COST Technical Appendix: Method, schedules, bridge and sensitivities

The COST research family (page counts are those of the approved PDF editions, which this site does not host):

• Deep Dive: A very good business, priced for a better one (58 pages). Public preview; the Deep Dive page says how to read the full report.
• Research Note: Costco at the $1,040.55 valuation close: the franchise is not in doubt, the price is (7 pages).
• Technical Appendix: Method, schedules, bridge and sensitivities (29 pages).
• Learning Companion: How to read the Costco research (12 pages).

---

The model behind the Costco Deep Dive and its Blockhaven Rating of 1 / 5, POOR, at the $1,040.55 close on 11-02-2026: the eleven-year record the forecast extends, how the value is built, the annual schedules of all three scenarios, the equity bridges, the terminal, the discount rate, what the price fits, the sensitivities and the nine challenged assumptions, as the Deep Dive states them.

How we valued it

Company-level dollar amounts print in billions of U.S. dollars, or in millions where an amount is smaller than a billion, and the unit is always written; per-share values, prices and dividends are in dollars. Tables name their unit in the heading or the row label.

We built the value from two revenue lines rather than one. Membership fees and net sales each grow at their own rate, fading in a straight line over ten years (FY2027 to FY2036). Operating income is fees plus the merchandise margin times net sales. We tax it at 25%, add depreciation, subtract capital spending and the growth in working capital, and discount the result at one rate for all three scenarios. Costco's working capital works in its favor: suppliers fund more than all of its inventory, and members pay fees in advance, so growth releases cash instead of absorbing it. After year ten we use a value-driver terminal: the terminal value is next year's after-tax operating profit times one minus growth over return on new capital, divided by the discount rate less growth.

Four method choices matter, and each follows the evidence. FY2027 capital spending is management's stated intention of about $7.5 billion (a plan, not a result), which is well above what our capital-intensity rule alone would imply and takes net sales per dollar of net PP&E down to 7.98 from 8.34. Other operating liabilities (accrued salaries, member rewards and other current liabilities, less receivables and other current assets) fund the business too, at about 3.3% of net sales, and we count that release of cash. Unsettled card receivables of $2.84 billion sit inside reported cash and are operating working capital rather than surplus cash, so they come out of the bridge. Cash flows are discounted at mid-year, because they arrive through the year. The paragraph after the sensitivity table shows the effect of each one.

The balance sheet is dated 08-30-2026 and the price is dated 11-02-2026, so we carry our value forward 64 days at the discount rate and deduct the $1.47 dividend that went ex on the 10-30-2026 record date (our reading of the settlement convention, not a statement in the filing). Costco has filed no balance sheet since 08-30-2026, so cash, debt and share count at the price date are unknown. We do not estimate them. Share repurchases at or near the price would not change value per share by a visible amount in this window.

Since the valuation date

Every value in this report is dated 11-02-2026, the session whose close we rated, and none has been rolled forward. Two later settled closes are in our record. COST closed at $1,053.56 on 11-06-2026 and at $1,066.18 on 11-09-2026, the November 9, 2026 regular-session closing price, which is $25.63 or 2.5% above the $1,040.55 valuation close and $12.62 or 1.2% above the 11-06-2026 close. The 11-09-2026 close is a separate market snapshot, taken after that session settled, and it is the only evidence added after 11-07-2026. The 10-year Treasury par yield ended 11-06-2026 at 4.86%, five basis points below the 4.91% we built the discount rate from. We have no Treasury reading for 11-09-2026, so the yield comparison stops at 11-06-2026.

Two things moved and neither changes the reading. The price rose, so every scenario now sits slightly further below it, and the upside at the most favorable end of the rate band, $741.88, is 30.4% below the 11-09-2026 close. The yield fell, and we have not rebuilt the rate. The stored sensitivities give the scale: on the reference, each percentage point of discount rate is worth about $55.03 a share on the high side of the band and $86.45 on the low side, so a five basis point move is worth about $2.75 to $4.32 a share if all of it passed into the rate, against a $637.60 gap between the reference and the 11-09-2026 close. That is a size for the move and not a new valuation.

What did not move is the evidence the values rest on. In the SEC filing index we read on 11-07-2026, Costco had filed no periodic report since the Form 10-K of 10-07-2026 and no event filing since the Form 8-K of 10-16-2026, and the only later filings were insider transaction reports; we did not read the index again for the 11-09-2026 close. There is no new balance sheet, and the share count and our assumptions are as they were. The one new company document since the valuation is the October sales release, read below. A price about 2.5% higher and a yield five basis points lower leave the gap between price and value where it was, and they give no reason to argue for a different level.

What the 11-09-2026 close fits

The break-even readings in this report were solved at the 11-02-2026 close. We solved them again, with the same model routines and the same other drivers, at the November 9, 2026 regular-session closing price of $1,066.18, and the table sets the two side by side. Each is a conditional combination that the price fits and not a statement of what investors believe or what the price requires, and neither is a new valuation: the values of 11-02-2026 are unchanged.

Every break-even moved the way a higher price says it should, and by a small amount: the reference matches the close at a rate of 5.19% instead of 5.24%, which is still below the 20-year and 30-year Treasury yields, and the margin that matches it rises to 7.96%, so FY2036 operating income would have to be 2.55 times the reference's, where it was 2.49 times at the earlier close. The relation between price and value is the one we rated. At the 11-09-2026 close the reference is 59.8% below the price and the upside 44.4% below, and inside the rate band we support no scenario or stress we report reaches it: the closest are the upside's 20-year fade at the band's low rate, $951.66, and the reference's 50-year fade at that rate, $900.95, which are the longest fades we tested for each. The values that do exceed the close are the same ones that exceeded it on 11-02-2026, the upside at a 6.00% rate, $1,276.41, and its 20-year fade at that rate, $1,711.18, and a 6.00% rate is one we do not support. The 11-09-2026 close is 25.4% above the $850.00 low and 2.6% below the $1,094.32 high of the 500 sessions captured to 11-02-2026, so the market record the rating cites is unchanged, and no threshold the report relies on has been crossed.

The October sales release

On 11-04-2026 Costco reported net sales of $24.50 billion for the four weeks ended 11-01-2026, up 12.6% from $21.75 billion a year earlier (the percentages are as Costco reported them), and $54.52 billion for the first nine weeks of FY2027, up 12.8% from $48.33 billion. Comparable sales and digitally enabled sales grew as follows, and Costco reported 943 warehouses.

This is a monthly operating datapoint and not a result. Nine of the 52 weeks of FY2027 are in it, and reported growth is lifted by gasoline prices and currency, as the gap between 11.3% and 7.5% in the nine-week comparable shows. The release reports three different measures, and each answers a different question. Net sales grew 12.8% in the nine weeks, and that includes new warehouses, gasoline prices and currency. Total company comparable sales grew 11.3%, which leaves out new warehouses but keeps gasoline and currency, and 7.5% once gasoline and currency are removed as well. Our reference path assumes net sales growth of 8.0% for the whole of FY2027, a rate that includes new warehouses and everything else in net sales, over 52 weeks and not nine. Because the periods and the definitions differ, the nine-week figures neither confirm nor contradict that assumption: subtracting a short-period comparable rate that excludes new warehouses, gasoline and currency from a full-year total-sales rate would not compare like with like. We have no stored or captured figure for what new warehouses added in these weeks, so we do not build a bridge between the measures. The release also says nothing about capital spending or membership fees, so our $7.5 billion capital intention and our fee growth path are not tested by it. It changes no assumption and moves no value.

Eleven years of financial history

The forecast in this report extends a record, and the record is longer than the three years of statements the model reads directly. The eleven fiscal years FY2016 to FY2026 below come from Costco's XBRL company facts as registered with the SEC, each figure read from the filing that reported it and each derived ratio computed from the figures shown. Two limits apply. Net sales and membership fees are tagged separately only for FY2016, FY2017 and FY2023 to FY2026, so the fee history has a gap for FY2018 to FY2022 that would need Costco's earlier Form 10-Ks, which this report has not captured; the gap is unknown and nothing is interpolated. And FY2017 and FY2023 were 53-week years, so growth in those years and the years after them is distorted by an extra week, which is 1.9% of a 52-week year. Amounts in these tables are billions of dollars unless stated, and each of the four history tables is printed in two parts with the same rows in each, FY2016 to FY2021 and then FY2022 to FY2026, except the fee table, whose parts show only the years it observes. A figure the record does not have for a year, such as growth in the first year of a series, prints as a dash.

The income statement's first six years, FY2016 to FY2021, run through the steady stretch and into the surge.

Its last five years, FY2022 to FY2026, are the base the forecast extends.

Over the ten years from FY2016 to FY2026, total revenue grew 155% in total, a compound 9.8% a year, operating income 218%, or 12.3% a year, and diluted EPS 289%, or 14.6% a year. Earnings outgrew sales because the operating margin on total revenue rose from 3.09% to 3.85%, and diluted EPS grew slightly less than net income, 289% against 293%, because weighted diluted shares rose 0.7% over the decade.

What the years teach

FY2016 to FY2019 were a steady, unremarkable stretch: total revenue grew between 7.9% and 9.7% a year, the operating margin hovered near 3.10%, and the one large change was below the operating line. The effective tax rate fell from 34.3% in FY2016 to 22.3% in FY2019, which lifted net income faster than operating income. We read that as the effect of the December 2017 U.S. tax law, which is our judgment and is not stated in the filings we captured. The rate has stayed between 24.0% and 25.9% in each of the seven years from FY2020 to FY2026, which is why the model taxes operating profit at 25%. The lesson for the forecast is that the early-decade earnings growth owes something to a tax rate that cannot fall again.

FY2020 to FY2022 were the surge. Total revenue grew 17.5% in FY2021 and 15.8% in FY2022 and operating income 23.4% and 16.2%. SG&A fell from 9.83% of total revenue to 8.71%, a gain of more than a point, while merchandise costs rose from 86.91% to 87.85%, nearly a point: the company let sales growth leverage its expenses and gave most of the gain back in price. Inflation also took its toll through the last-in, first-out inventory method, which the registered company facts tag for FY2022 at a charge of $438 million (the other years are unknown here, apart from $142 million in FY2025 and $206 million in FY2026). The lesson is that a surge year does not tell us the steady-state growth rate, and that the operating margin on total revenue held at 3.42% and 3.43% through the inflation of those two years.

FY2023 and FY2024 were the normalization. FY2023 had 53 weeks and total revenue grew 6.8%; FY2024, with 52 weeks against 53, grew 5.0%, the slowest year of the decade, and partly because of the extra week in the base. Operating income nonetheless grew 14.4% in FY2024 as the merchandise margin kept climbing. FY2024 is also the year of the special dividend: $15 a share, about $6.66 billion in cash according to the FY2025 Form 10-K, which is why dividends paid were $9.04 billion in FY2024 against $2.18 billion in FY2025 and why net cash fell from $8.78 billion to $5.25 billion.

FY2025 and FY2026 are the base for the forecast. The September 2024 membership fee increase lifted fee growth to 10.3% and 11.0%, the merchandise margin reached 1.94%, and capital spending stepped up from 1.85% of total revenue in FY2024 to 2.12% in FY2026, 2.41 times depreciation. Free cash flow reached $9.38 billion in FY2026, and the balance sheet ended the year with $15.1 billion of cash and investments beyond the debt.

The fee and margin table prints the years the record observes, FY2016 and FY2017 first. Net sales and fees are not tagged separately for FY2018 to FY2022, so those five years are unknown and are left out of the table rather than printed as columns with no figure in them.

From FY2023 the record observes every line again, FY2023 to FY2026, except fee growth in FY2023, which would need the unknown FY2022.

The fee table shows two things the income statement hides. The merchandise margin has more than doubled since FY2016, from 0.88% to 1.94%, and the analytical residual it implies, not an audited profit contribution, grew from $1.03 billion to $5.78 billion, so the fee, which was 72.1% of operating income in FY2016, is 50.6% now because the goods have earned more, not because the fee has shrunk (fees rose from $2.65 billion to $5.91 billion). That is also a warning about how to read the ratio. It is arithmetic, not a disclosed fee profit, and it moves when the merchandise margin moves. The second thing is the margin's direction: gross margin fell while SG&A fell faster, which is what a company does when it prefers volume to margin and its operating costs scale. The FY2026 gross margin of 11.09% is below FY2016's 11.35% and SG&A of 9.15% is far below FY2016's 10.40%.

The cash-flow table's first part covers FY2016 to FY2021, including two of the three years of unusually large dividends.

The second part covers FY2022 to FY2026, including the third.

The cash-flow table is about reinvestment and returns. Free cash flow totaled $56.5 billion over the eleven years, dividends $30.0 billion and repurchases $5.79 billion, so Costco returned 63% of its free cash flow, and the special dividends account for a large part of what it returned. Only the FY2024 special dividend is documented in the filings this report holds, about $6.66 billion in cash; the FY2017 and FY2021 payouts are our inference from size alone ($3.90 billion and $5.75 billion of dividends against $1.50 billion in FY2022), because no registered filing describes them. Free cash flow has run close to net income in recent years (102% in FY2026). Capital spending, between 2.11 times and 2.41 times depreciation in the last three years, is only partly covered by depreciation: in FY2026 depreciation and amortization were $2.67 billion against $6.44 billion of spending, and the rest came out of operating cash flow of $15.82 billion, which is what leaves the free cash flow of $9.38 billion in the table. The stock-based compensation line matters for how to read that: operating cash flow adds stock compensation back, but the model does not, because operating income already bears it. The share count has barely moved, with weighted diluted shares up 0.7% over the ten years; repurchases exceeded stock compensation in FY2025 ($903 million against $860 million) and fell short of it in FY2026 ($848 million against $924 million).

The balance sheet table follows the same split, FY2016 to FY2021 first.

FY2022 to FY2026 complete it.

The balance sheet carries the third lesson. Payables exceeded inventory in every year from FY2018 to FY2026 except FY2022, when they fell just short of it: 1.17 times in FY2026 against 0.85 times in FY2016, with inventory turning in 27 days and suppliers paid in 31. Suppliers fund the stock, and the members' fees arrive in advance and sit in deferred membership fees of $3.01 billion. Net property and equipment of $35.6 billion supports total revenue of 8.51 times that amount, down from the 9.21 times of FY2022, and lower in each of the four years since, as the company has built faster than sales have grown; the forecast's capital-intensity assumption is a bet on where this ratio settles.

Normalizations and what they do to cash

We make few adjustments, and each is stated. The 53rd week in FY2017 and FY2023 is left in the reported figures and flagged. The LIFO charge stays in earnings, because it is a real cost of inflation. Management's statement that the fourth quarter of FY2026 included a non-recurring benefit of $0.15 a share from refunds of IEEPA tariffs, net of partial reinvestment in prices, takes FY2026 EPS from $20.76 to $20.61; that is the only adjustment to earnings the model makes, it is management's number and not a reported one, and it is why the reference margin is 1.91% and not 1.94%. The special dividends are a cash event with no effect on operating cash flow. Operating leases are inside operating income as rent, finance leases of $1.41 billion are deducted as debt in the bridge, and stock compensation is treated as an expense. The effective tax rate in the model is the flat 25% that the seven-year range of 24.0% to 25.9% supports.

What the history says about the forecast is partly reassuring and partly not. Reassuring: the reference's ten-year net sales growth of 6.0% a year is below the 9.8% a year at which total revenue grew over the last ten (net sales are not tagged for every year, so the history's growth is on total revenue), and the operating margin on total revenue, observed in all eleven years, fell by about 0.08 of a point at most, from FY2022 to FY2023, the 53-week year. Less reassuring: in the years the record observes the merchandise margin (FY2016, FY2017 and FY2023 to FY2026) it rose in every year-on-year step the record observes, FY2016 to FY2017 and each step from FY2023 to FY2024 through FY2025 to FY2026, while FY2018 to FY2022 are unobserved, so the step from FY2017 to FY2023 is unobserved too and we make no claim about it, and the reference holds the margin where the last year left it; and the reference's capital intensity from FY2028, shown in the forecast section, is lighter than any year in these tables.

How the forecast works

The forecast has nine drivers, set once for each scenario and held fixed: net sales growth in FY2027 and in FY2036, with a straight-line fade between; membership fee growth over the same two points; a merchandise margin reached by FY2029; net sales per dollar of net PP&E from FY2029; inventory less payables as a share of net sales; terminal growth and the return on new capital. Everything else follows from them by the arithmetic of the model: operating income is fees plus the merchandise margin times net sales; after-tax operating profit is that less 25% tax; depreciation is 0.9% of net sales, as in FY2026; capital spending is depreciation plus the growth in net PP&E that the sales-to-PP&E ratio requires; and working capital moves with sales, with deferred membership fees moving with fees. This section sets out the evidence for each driver, the links among them, and the annual schedule that results for all three scenarios.

The drivers and the evidence for each

Net sales growth. FY2026 net sales grew 10.1%, and the arithmetic of that growth is 8.3 points from comparable sales and 1.8 points from new warehouses and other differences; with gasoline and currency removed, total net sales grew 8.3%, after 9.8% the year before. The reference starts FY2027 at 8.0%, against 8.3% for FY2026 on that basis (a difference of -0.34 points). That comparison is like for like only because both figures are full-year total net sales including new warehouses; the nine-week figures in the October release are not, as the section on the release explains. The reference fades to 4.0% by FY2036. Planned openings support the first of those years: management intends up to 33 openings in FY2027, 28 net of relocations, which is 3.0% unit growth, and new units have added about two points of growth in each of the last two years, so the remaining six or so points must come from comparable sales, which excluding gasoline and currency grew 7% in FY2026 and 8% in FY2025 as printed. The fade is the judgment: ten years at a compound 6.0% is below the 9.8% a year at which total revenue grew over the last ten, and we adopt it as a base rate for a company whose revenue has already grown 155% in a decade and that has a finite number of sites. The downside runs at 4.0% and the upside at 7.5%. Moving the growth path from its reference to 6.0% fading to 3.0%, with fee growth moved alongside, changes the value by -$40.23 a share, and to 9.0% fading to 6.0% by $42.35.

Membership fee growth. The reference assumes 7.5% in FY2027 fading to 4.5%, and no further fee increase in the ten years, which is a judgment because the filings describe one increase, in September 2024, and no cadence. Fees need no capital in the model, which is why this driver is worth more than its size suggests; the unit-economics section shows why its start looks high against member growth. A lower start of 6.0%, fading to 3.5%, changes the value by -$20.86 a share, and a further increase at 8.5% fading to 6.0% changes it by $22.08.

Merchandise margin. The reference holds 1.91%, FY2026's margin without the tariff benefit, from FY2029, reached in three equal steps. The margin climbed 0.15 points a year on average over FY2023 to FY2026, from 1.49% to 1.94%, and from 0.88% in FY2016, but most of the three-year climb came in its first year: from 1.79% in FY2024 to 1.87% and then 1.94%, each of the last two steps was less than a tenth of a point. Extrapolating the climb is as defensible as stopping it, and we stop it because management states that it reinvests efficiency in price. The upside's 2.25% is about two more years of the three-year average climb, or about four at the pace of the last two years; the downside's 1.55% is close to FY2023.

Capital intensity. FY2026 net sales were 8.34 times year-end net PP&E. FY2027 capital spending is management's intention of $7.5 billion, which takes the ratio to 7.98, and the ratio then moves to the scenario level, 8.3 in the reference, by FY2029. Because depreciation follows sales and capital spending is depreciation plus the growth in PP&E, the move from 7.98 to 8.3 makes capital spending fall to $5.29 billion in FY2028 from $7.5 billion in FY2027 even as sales grow. As a share of total revenue that is 1.50% in FY2028 and 1.34% by FY2036, against 2.12% in FY2026 and a range of 1.69% to 2.23% across the eleven years of history. This is the assumption in the reference path that looks lightest against the record, and the one-at-a-time sensitivity on the ratio, $416.59 at 7.5 and $434.98 at 8.8, understates what it is worth. Holding capital spending at FY2026's share of total revenue from FY2028 would cost about $37.12 a share, taking the reference to $391.46, -8.7% against the reference; holding it at FY2020's 1.69%, the lowest year of the eleven, would cost $11.71, to $416.86. These figures are an illustration read from the stored schedule and not a model run; they change nothing in the valuation, and they belong in the bear case.

Working capital. Inventory less payables is -1.1% of net sales, FY2026's level and the most supplier-funded of the last four years (inventory less payables was -$3.27 billion in FY2026 and -$1.67 billion a year earlier). Management attributes the FY2026 reduction in its net investment in inventory to faster inventory turns and improved payment terms with suppliers. Other operating liabilities, net of receivables and other current assets, are -3.3% of net sales, and deferred membership fees are 50.9% of the year's fees. All three release cash as the company grows, which is why the model shows an annual inflow in every year of the reference, between $1.14 billion (FY2036) and $1.33 billion (FY2030).

Tax and terminal. Tax is 25% in every year, the seven-year range of the effective rate rounded. Terminal growth and the return on new capital, 3.0% and 18% in the reference, are discussed under terminal economics.

How the drivers link

The links are what make the schedule more than a growth rate. Growth in net sales raises capital spending through the sales-to-PP&E ratio, which is why a faster-growing path spends more even at the same ratio, and it releases working capital because suppliers and members fund the stock and the fees in advance. Growth in fees raises operating income one for one with no capital attached. The merchandise margin applies to net sales, so it multiplies the first link. In the reference, a dollar of added net sales earns about its margin of 1.91% before tax and needs capital of about one over 8.3, so it earns a small return on its own capital before the working-capital release and a large one after it; a dollar of added fees earns a dollar. That asymmetry is why moving growth in net sales alone is worth $413.46 to $444.52 while moving fees and sales together is worth $388.34 to $470.92.

The annual schedules for all three scenarios

Each table is the model's output at the stored inputs and the built central rate of 8.66%; present values are at the 08-30-2026 balance sheet date, with mid-year discounting. Free cash flow is after-tax operating profit plus depreciation less capital expenditure plus the working-capital and deferred-fee flow shown.

Reference path

The reference's free cash flow falls from $9.38 billion in FY2026 to $6.08 billion in FY2027, because capital spending steps up to the management intention and FY2026's working-capital inflow does not repeat at that size, then recovers to $14.2 billion by FY2036. Operating income grows from $11.7 billion to $20.7 billion, a compound 5.9%, and the reference's cumulative free cash flow over the ten years is $109.3 billion on cumulative capital spending of $67.1 billion.

Downside path

Upside path

Read together, the three paths differ more in operating income than in sales: the upside's net sales in FY2036 are $612.2 billion against the downside's $439.9 billion, while operating income is $26.2 billion against $15.6 billion, because the margin and the fees both differ. Fees are 56.2% of the downside's FY2036 operating income and 47.5% of the upside's, so the downside is the path in which the fee carries the business and the upside the path in which the goods do.

The three scenarios

The scenarios are conditional paths. They carry no probabilities, none is a forecast, and the reference is the path we find best supported by the record, not a midpoint of the other two.

The bold values use the central discount rate of 8.66%. The two rate rows use the ends of the band built below, and each is carried to the price date at its own rate. The same rate applies to all three scenarios, because the scenarios already differ in their cash flows and stacking scenario risk on the rate as well would count it twice.

The reference keeps the FY2026 merchandise margin excluding the tariff benefit (1.91%). That margin has risen for three years and Costco says it reinvests efficiency in price, so holding it is the midpoint between continuing the climb and giving some of it back. We assume no further membership fee increase in the ten years, which is a judgment: the filings show only the September 2024 increase and no cadence, and one more increase, with fee growth of 8.5% fading to 6.0%, would add to the value (the sensitivity table shows by how much). We also keep inventory less payables at -1.1% of net sales, the most supplier-funded level of the last four years, because management attributes the FY2026 improvement to faster inventory turns and improved payment terms with suppliers. That is a judgment that supplier funding stays near this level; if it drifted back toward -0.4%, the value would be $421.46.

The downside is an enduring impairment, not a trough. It keeps the merchandise margin at 1.55%, close to the 1.49% Costco earned in FY2023, for good, with net sales growth about a third lower than the reference's (4.0% a year compounded over the ten years, against 6.0%) and weaker capital efficiency. A temporary dip that recovered would be worth more than this. The upside combines a faster sales-growth path (the model has no warehouse count), fee growth that includes a further increase, a margin of 2.25% and a long, high-return terminal. Its 30% return on new capital is about what Costco measured on its operating capital in FY2026 on a narrow basis (29.8%), so it is generous in combination rather than in any one number.

The reference path

Free cash flow falls in FY2027 because capital spending jumps to management's intention and because FY2026's working-capital inflow does not repeat at that size. It then recovers as growth slows and the warehouse base matures.

Equity bridge and the terminal

The terminal value is 60.2% of enterprise value, which is ordinary for a ten-year model of a growing business and also a reminder that most of the value sits after the period we can see. We checked the step into it. In the last explicit year the reference reinvests 8.5% of its after-tax operating profit; in the first terminal year, at 3.0% growth and an 18% return on new capital, it reinvests 16.7%. Returns tell the same story from the other side. On the broad basis the forecast uses (net PP&E plus working capital, with fees received in advance and other operating liabilities deducted), Costco earned 45.0% on operating capital in FY2026 and the reference earns 45.9% in FY2036; on the narrower basis that leaves out other operating liabilities it earned 29.8% in FY2026. The terminal then assumes that each new dollar of capital earns 18%, far below either figure.

That fade is a judgment, and we should say what supports it. A measured average return on an old, depreciated base overstates the return on a new warehouse built at today's cost, and Costco's ratio of total revenue to net PP&E has fallen in each of the last four years. Competition for sites and members, and a growing international mix, argue against returns on new capital staying at today's level forever. Nothing in the record measures where they settle, so we tested 12%, 25% and 30%: the reference is worth $404.17 at 12%, $442.24 at 25% and $448.10 at 30%. It is a real lever but a modest one next to the discount rate.

Terminal economics

Most of the value in each scenario sits beyond the ten years we forecast explicitly: 60.2% of the reference's enterprise value, 56.5% of the downside's and 66.1% of the upside's. A terminal value is a claim about what happens after the period we can see, so the useful test is to put the last explicit year beside the first terminal year and ask whether the step between them is one the business could make. The table does that for all three scenarios, using the value-driver formula the model uses: terminal value equals the first terminal year's after-tax operating profit times one minus growth over the return on new capital, divided by the discount rate less growth.

Three features of the step deserve attention. First, free cash flow changes from FY2036 to FY2037 by -10.6% in the downside and -6.1% in the reference, and is flat in the upside, even though after-tax operating profit grows by the terminal growth rate in each. The reason is reinvestment. In FY2036 the reference reinvests 8.5% of its after-tax operating profit, because the capital-intensity path has capital spending close to depreciation and working capital still releasing cash; in the terminal year it reinvests the 16.7% that growth of 3.0% requires at an 18% return on new capital. The terminal value therefore charges the growth it assumes, and it is not flattered by the unusually light reinvestment of the final explicit year. Second, the terminal value is 11.4 times FY2036 operating income and 14.7 times the first terminal year's after-tax operating profit, which is lower than the 15.2 times operating income that the reference's own enterprise value represents today and far below the 38.7 times that the market price represents. We are assuming that the multiple compresses as growth falls from 4.0% to 3.0%, and that is the point at which our view and the price part company most visibly. Third, the cash yield the terminal pays is 5.7%, which is the discount rate less growth, 5.66%, reached from the other side.

The independent check on the return assumption

The terminal's weakest input is the return on new capital, and the Costco record gives two independent ways to test it. The first is the measured return on the existing base. On the broad basis the forecast uses, net PP&E plus working capital with fees received in advance and other operating liabilities deducted, Costco earned 45.0% on operating capital in FY2026, and 29.8% on the narrower basis that leaves out other operating liabilities. The reference earns 45.9% on the broad basis in FY2036, and the terminal assumes that each new dollar earns 18%, which is 39% of that figure; the downside's 12% is 37% of its own FY2036 return and the upside's 30% is 52% of its own. The second check is the no-growth earnings-power lens: FY2026 operating income without the tariff benefit, taxed at 25%, gives $8.70 billion of after-tax profit, worth $252.17 a share capitalized at the central rate with no growth. The reference adds $176.41 to that, or 41.2% of its value, for ten years of growth and a growing terminal.

The fade in the return is a judgment, and the reasons for it are that the measured average return on an old, depreciated base overstates what a new warehouse built at today's cost earns, that Costco's ratio of total revenue to net PP&E has fallen in each of the last four years, and that competition for sites and members and a larger international mix argue against returns on new capital staying at today's level for ever. Nothing in the record measures where they settle. Our first review of the model argued that the evidence points higher than 18%, because Costco's measured return on operating capital is nearer 30% on the narrow basis and the reference's own long-run drivers imply about the same on the capital a dollar of new sales needs once fees come with it. We kept 18% and tested the alternatives: the reference is worth $404.17 at 12%, $442.24 at 25% and $448.10 at 30%, changes of -$24.41 and $19.53 for the ends, and terminal growth of 2.5% or 3.5% moves it by -$13.46 or $15.98. These are real levers, and modest next to the discount rate. The terminal value is only as good as the horizon it follows, and the bull case below is largely an argument about that horizon.

Valuation methods and the equity bridge for each scenario

Three lenses sit side by side in this report, and they measure different things, which is why they give different numbers. The ten-year model with a value-driver terminal, our main method, values the cash the business is forecast to produce, including the growth we assume. The earnings-power lens values only what the business earns today, capitalized with no growth, so the difference between the two, $176.41 a share on the reference, is the value we put on growth. And the market price, read as a multiple, shows what an investor pays for today's earnings: the close of $1,040.55 implies an enterprise value of $452.4 billion on the model's bridge, 38.7 times FY2026 operating income and 51.6 times after-tax operating profit, against 15.2 times and 20.3 times for the reference. The two enterprise values are at different dates (the market's at 11-02-2026, the model's at 08-30-2026), so the comparison is one of scale, and it says the market's enterprise value is 154% above the reference's. No peer multiple is used, because no peer in the record is priced on the valuation session.

The equity bridge for each scenario

The bridge from enterprise value to a value per share is the same arithmetic in every scenario; only the enterprise value differs. Cash and short-term investments of $21.30 billion less $2.84 billion of unsettled card receivables, which are operating working capital and not surplus, give surplus cash of $18.46 billion, or $41.46 a share, which is 4.0% of the price. Debt and finance leases are deducted, operating leases are not, and the result is divided by 445.266 million diluted shares. The value is then carried from the balance-sheet date to the valuation date at the central rate and the $1.47 dividend deducted.

The reference's value per share at the balance sheet date is $423.83 and at the valuation date $428.58; the carry is worth $4.75 a share, which is small beside the gap to the price. In every scenario the explicit period carries 39.8% (reference), 43.5% (downside) and 33.9% (upside) of enterprise value, and the close is 3.41 times, 2.43 times and 1.76 times the three values.

Why the reference is the best-supported path and not a midpoint

The reference is not the average of the other two. The midpoint of the downside and the upside is $448.89, which is $20.31 above the reference, and the distances are unequal on purpose: the reference is $123.50 above the downside and $164.12 below the upside. It is the continuation of FY2026's economics that the record best supports, with the FY2026 merchandise margin held at its tariff-free level, fee growth and sales growth fading from the last year's pace, management's capital intention in the first year and no further fee increase. The downside is an enduring impairment, not a trough, and the upside is generous in combination and not in any one number. The scenarios carry no probabilities, and nothing in this report weights them.

What separates them is easier to see by moving one group of drivers at a time. The table takes the reference, replaces a single group of its drivers with the downside's or the upside's, and reports the value; the five changes do not add to the whole gap because the drivers interact, and the interaction is shown and never added to anything.

Growth, the terminal and the merchandise margin account for most of the distance in both directions, and capital intensity and supplier funding for little. That ordering is also the ordering of what we do not know: how long Costco keeps growing, and what each new dollar of capital earns, are the two questions the record answers least well, and the discount rate, which sits outside this table, is larger than any of them.

The discount rate: how we built it and what we assume

The rate is the largest lever in the model, so it is built from dated inputs wherever a dated input exists, and the inputs that are judgments are labeled as such. The risk-free rate is the official U.S. Treasury 10-year constant-maturity par yield for 11-02-2026, 4.91%. The 20-year yield that day was 5.31% and the 30-year yield 5.28%. Cost of equity is the risk-free rate plus beta times an equity risk premium. Beta of 0.85 and a premium of 4.5% are our assumptions. Neither is measured here: no regression beta is available to us, and the premium is a convention we use across our research. We show a band for both, beta from 0.70 to 1.00 and premium from 4.0% to 5.5%.

Debt is a small part of Costco's capital, so the assumed debt spread barely matters. The band is driven by beta and the premium, the two assumed inputs that matter. At the low end the reference is worth $515.76 and at the high end $337.89; the upside is worth $741.88 at the low end and $447.96 at the high end. The price is above every scenario at every rate in the band, including the upside at its most favorable end.

The rate that would make the reference equal the price, holding everything else, is 5.24%. That is about 33 basis points above the 10-year Treasury yield, so matching the price would mean owning a business with Costco's risks for little more than the return on a ten-year government bond, and it is below the 20-year and 30-year yields. The upside alone would match the price at 6.52%, still well under the low end of our band. We cannot rule out that the market accepts a return that low, and we do not think it is the rate that fits a retailer whose margin is a choice.

We checked that the pieces of the valuation agree with one another. The cash flows are those of the whole firm, before interest, so a rate that blends equity and debt is the matching rate: Costco's interest income of $711 million and interest expense of $145 million are outside operating income and outside the flows, surplus cash is added in the bridge, debt and finance leases are deducted, and operating leases are not deducted because their cost is already in operating income. The flows are nominal and so is the rate; terminal growth of 3.0% sits against 10-year breakeven inflation of 2.29% (the 4.91% nominal yield less the 2.62% real yield) and below the risk-free rate. Discounting is mid-year throughout, and one rate serves all three scenarios. Carrying the reference equity value to the price date at the cost of equity instead of the blended rate would add $0.05 a share, which is immaterial.

What the price fits

The price fits many combinations of growth, margin and rate, and the model shows a few of them. Holding the reference's other drivers and the central rate, the model matches the price at a long-run merchandise margin of 7.72%, which is 4.03 times the 1.91% Costco earned in FY2026 without the tariff benefit. Because that test holds fees fixed, a fairer way to say it is in operating income: matching the price this way means FY2036 operating income 2.49 times the reference's. In Costco's own presentation that is an operating margin on net sales of 9.71% in FY2036, against 3.93% in FY2026. With SG&A at FY2026's 9.15% of net sales it would need a gross margin of 16.9%, against 11.09%; with gross margin at its FY2026 level it would need SG&A of 3.4% of net sales, against 9.15%. That describes how far the cash flows would have to move, not a gap in definitions: the merchandise margin and Costco's reported lines are the same quantities, as the bridge above shows.

These are conditional break-even combinations, not a statement of what investors believe, and a reader can reasonably attribute the gap to a lower rate, a longer high-growth period, or both.

A second lens does not depend on our forecast. FY2026 operating income without the tariff benefit, taxed at 25%, is $8.70 billion of after-tax profit. Capitalized at the central rate with no growth and run through the same bridge, it gives $252.17 a share. That is what the business earns today before any growth; the reference value of $428.58 is what ten years of growth and a growing terminal add to it. At $1,040.55 the market pays 50.1 times FY2026 earnings, for an annualized cash dividend yield of 0.57% at the declared quarterly rate and a free cash flow yield of 2.0%.

The market has also been consistent. Over the 500 sessions to 11-02-2026 that we captured, COST closed between $850.00 and $1,094.32, and the 11-02-2026 close is 17.4% above the first close in that window and 22.4% above the low. Our reference sits 49.6% below even the low. Either the market has paid more than the business was worth for two years, or our framework values this kind of durability too cheaply. We take that second possibility seriously, which is why we develop it as the bull case below.

The market-implied table

The break-evens above are single combinations. The table lays the ones we can compute side by side, each holding the rest of the reference fixed and each conditional: they describe what the price fits, not what investors believe and not what the price requires. They are solved at the 11-02-2026 close; the section "What the 11-09-2026 close fits" solves the same routines at the 11-09-2026 close and sets the two side by side.

Two of those rows are about time, and they are the least intuitive. At the central rate, stretching the reference's growth fade from ten explicit years to fifty never reaches the close, and a search to eighty years does not either; at the band's low rate it takes 68 years. Those are the reference's durations; for the upside we tested a 20-year fade only, so a longer upside fade inside the band is untested. The table below shows how the value responds to duration at each of the two rates, and the ladder after it shows the same for the discount rate, where the reference and the upside are each run at illustrative rates inside and outside the band.

Read together, the ladder and the fade say that the rate does most of the work and the duration the rest. A reader who believes the right rate is 6.00% and the right operating case is the upside's holds a view that fits the price, and a reader who believes the rate is 7.65% or above does not, among the durations we tested (the reference's fade to fifty years in the table, the upside's to twenty), unless the reference's growth lasts 68 years at the band's low rate. We do not think the first view is supported by the record, because a 6.00% rate asks for a cost of equity about a point above the government bond on a stock that has ranged between $850.00 and $1,094.32. And we do not rule it out, which is why the beta, premium and spread that build the rate are stated as assumptions.

The cross-check built a different way

The earnings-power value and the return-on-capital check in the terminal section are the independent tests of the cash-flow model. The first needs no forecast: today's after-tax operating profit of $8.70 billion, capitalized with no growth, is worth $252.17 a share. On reported FY2026 operating income taxed at 25%, with the tariff benefit left in, the market pays 51.6 times after-tax operating profit for the enterprise, and so values growth at a price we cannot reproduce. The second tests whether the model's returns are in line with the record: the reference earns 45.9% on operating capital in FY2036, against 45.0% measured in FY2026 on the same basis, and the terminal assumes only 18% on each new dollar. Both checks say the model is not conservative about the business, and not generous either.

Sensitivities

Each row moves one assumption in the reference and leaves the rest. Growth rows are shown two ways, because Costco's fee stream needs no capital while extra sales do, so a sales-only change understates what growth is worth.

The rate row reads lower input first, so the low-end rate gives the higher value. The method choices described earlier show their effect the same way: using end-of-year discounting gives $412.09, leaving out the other operating liabilities gives $416.41, using our capital-intensity rule for FY2027 capex instead of management's intention gives $429.02, and treating a further 20% of cash as needed for operations gives $420.16. One further membership fee increase, with fee growth of 8.5% fading to 6.0%, gives $450.66.

The nine assumptions we challenged

Before this report was finished, two separate checks were made. A second implementation of the model, written apart from it, sharing none of its code and reading the registered filings, market data and Treasury yields directly, rebuilt the discount rate and its band, the three scenario values at the central rate and at both ends of the band, and the break-even discount rates and merchandise margin, and it matched every one; it did not rebuild the history tables, the annual schedules, the bull and bear lenses or the effects in the table below, which come from the model's own code at the stored inputs. And every assumption that moves the value was challenged in a review by someone who had not written it. The challenge was mixed: on some assumptions the evidence agreed with us, on some it disagreed immaterially, and on one, the discount rate, it disagreed on the foundation, which is why the rate in this report is built from a dated Treasury yield and not typed. The table puts each assumption's input, our test range and the effect on the reference at the rate we now use, beside the effect our first review recorded, which was computed on an earlier, simpler version of the model at a typed 8% rate and is kept for the record.

Each assumption is discussed here under the name the review gave it.

Discount rate. The review disagreed with the foundation and not the level: the rate was typed, and it is the largest single lever. The built rate of 8.66% answers that. At 7% and 9% the reference is worth $595.59 and $405.85, effects of $167.01 and -$22.73 from a base that is now 8.66%, which is why these figures differ from the first review's $110.95 and -$73.80 from 8%. The reference reaches the price only at 5.24%.

Long-run merchandise margin. The review agreed that holding the latest margin without the tariff benefit is a defensible center, and noted that the stated reason, a narrow band, was really three rising years at the top of the record. We rewrote the reasoning. At FY2023's 1.49% the reference is worth $384.34 and at the upside's 2.25% $464.39.

Terminal RONIC (return on new invested capital). The review disagreed immaterially: the evidence puts the return higher than the 18% we use, because Costco measured nearer 30% on its operating capital. We kept 18% as a stated fade, and the terminal-economics section gives the reasons. At 12% the reference is worth $404.17 and at 30% $448.10.

Sales-to-PP&E ratio. The review disagreed immaterially on the first forecast year, which contradicted management's capital intention; we now use the intention, which makes the first-year ratio 7.98. At the low test of 7.98 held and the high test of 8.60 (FY2024's) the value changes by -$4.56 and $3.91. The forecast section shows that the ratio is a smaller lever than the capital-spending share it implies, which is larger.

Net sales and fee growth path. The review agreed with the reference: it starts below FY2026's 10.1% because gasoline and currency contributed, and the fade to 4.0% is a base-rate judgment that the counter-thesis contests. Moving the two growth rates together to 6.0% or 9.0% fading to 3.0% and 6.0% gives $388.34 and $470.92.

Membership fee growth (pricing). The review disagreed immaterially and asked us to say plainly that the reference assumes no further fee increase for ten years on a stream whose ratio to operating income, an arithmetic ratio and not a reported fee profit, is 50.6%, which we now do. A start of 6.0% gives $407.71 and one further increase, fee growth of 8.5% fading to 6.0%, gives $450.66.

Terminal growth. The review agreed with 3.0%: nominal growth in line with a mature retailer, and the value-driver formula already charges the reinvestment growth needs. At 2.5% and 3.5% the value is $415.12 and $444.55.

Inventory less payables / net sales. The review disagreed immaterially: the reference holds the FY2026 level of -1.1%, the most supplier-funded of the last four years, and a reversion is asymmetric. At -1.3% the value is $430.61 and at -0.4% $421.46.

Surplus share of cash and short-term investments. The review disagreed immaterially and sized the unsettled card receivables, which we now remove from cash. Treating a further fifth of cash as needed for operations gives $420.16, -$8.41.

The review's counter-thesis was about duration, not margin: Costco has grown total revenue 9.8% a year over ten years, the reference fades to 4% within ten, and a company that funds its inventory with suppliers' money can compound for longer. That is the case the bull section develops. Its outside view was that no consensus or third-party research is captured and none is inferred, that Costco itself repurchased shares at an average of $964.62 across the fourth quarter of FY2026 (transaction prices, not a statement of value), and that the model and the price differ in duration and rate far more than in near-term economics. The review's overall reading was mixed, and it did not move the reference toward the price to close the gap.

Sources and basis

Reported figures come from Costco's Form 10-K for the 52 weeks ended 08-30-2026 (filed 10-07-2026) and its comparatives, the fiscal fourth-quarter earnings release and supplemental information (09-24-2026), the dividend declaration (10-14-2026) and the SEC's XBRL data for the company. The risk-free rate and breakeven inflation come from the U.S. Treasury's daily par yield curve and daily real yield curve for 11-02-2026. Peer context comes from Walmart's Form 10-K for the year ended 01-31-2026 (filed 03-13-2026), including its Sam's Club U.S. segment, and BJ's Wholesale Club's Form 10-K for the year ended 01-31-2026 (filed 03-12-2026), both read from captures made on 11-03-2026. The eleven-year history comes from Costco's SEC XBRL company facts, and the membership, warehouse, cohort, segment and sales-mix figures from tables and text in the FY2026 and FY2025 Form 10-Ks. The FY2018 to FY2022 split of net sales and membership fees, and the paid-member and warehouse counts for those years, are unknown to this report because the earlier Form 10-Ks are not among the documents it holds, and nothing is estimated in their place. Evidence is gathered through 11-09-2026, and the only item observed after 11-07-2026 is the November 9, 2026 regular-session closing price of COST, a separate market snapshot taken after that session settled; the filings, the model, the valuation, the Treasury yields and the disclosures are as of their earlier dates. Walmart's Form 10-Q for the quarter ended 07-31-2026 and Target, Kroger and Amazon filings were not used for this section's comparisons. The valuation price is the November 2, 2026 regular-session closing price of COST. The later closes through 11-06-2026 and a second set of Treasury yields were read from the same kinds of source on 11-07-2026, Costco's SEC filing index was read the same day, and the October sales release is the company's own news release of 11-04-2026. The November 9, 2026 regular-session closing price was read on 11-09-2026. Closing prices are split-adjusted and not dividend-adjusted. None of them changes a value in this report. Figures in this report are USD and reported GAAP unless stated. Reported figures, management statements (the capex intention, the tariff benefit, the fee-increase contribution), our calculations, our assumptions and our judgments are different things, and we say which is which where it matters. Beta, the equity risk premium and the debt spread are our assumptions and have no source. Management's statements are its own, and a calculation is not a reported figure. No third-party research or consensus was used.

Disclosure

This is research, not individualized advice. The Blockhaven Rating is our assessment of risk and reward at the stated price and horizon, here the ten years of our model; it is not a price objective and is never a recommendation to transact.

Important disclosures

Blockhaven research is provided for general informational and educational purposes. It is not individualized investment advice and does not take into account any reader's financial circumstances, investment objectives, risk tolerance, tax situation or portfolio.

Nothing in this report is an offer or a solicitation to buy or sell a security or other financial instrument. Reading or accessing Blockhaven research does not by itself create an investment-advisory, fiduciary, brokerage or client relationship.

Investing involves risk, including the possible loss of principal. Scenario values, terminal values, valuation ranges, sensitivities and forward-looking assumptions are analytical tools, not guarantees or promises of future performance, and actual results may differ materially.

Blockhaven uses information from company filings, company publications, market-data providers and other sources it believes to be reliable, but that information may be incomplete, delayed, revised or incorrect.

Opinions, ratings and interpretations reflect Blockhaven's research judgment as of the report date and may change as new evidence becomes available.

Positions and conflicts

As of November 6, 2026, Blockhaven's owner held no shares, options, or other direct derivatives in Costco Wholesale Corporation; had no material positions in Costco Wholesale Corporation through entities the owner controls; and had no paid financial relationships with Costco Wholesale Corporation.

© 2026 Blockhaven LLC. All rights reserved.