The setup
Winbond's third quarter was all price, and our call is that the fourth-quarter contract step lands: October revenue, filed by , at least 10% above September's NT$28.25B, that is NT$31.07B, and October–December revenue of at least NT$93.22B. October below September refutes it. Revenue rose 37.6% to NT$82.33B while CMS bit shipments fell by a high-single-digit percentage and Flash bits were flat. The memory the fabs made at about 100% utilization but did not ship sat on the balance sheet as a NT$3.60B inventory build.
Winbond Electronics (華邦電, 2344.TW) reported on , at its 15:00 Taipei investor conference. Consolidated revenue was NT$82.33B (~US$2.57B at ~31.99 NT$/US$, the deck's third-quarter average, approximate), up 37.6% on the quarter and 278.2% on the year. Gross profit was NT$58.10B (~US$1.82B), a 70.6% margin, against 66.2% in the second quarter and 46.7% a year ago. Operating profit was NT$43.78B (~US$1.37B), a 53.2% margin. Net income attributable to the parent was NT$35.42B (~US$1.11B), up 45.7% on the quarter, and EPS was NT$7.58 (~US$0.24). Nine-month EPS is NT$15.32.
Put the quarter against the company's own history before reading anything else. The deck's corporate-value slide lists annual EPS of NT$3.41 for 2021, NT$3.23 for 2022, a loss of NT$0.29 for 2023, NT$0.14 for 2024 and NT$0.88 for 2025. One quarter earned 2.22 times the best full year in that series. Nine-month EBITDA of NT$101.55B is 3.38 times the NT$30.09B of 2021. This is not a good year for Winbond. It is another company's income statement on Winbond's letterhead. The question the print raises is how much of it is the plant and how much is the price.
The answer in the deck is unusually plain. Revenue in the memory business, 90.2% of the company, rose on price alone. The Customized Memory Solution line (CMS, the DRAM business under its new name) grew 50% on the quarter with blended ASP up "approx. sixties%" in US dollars and bits shipped down. Flash grew 34.0% with ASP up "low-thirties%" and bits flat. In August we read the second quarter as a supplier reserving capacity rather than selling into the shortage; the third quarter is the same decision, made again at a higher price, and this time the bits it held back show up on the balance sheet.
The numbers
Winbond Electronics Q3 2026 · consolidated · NT$ billions unless stated
| Line | Q3 26 | Y/Y | ||
|---|---|---|---|---|
| Sales | ||||
| Revenue | 82.33 | |||
| Memory revenue | 74.28 | |||
| Logic IC revenue | 7.59 | |||
| Gross margin | 70.6% | |||
| Memory gross margin | 73.9% | |||
| Profit | ||||
| Operating profit | 43.78 | |||
| Operating margin | 53.2% | |||
| Net income to parent | 35.42 | |||
| EPS (NT$) | 7.58 | |||
| Cash and balance sheet | ||||
| Operating cash flow | 45.87 | |||
| Capex | 15.95 | |||
| Depreciation & amortization | 7.60 | |||
| Inventory | 29.38 | |||
| Cash & hedging assets | 84.86 | |||
| Equity | 225.54 | |||
Source: Winbond 3Q26 and 2Q26 investor conference decks (consolidated, unaudited); monthly revenue from TWSE MOPS. Notes ↓
One piece of context before the argument. The deck reports in New Taiwan dollars and the ASP changes in US dollars; the exchange rate moved 1.3% on the quarter (31.58 to 31.99 NT$/US$), so about a point of the revenue growth is currency and the rest is the contract price.
The deck's own description of the quarter is two lines on the memory-results slide. For CMS: revenue "+50%", bit shipment "Decreased high-single digit%", blended ASP (US$) "Increased approx. sixties%". For Flash: revenue "+34.0%", bit shipment "Flat", blended ASP "Increased low-thirties%". Decode: none of the quarter's growth came from selling more memory. Price carried all of it, and in the larger line price had to make up for fewer bits as well. A supplier at about 100% utilization that ships fewer bits is not short of demand. It is choosing where the bits go, and when.
The rest of this dive is for paid subscribers.
The headline numbers are above. The 12 sections below carry the mechanism, the peer read, the valuation work, and the dated tests that decide it.
- Nine quarters: the level before the delta
- All price, fewer bits
- The bits it made and did not ship
- The depreciation bill
- The bond, the buy-in and the cash
- Connecting the dots
- The December bar
- Valuation
- The strongest case against this read
- What we'd watch
- Grading the August calls
- Bottom line
Not ready to pay? Get the free Monday issue and one full dive a week by email.
Already a subscriber? Sign in.