question, one must take the income statement apart and also read the story behind the share price.
The core business is climbing while the books give back
Start with the nature of the growth. First-three-quarter revenue rose 197.11% year-on-year, roughly triple the year-earlier level; net profit attributable to shareholders rose 921.55% year-on-year, roughly ten times the year-earlier level. This scissors gap shows that the increment came mainly from "selling at higher prices," not just "selling more." The company's own wording is "rising volume and price," but judging by the degree of profit amplification, price contributed far more. Costs rose more slowly than selling prices, and the higher prices fell almost entirely into profit, which is the most typical operating leverage in a price-upcycle.
Now look at the quality of profit. First-three-quarter net operating cash flow was 10.124 billion yuan, up 463.81% year-on-year, about 90% of net profit attributable to shareholders. Breaking down the structure further: of the 11.066 billion yuan in net profit attributable to shareholders, the portion after deducting non-recurring gains and losses was 9.434 billion yuan, with about 85% coming from the core business; the remaining 1.632 billion yuan was non-recurring gains and losses, mainly from fair value changes on securities investments. Judging by cash collection and profit structure, this is a fairly solid set of accounts.
The real details are hidden in the third quarter. Single-quarter revenue was 8.732 billion yuan, up 225.65% year-on-year and about 18% quarter-on-quarter; net profit attributable to shareholders was 4.209 billion yuan, up 728.93% year-on-year but about 22% lower quarter-on-quarter, versus 5.396 billion yuan in the second quarter. Looking only at that line, it is easy to conclude that growth has peaked. But from another angle, the conclusion changes.
Third-quarter gross margin was about 70.3%, up another 3.8 percentage points from the second quarter, while the first and second quarters were 57.1% and 66.6%, respectively, rising through all three quarters. The quarter-on-quarter decline in net profit came from non-recurring gains and losses: more than 1.9 billion yuan in the second quarter turned into a negative of more than 300 million yuan in the third quarter. Working backward, fair value change gains for the first three quarters were 1.878 billion yuan, while the first half alone was already 2.228 billion yuan, meaning this item gave back about 350 million yuan in the third quarter. Excluding this book volatility, the core business did not deteriorate. Third-quarter operating cash flow was 4.076 billion yuan, about 97% of net profit attributable to shareholders, and the money earned was genuinely collected.
So the more accurate statement is this: the core business is still climbing, while book investment gains are giving back.
A 70% gross margin: market conditions or a moat?
A good set of accounts also needs to be read calmly. Third-quarter revenue growth slowed to 18% quarter-on-quarter, and the profit side was similar: the quarter-on-quarter growth rate of non-GAAP net profit fell from a high in the second quarter to 31% in the third quarter. Slowing growth after a higher base is common sense, but the capital market never trades the water level itself, only the slope of the water level's rise.
The deeper question is: who gave the 70% gross margin? Was it a technological barrier or a supply-demand gap? The memory industry has a simple rule: price increases are the best advertisement for capacity expansion. High prices encourage peers to add capacity while suppressing downstream demand, and once new capacity is gradually released, today's shortage may become tomorrow's oversupply.
Moreover, the slope of prices is already changing. According to TrendForce forecasts, the quarter-on-quarter increase in conventional DRAM contract prices narrowed from 58% to 63% in the second quarter to 13% to 18% in the third quarter. Prices are still rising, but the steepest part may already be over. That is why the source of gross margin matters more than the number itself. If it comes mainly from a supply-demand gap, then it is more like a gift from the cycle than a card the company accumulated on its own. At that point, operating leverage becomes a double-edged knife: it is a booster on the way up and an amplifier on the way down.
The curtain falls on a "shadow"
Back to the opening question: why isn't the share price buying it? The answer is not entirely in the financial statements, but also in the "story." For more than a year, funds treated GigaDevice as a "shadow stock" of ChangXin Memory Technologies. Founder Zhu Yiming is also chairman of ChangXin Memory Technologies, and GigaDevice directly held about 1.80% before ChangXin's listing, diluted to about 1.6% after the listing. The appeal of a shadow stock is that it lets people trade the imagination of the main entity in advance.
On July 27, ChangXin Memory Technologies listed, closing up 465.82% on its first day with a total market value of 3.28 trillion yuan, topping the A-share market. Once the shadow had the entity itself, the premium lost its reason to exist: that day GigaDevice closed down 5.03%, and by July 29 it had fallen more than 20% over three trading days. Combined with a late-July Morgan Stanley report warning that memory contract prices might peak in the fourth quarter, the market's pricing logic for the company began to shift.
Afterward, Chairman Zhu Yiming proposed a share buyback of 1 billion to 2 billion yuan and planned to increase his stake by no less than 1 billion yuan. This was a show of confidence, but it should also be viewed objectively. The increase plan will not begin until December 13, while from May to June, Zhu Yiming had just reduced his holdings by about 11.11 million shares, cashing out about 4.4 billion yuan. By late October, the company's buyback had already used about 1.1 billion yuan, with the highest buyback price at 411.95 yuan, also above the current share price. Confidence can support sentiment, but it cannot support the cycle.
Conclusion
Zooming out, GigaDevice's long-term opportunities are not lacking. Some views hold that the memory supply shortage may last until 2028, making the boom window longer than in the past; AI is moving from the cloud to endpoints, and the dual demand from phones, cars, and industrial equipment for memory and MCUs is exactly the battlefield this company knows best; coordination with ChangXin is also deepening, with related-party transaction quotas rising sharply this year, upstream supply more assured, and the fair value of its ChangXin stake reaching 10.062 billion yuan by the end of June, a substantial strategic reserve.
But the other side of the coin is equally clear. The fabless model has no heavy-asset depreciation burden and is highly elastic when prices rise, but that also means capacity bargaining power is not in its own hands. The company's products are concentrated in NOR Flash, SLC NAND, and niche DRAM, and it has not entered core AI computing categories such as HBM, DDR5, and enterprise SSDs, so it benefits from peripheral demand rather than the main battlefield. In addition, as investment income takes a larger share, profit volatility is amplified, making valuation naturally harder to pin down.
So the safer way to read this third-quarter report is to split profit into two halves: one half is non-GAAP, which reflects the temperature of the core business; the other half is fair value, which reflects the weather of investments. Next there is only one exam question: when the price-increase dividend recedes, what is left for GigaDevice? When contract prices peak, how long gross margin can hold, and whether MCUs and self-developed DRAM can take over are all sub-questions of that exam. While the tide is still rising, the most important question is not how fast you are swimming, but whether you are swimming or being pushed along by the tide. (This article is only personal analysis and does not constitute any investment advice. Investing involves risk, and caution is required when entering the market.)
Comments