返回文章列表
stocksNovember Releasestsmc-october-revenue-currency
📊

What TSMC’s October Sales Report Can Tell You Before Earnings

TSMC’s October monthly release updates the quarter’s available sales evidence. A currency-aware reading shows the implications of a single monthly number for a quarterly outlook.

iBuidl Research2026-11-1816 min 阅读

TL;DR

TSMC reported October consolidated revenue of about NT$586.89 billion on November 10, up 14.7 percent from September and 59.7 percent from a year earlier. The company had guided to US$52.3–54.3 billion for the fourth quarter. At NT$31.8 per US dollar, a constant-rate illustration leaves an average of roughly NT$538–570 billion for each of November and December to reach that range. The result is a benchmark for the two months still to be reported, subject to currency translation, mix, timing, and any updates to the outlook.

A number goes on a spreadsheet as soon as TSMC publishes it. October sales are in NT dollars. The next tab contains quarterly guidance in US dollars. A hurried reader subtracts the reported month from the guidance and gets a result that can look plausible after some spreadsheet formatting. The analyst who owns the model leaves the result blank. The two numbers have to agree on currency and on the length of the reporting period before the subtraction can mean anything.

TSMC's monthly disclosure is useful precisely because a reader can do this exercise while the quarter is still under way. The November 10 release gives the consolidated October observation, its immediately previous comparison, and its year-over-year comparison in the same currency. It supplies a fresh input without giving an outcome for the rest of the quarter. The analyst can replace one uncertain cell and then ask how much the two unreported months would need to contribute under explicitly stated assumptions. October revenue release

Worksheet tab 01 — Put the three prices of time in their own columns

The first column marks October in the 2026 fourth calendar quarter, even though the release appeared in November. The next assigns the date on which the data became available. A third records whether the line represents actual reported sales or an assumption. If someone later investigates what analysts could know in early November, both the activity month and its publication date will matter. Data saved only by release date can place October sales in a November forecasting cohort by accident.

TSMC reports approximately NT$586.89 billion for October and compares it with September's approximately NT$511.86 billion. Their difference is about NT$75.03 billion. Dividing by the September basis gives roughly 14.7 percent, matching the month-to-month figure. October's reported year-over-year growth is 59.7 percent. That denominator answers a different question about the same month in the preceding year. The two comparisons can both appear alongside the actual observation. TSMC September announcement and monthly consolidated revenue records

When entering the figure, choose one scale and preserve it. The worksheet below uses billions. An issuer table using millions could put 586,890 in a cell that the worksheet expects to read as 586.890. The unit error would overwhelm every scenario. Some source tables report exact figures in millions and the press language rounds them in billions. For a ledger reconciliation use the appropriate precision and document the choice. For a sensitivity intended to show order of magnitude, the rounded press values here are enough.

The next row in the worksheet should contain the date to which the quarterly guide applies. October is the first month of the quarter ending in December. Revenue is a flow earned over the period, so the quarterly measure calls for all three months under a comparable accounting basis. A single day's shipment or a cumulative ten-month statistic cannot directly stand in for the still-pending November and December amounts.

This also prevents a common arithmetic shortcut. An average for the prior twelve reported months can help describe historical levels. It does not create an independent observation in the December quarter. An analyst can use a historical assumption openly and still mark actual observations as actual. The update from the latest monthly data should change one part of a model, with the origin of every other part apparent to a second reader.

Worksheet tab 02 — Put dollars and NT dollars on one basis

In its October 15 quarterly release, TSMC forecast fourth-quarter revenue of $52.3–54.3 billion. The same release gives an exchange-rate assumption of NT$31.8 per US dollar in connection with its margin outlook. This article uses that rate as a constant for a simple bridge between the US-dollar revenue guide and observed monthly local-currency revenue. The computed local-currency levels are illustrations. They have a different disclosure status from the company's stated US-dollar guidance. Third-quarter earnings release and fourth-quarter outlook

At the lower US-dollar endpoint, multiplying 52.3 by 31.8 gives NT$1,663.14 billion. The upper endpoint gives NT$1,726.74 billion. The midpoint US-dollar figure of $53.3 billion converts to NT$1,694.94 billion on the same simplified assumption. Only after those conversions does the comparison with an October amount in NT dollars have compatible units. Use the rate in the direction that follows its stated unit. Here one US dollar corresponds to 31.8 NT dollars.

Subtract the October actual. The remaining total would be NT$1,076.25 billion at the low end and NT$1,139.85 billion at the high end. An equal two-month average would be NT$538.125 billion to NT$569.925 billion. The comparable midpoint average is about NT$554.025 billion. The exercise does not require the two months to be equal. This expression lets the analyst assess how a future November figure changes the December amount still needed.

Constant-rate worksheet itemLower guide endpointMidpointUpper guide endpoint
US dollar revenue in billions52.353.354.3
Converted quarter in NT$ billions at 31.81,663.141,694.941,726.74
Less October, in NT$ billions586.89586.89586.89
November and December total, in NT$ billions1,076.251,108.051,139.85
Equal monthly average, in NT$ billions538.125554.025569.925

October exceeds both ends of the remaining-month average under these assumptions. That puts the initial observation into perspective. Repeating October twice would yield about NT$1,760.67 billion for the quarter, translating to roughly $55.37 billion on the constant-rate example. The implied outcome would exceed the upper guide endpoint in that specific exercise. The difference answers a conditional question about repeated identical months. It does not establish that October's shape will recur.

A good updating rule turns any new monthly figure into a remaining amount. Suppose November were NT$550 billion as a purely fictional input. The implied December requirement would then range from NT$526.25 billion to NT$589.85 billion at the two guide endpoints, with NT$558.05 billion at the midpoint. A lower November result increases the needed December amount by the same difference in NT dollars. The consistency check is easy to repeat once the issuer reports its actual number.

Worksheet tab 03 — Change the exchange-rate assumption without changing the demand story

In a second sheet put the same operational midpoint of US$53.3 billion next to three hypothetical conversion rates. At 31.0 NT dollars per US dollar it requires about NT$1,652.3 billion for the quarter. After reported October, the two-month average would be NT$532.705 billion. At 31.8 the corresponding average is about NT$554.025 billion. At 32.6 it rises to NT$575.345 billion. Holding the target in US dollars constant means a higher number of NT dollars per US dollar requires a greater local-currency amount.

The inverse question is equally useful. Keep the same local-currency quarter total and change only the conversion. A quarter with NT$1,694.94 billion, the first sheet's modeled midpoint, would be about $54.68 billion at 31.0 or about $51.99 billion at 32.6. Both are mathematical sensitivities using the same nominal local-currency total. A real revenue comparison can involve transactions in several currencies, and the economic shape of the quarter may change as rates move. The example isolates the conversion input so the reader can see its direction.

The size of such divergences is visible in the Q3 press release. TSMC reported NT$1,494.24 billion and US$46.72 billion for the same quarter. The return from dividing the first by the second is roughly NT$31.98 per US dollar as an implied ratio of the reported totals. The release also reports 50.9 percent year-over-year growth in NT dollars and 41.2 percent in US dollars. The difference makes it worth retaining a currency label on each percentage. The implied ratio is a summary of amounts already translated and rounded by the company, with no claim that it is a specific daily or transaction-level rate.

When building a forecast, use the company-reported historical US-dollar amounts where the question is historical US-dollar growth. A translation at a single assumed future rate has a defensible place in a sensitivity but should not replace those reported data points. If an analyst wants to compare operational growth without exchange-rate movements, they need to describe a constant-currency method and the source amounts it uses. A plain division between two headlines will give a result but cannot repair an incompatible historical series.

An FX desk and a semiconductor analyst can use the same worksheet for different decisions. The FX desk varies the rate while holding forecast local sales steady and observes how a translated total responds. The semiconductor analyst examines local sales and supply assumptions while holding one rate steady for the comparison. An actual forecast can change both. Reporting independent cases first makes the combined scenario understandable. It also helps prevent an exchange-rate movement from being misdescribed as higher wafer demand.

Worksheet tab 04 — Ask which operational stories can fit the monthly total

One month of revenue can reflect several changes at once. Delivered quantity, technology mix, packaging and other sold services, realized pricing, and the timing of recognized revenue can contribute differently. October's consolidated number is the result across the covered business for that period. The monthly release discloses a sales amount, and the quarterly release offers additional breakdowns and management explanations. A reader can use the October data without making it answer a question it could not identify.

For example, TSMC disclosed in October that 2-nanometer shipments made up 8 percent of third-quarter wafer revenue and expected a strong ramp in the fourth quarter. The denominator in the disclosed percentage is wafer revenue for the prior quarter. October's total is consolidated monthly revenue. Multiplying the latter by eight percent would use a share from one period and scope on a different period and population. The reported node-mix share supplies context for demand and pricing, with no basis for turning it into an October-only estimate without more data. TSMC Q3 earnings materials

A simple fictional manufacturer can show why. It sells eighty units of a standard product at one unit of currency each and twenty advanced units at three each, for total product revenue of 140. One month later it sells the same total physical quantity, but the mix is seventy standard and thirty advanced. Product revenue becomes 160 even while unit volume is flat. Add a separately billed service and consolidated revenue can rise again with unchanged product shipment count. This example illustrates the information a revenue headline leaves open, with no assumption about TSMC's actual price or volume.

There are dates within a month that matter too. One large customer shipment can become recognized before month-end or during the next month under the applicable terms. The resulting timing difference may largely cancel within a quarter. It can still create a sharp month-to-month comparison. Before extrapolating a monthly change, an analyst can look at the prior monthly sequence and management's description of seasonal effects or ramps, then mark what evidence would distinguish a shift of recognition from persistent acceleration.

Worksheet tab 05 — Use scenarios built from real questions

An analyst working before October reports might assign one expectation for each month. Once the amount arrives, there are two legitimate choices. Keep the independent assumptions for the remaining months and let the modeled quarter change, or keep the stated quarterly target and let the worksheet calculate the implied remaining amount. The first produces a forecast or sensitivity based on a view of the unreleased months. The second produces a hurdle required to land on the chosen guide. Mixing the two without labels lets a target masquerade as independent evidence.

Imagine a worksheet that initially assumes NT$550 billion in each of the three months. Its modeled quarter is NT$1,650 billion. Replacing October alone with NT$586.89 billion lifts the modeled quarter to NT$1,686.89 billion. At 31.8 this yields about US$53.05 billion. The two unchanged monthly figures remain hypotheses. A future forecast can revise them because a separate supplier disclosure, customer timing, or calibrated seasonal model changes the analyst's view. The one new reported number does not directly establish those changes.

The target-driven version begins with the US$53.3 billion midpoint and on the same conversion assumption requires the remaining average previously shown. If November is assumed to be NT$520 billion for a timing scenario, the December figure needed to reach that midpoint rises to NT$588.05 billion. If November were NT$580 billion, December would need NT$528.05 billion. These assumptions can be discussed as examples of early and late delivery in a quarter. They do not tell a reader which distribution of sales TSMC will report.

A neutral sensitivity sheet can assign no probability. It can state how far each row lies from the guide. With two assumed months at NT$530 billion apiece, the local total would be NT$1,646.89 billion, or about $51.79 billion at 31.8, below the lower threshold in the illustration. With two NT$570 billion months, the local total would be NT$1,726.89 billion, giving roughly $54.30 billion and almost exactly the upper bound. Rounding the published month and scenario values makes a few million dollars around that edge uninformative.

Compare with a prior shape before repeating October

The company reported NT$367.473 billion in October 2025, NT$343.614 billion in November, and NT$335.003 billion in December. Their sum is NT$1,046.090 billion using the published million-level values. October was the largest of those three months. That observed pattern is one possible comparison with the exercise that repeats October 2026 three times. A single prior year supplies a useful example of dispersion within one quarter while giving little evidence by itself for a stable seasonal rule. October 2025 release, November 2025 revenue report, and December 2025 release

One purely mechanical comparison scales each previous-year month by the same factor as the latest October observation. Dividing 586.89 by 367.473 gives about 1.5971. Applying that multiplier to the prior November gives NT$548.78 billion and to prior December gives NT$535.03 billion, with values rounded to two decimal places in billions. October plus those two illustrative months gives approximately NT$1,670.71 billion. At 31.8 per US dollar this equals roughly $52.54 billion, near the lower portion of the guidance range.

The mechanical result differs materially from the flat-October scenario of $55.37 billion. Both use the same latest observation and conversion. Their difference comes only from the assumed shape of November and December. A reader can see how strongly a forecast depends on its time profile instead of treating two published monthly scenarios as studying different levels of October sales. The prior shape could be wrong here because capacity, mix, customer demand, and invoicing have changed. It serves as a sensitivity supported by an inspectable historical series.

An operations team could use the prior sequence as a reference and then set a test for its own forecasts. If the November observation comes in above that scaled path, the arithmetic quarter estimate can absorb the difference immediately. Updating the December assumption requires identifying a connection, such as invoices brought forward from December or new incremental orders expected to carry through. The first reassessment follows from a number. The second needs a claim about what the number represents.

A separate chart might compare the proportion of the quarter represented by October. With the company's reported October figure and the simplified low-quarter total of NT$1,663.14 billion, the proportion is about 35.3 percent. Against the high-quarter total of NT$1,726.74 billion it is about 34.0 percent. The prior October made up about 35.1 percent of its quarter's summed monthly series. These proportions suggest a well-understood decomposition of the scenarios. They have no independent predictive power just because they look close.

Reconstruct the reporting base behind the growth rates

The October bulletin also reports approximately NT$4,485.62 billion for January through October, up 43.3 percent from the same period a year earlier. The October 2025 release put that prior ten-month sum at about NT$3,130.437 billion in its data table. These cumulative numbers tell a broader story than a single month. Someone reading the growth headlines can reconstruct why a month growing almost sixty percent pushes the ten-month rate upward by only a few percentage points.

Subtract current October from the current ten-month amount to obtain roughly NT$3,898.73 billion for January through September. The same operation on the prior values gives NT$2,762.964 billion. The implied increase for the nine months is about 41.1 percent on those inputs. October's prior base amounts to roughly twelve percent of the prior ten-month sum. Growth from the other nine months carries most of the weight. A simple average of 41.1 and 59.7 would overstate October's weight and would yield a result inconsistent with the published period.

The exercise shows how to compare a cumulative amount and an individual update without suggesting that October changed the earlier nine months. To build a chart for the December quarter, keep the year-to-date tab as context about the base and track October through December separately. A reader who asks what is required to reach year-end guidance can follow its relevant period. A reader who asks about one month's momentum sees the comparison denominators.

Worksheet tab 06 — Leave space for revision and reconciliation

The calendar on TSMC's investor site says the next monthly sales release is scheduled for December 10 at the published Taipei time. An analyst can make a note to replace the November placeholder when the actual appears. The original forecast remains available as a dated view and the new version can show the variance due to each revised input. This lets readers evaluate how the forecast developed without confusing the improvement from newer data with a better model. TSMC financial calendar

A data table can also receive a small correction after initial publication. Store the values with publication time and precision, and preserve the old calculation if it informed a public claim. When the consolidated quarter is available, use the company's actual statements and reconcile against the sum of recorded monthly sales on the applicable basis. Differences can reflect rounding, revisions, or other presentation details depending on the underlying reports. A simple check can identify a mismatch without supplying a cause on its own.

A narrow uncertainty check asks how large a rate error would matter near the boundary. Start from the modeled midpoint total in NT dollars. Increasing the conversion denominator by one percent at unchanged local revenue divides the US-dollar result by 1.01, taking it from $53.3 billion to about $52.77 billion. An assumption this small can move a closing estimate by roughly half a billion dollars. A model that compares reported sales within a few million of a translated threshold should state the much larger practical sensitivity to currency and rounding beside that comparison.

Finally give the user of the worksheet one sentence they can attach to a chart. October provides NT$586.89 billion of directly observed consolidated revenue for the quarter. The quarterly guide is expressed in US dollars. November and December remain assumptions in any calculation made from today's inputs. A stated exchange-rate convention makes those numbers comparable for the exercise. This sentence can stay beside the chart even when future releases replace every placeholder.

The value of the new observation is a smaller unknown range. What remains to reach a guide is a calculation. Whether the company will reach it is a separate empirical question. The analyst can answer the former now, explain how the currency assumption moves it, and leave the latter to the remaining monthly reports and the consolidated quarter.

Sources

更多文章