A beat on every line it guides
Start with what isn't in dispute. In May, WT guided Q2 revenue at NT$560B to NT$590B (about US$17.39B to US$18.32B) and printed NT$590.69B (about US$18.34B), up 127.6% on the year and 19.5% on the quarter. Operating profit of NT$12.99B (about US$403M) cleared a NT$11.20B to NT$12.39B range (about US$348M to US$385M). Net income to the parent, NT$9.71B (about US$302M), beat the NT$8.57B (about US$266M) top of its guide by 13.3%, and basic EPS of NT$7.68 (about US$0.24) is the highest WT has printed, against NT$2.28 a year earlier. First-half revenue passed NT$1.08T (about US$33.69B).
The acquisition doesn't flatter the comparison. WT closed its US$3.8B purchase of Future Electronics on , so Future sits inside both Q2 2025 and Q2 2026 for the whole quarter, and the 127.6% is growth on a like-for-like footprint. The filings don't split Future out. Management said Future's European and US business grew about 40% (unverified — third-party summary of the call).
On the call, management put it plainly: "The demand we are seeing is very, very strong, and next year's growth will also be very positive" (unverified — translated third-party summary of the call). The filing backs the first half of that sentence. What the quote leaves out is who paid to carry the demand, and for a distributor at a 3.41% gross margin that is the only question the income statement can't answer.
A thinner margin on a much bigger base
Gross margin fell to 3.41% from 4.32% a year ago and 3.45% in Q1. The reason is mix. Data center and server was 59.2% of revenue against 36.0% a year earlier and 56.9% in Q1, up 274% year over year on the investor-conference deck. On the deck's rounded percentages that is roughly NT$349.69B (about US$10.86B) of data-center revenue against about NT$93.42B (about US$2.90B) a year ago. Everything else, about NT$241.00B (about US$7.48B), grew roughly 45%: slower, and still fast.
The margin that lost was gross; the one that won was operating. Operating expenses rose only 9%, to NT$7.15B (about US$222M) from NT$6.56B (about US$204M), so the expense ratio fell to 1.21% from 2.53% and operating margin widened to 2.20% from 1.80%. A data-center dollar carries less gross profit and almost no extra overhead. That trade works as long as the dollar doesn't also need a balance sheet.
The rest of this dive is for paid subscribers.
The headline numbers are above. The 9 sections below carry the mechanism, the peer read, the valuation work, and the dated tests that decide it.
- Who paid for the build
- An 18-day cycle, and what it nets
- Two years of the same level
- Two months into Q3
- WPG took the other road
- The strongest case against me
- What decides it
- The trade (analysis, not advice)
- Bottom line
Already a subscriber? Sign in.
Who paid for the build
Between December and June, receivables rose NT$179.09B (about US$5.56B) and inventory NT$58.84B (about US$1.83B): NT$237.94B of new working capital in six months, taking receivables plus inventory to NT$667.71B (about US$20.74B). For a distributor growing this fast that number normally turns up as borrowing. It didn't. Accounts payable rose NT$239.81B over the same six months, to NT$557.09B (about US$17.30B), more than the whole build. Payables are now larger than receivables of NT$349.11B (about US$10.84B), and 69% of total assets. Net working capital finished June at NT$110.62B (about US$3.44B), below March and below .
The cash statement says the same thing from the other side. First-half pre-tax profit was NT$21.25B (about US$660M) and operating cash flow NT$20.73B (about US$644M), against NT$13.03B (about US$405M) a year earlier. Receivables and inventory absorbed NT$230.27B (about US$7.15B) of cash; payables supplied NT$233.55B (about US$7.25B). Q2 alone generated NT$29.17B (about US$906M) after a negative NT$8.44B (about US$262M) in Q1. Net debt fell to NT$27.63B from NT$55.66B (about US$1.73B) in March, and total borrowings, including the convertible bond, to NT$64.78B (about US$2.01B) from NT$94.14B (about US$2.92B).
The deck states the result in one line: "Net gearing ratio 22% 49% 67%," for June, March and a year ago. Gearing fell by more than half in a quarter when revenue grew 19.5%. A distributor normally levers up into a boom. WT levered down, because its suppliers extended it the terms a lender would otherwise have had to. It is the library's vendor becomes the credit model read from the other end: here WT is the customer being financed.
The prepayment lines make the mechanism visible. Contract liabilities, cash customers paid WT before delivery, were NT$35.95B (about US$1.12B) in June against NT$0.78B (about US$24M) a year earlier. Prepayments, cash WT paid its own suppliers ahead of delivery, were NT$32.35B (about US$1.00B) against NT$1.39B (about US$43M). That's what an allocation market looks like on a distributor's balance sheet: customers pay to secure supply, WT passes the cash upstream and keeps a NT$3.60B (about US$112M) difference. WT isn't funding the allocation. Its customers are.
An 18-day cycle, and what it nets
The deck's headline efficiency number is the operating cash cycle: 18 days, down from 21 in Q1 and 36 a year ago. It's a real improvement, and the series underneath it says where it came from. Receivable days rose to 48 from 41 in one quarter. Inventory days fell to 47 from 50. Payable days went to 77 from 70. All of the Q2 improvement came from paying suppliers later, while customers paid WT more slowly. The composite hides the component here exactly as it did at HPE: a cycle that shortens by three days can carry a customer book that lengthened by seven.
Payable days have been 70 or higher for six straight quarters, 77, 78, 72, 83, 70 and 77 from Q1 2025, so the supplier terms aren't a one-quarter gift. What moved in Q2 is receivables: 48 days is the highest since Q2 2025's 49, and receivables grew 168.2% year over year against revenue growth of 127.6%. The customers buying data-center parts are taking longer to pay, and for now the suppliers are giving WT more time than that.
Two years of the same level
One quarter of falling working capital would prove nothing; March had gone the other way. The longer series is what makes this a finding. Net working capital has stayed between NT$96.96B (about US$3.01B) and NT$127.97B for nine straight quarter-ends while quarterly revenue went from NT$243.65B (about US$7.57B) in the quarter to NT$590.69B. As a share of annualized sales it fell from 12.24% in to 9.34% a year ago, 6.47% in March and 4.68% in June. WT has become a business where an extra dollar of revenue needs almost no extra working capital, because the supplier side scales with it. That is the opposite of the pattern I named “the ODM pays three ways” at TD SYNNEX, where trade working capital left its two-year band and kept climbing for three straight quarters.
Two months into Q3
The Q3 guide opens with a line worth reading slowly: "At an exchange rate assumption of 32.2 NT Dollars to 1 US Dollar." Q2's guide was set at 31.6. The weaker NT dollar adds about 1.9% to NT$ revenue on its own, so a midpoint of NT$592B (about US$18.39B), flat on Q2's NT$590.69B, is a slight decline in dollars. The range runs from NT$572B (about US$17.76B) to NT$612B (about US$19.01B); operating profit is guided at NT$13.27B to NT$14.81B (about US$412M to US$460M) and EPS at NT$7.55 to NT$8.52, midpoint NT$8.03 (about US$0.25). Profit grows on flat revenue because gross margin is guided up to 3.54% to 3.64%.
Then the monthly numbers. July revenue was NT$182.97B (about US$5.68B) and August NT$198.38B (about US$6.16B), together NT$381.35B (about US$11.84B). At the two-month average pace Q3 lands at NT$572.0B, the bottom of the range to the decimal. The midpoint needs September at NT$210.65B (about US$6.54B), 6.2% above August and essentially April's record NT$211.02B (about US$6.55B). The bottom needs only NT$190.65B (about US$5.92B), less than August. Seasonality helps: last September was NT$134.95B (about US$4.19B), 34.8% above , and the midpoint needs this September up 56.1% on the year, against 94.85% in July and 98.21% in August.
The company's own monthly explanation shifted, too. From April through July WT attributed growth to "demand growth in data center, communications and industrial applications" (translated from the monthly revenue filing). August's note reads "year-over-year growth in communications, industrial, automotive and data center applications" (translated). I wouldn't hang much on word order. But it's the first month since at least April that the explanation didn't lead with data center, and the first time automotive appears. Management described non-AI markets as in "sustained robust recovery" on the call (unverified — translated press report). If so, the non-data-center half is doing more of the lifting, which helps the margin guide. My lean: September lands inside the range and below the midpoint.
WPG took the other road
WPG Holdings, the other Taiwan distributor at this scale, shows what the same boom costs without the supplier float. Q2 revenue was NT$456.21B (about US$14.17B), up 82.2%, at a higher gross margin of 4.43% and a higher operating margin of 2.62%. But its receivables and inventory ran well ahead of its payables. On the same definition, receivables plus inventory minus payables, WPG's net working capital was NT$295.18B (about US$9.17B) in June, and its first-half operating cash flow was negative NT$98.90B (about US$3.07B). Net debt was about NT$194.21B (about US$6.03B). WPG earns more per dollar of sales and borrows to carry it; WT earns less and doesn't. On EPS they look like peers. On cash they are opposites.
WPG's July and August revenue came to NT$288.33B (about US$8.95B), and its June had run 39.77% above May, so it is used to a heavy last month. Its Q3 guide of NT$460B to NT$500B (about US$14.29B to US$15.53B) needs September at NT$171.67B (about US$5.33B) for the low end, 20.7% above August (unverified — guide range from a press report, not the filing). WT needs less of a jump to hit its range than WPG does.
The strongest case against me
The best bear argument is that I'm reading a lease as if it were a deed. WT doesn't control its payable days; its suppliers do, and in a market where customers are prepaying NT$35.95B for allocation, the suppliers can change terms when they choose. Each 10 days of payables is worth about NT$62.53B (about US$1.94B) at Q2's cost of sales, more than twice June's net debt. Receivable days already rose by seven in one quarter. If payable days went back to the 62 to 66 of 2024 while receivable days held at 48, net working capital would jump well past the March high and gearing would head back toward the 67% of a year ago. WPG's balance sheet is what the normal version looks like. On this reading the NT$27.63B of net debt is a snapshot of generous terms at the top of an allocation cycle, and the prepayment pair is evidence the cycle is at its tightest.
That names the right risk, and I take it seriously. I still come down on the other side for the next print, for two reasons. Payable days have held at 70 or more for six straight quarters, through a period when revenue more than doubled, so they are a feature of WT's supplier relationships rather than a Q2 gift. And the Q3 guide is flat in NT dollars, the easiest possible quarter for working capital: there is no growth to fund. The bear case is a 2027 risk. It isn't the Q3 base case.
What decides it
One number on one date: net working capital, receivables plus inventory minus payables, on the balance sheet, in the Q3 statements due by . At or below NT$127.97B, the March high, the supplier float held through a NT$592B-scale quarter and I'm right. Above it, the float has started to turn, the growth is back on borrowed money, and I'm wrong. Two earlier checks: September revenue, due by , against NT$210.65B for the midpoint and NT$190.65B for the bottom; and receivable days against Q2's 48. A second quarter of rising receivable days with falling payable days would be the first sign the bear case has started.
The trade (analysis, not advice)
The distributor trade on AI usually means buying the revenue and accepting the balance sheet that comes with it. WT is the rare case where the balance sheet improved as revenue doubled: net gearing 22% against 67% a year ago, net leverage 0.8x against 2.8x, and a 32.8% return on equity on the deck. Anyone who wants distribution exposure to the data-center build gets less funding risk in WT than in WPG, which put NT$98.90B into working capital in six months. The margin is thinner, 2.20% operating against WPG's 2.62%, and the Q3 guide lifts it only to 2.32% to 2.42%. It's the inverse of demand-long, balance-sheet-short: here the balance sheet is the strength, and it is on loan from the supply chain. What you are underwriting is not the demand, which is plain, but the supplier terms. I would want the revenue print and the November balance sheet before paying for the record EPS.
Bottom line
WT printed a record: NT$590.69B of revenue, NT$7.68 of EPS, data center at 59.2% of sales. The more interesting number is the one it didn't need, new borrowing. Suppliers added NT$239.81B of payables against a NT$237.94B build in receivables and inventory, operating cash flow was positive, and net debt halved. My view is that the float holds through Q3. The number that decides it is net working capital at or below NT$127.97B on .