Costco

IThe business

Costco sells a membership and then sells groceries, gasoline, televisions, tyres, prescriptions and cruises to the people who hold it. The US fee is $65 a year. An Executive membership costs another $65 and pays back 2% of qualifying purchases, up to $1,250 a year.³¹⁰ There were 84.1M paid members at the end of fiscal 2026, 42.3M of them Executive, and 150.4M cardholders once the free household cards are counted.³

The idea is old and simple. Carry a narrow range, fewer than 4,000 items in a warehouse, which the 10-K calls significantly less than other broadline retailers, and buy each of them in huge volume.³ Sell them in plain buildings, often on the pallet they came on, with short opening hours. Price them with very little margin. The 10-K says the volume and turnover let Costco operate “at significantly lower gross margins” than most retailers.³ In fiscal 2026 the gross margin on net sales was 11.09%.³ Kroger, one of the largest US supermarket chains, reported 22.9% for its 2025 year, about twice as much.¹¹

Then charge for entry. The fee does two jobs. It is most of the profit, and it is a reason to come back: once you have paid $65 or $130, you want to use it. Costco also says the format cuts theft, since entrances and exits are controlled.³

What it sells

Food and sundries were $116.8B of fiscal 2026 net sales, non-foods $77.4B and fresh food $41.7B. The fourth bucket, warehouse ancillary and other businesses, was $61.4B. That covers gasoline, pharmacy, optical, the food court, hearing aids, tyre fitting, e-commerce, business centres and travel.³ It grew 20% in the year, twice as fast as the core, led by gasoline and pharmacy.³ Gasoline alone was 11% of net sales from 774 stations.³

That mix shift is worth understanding, because it flatters the headline. Pump prices rose 12% in the year and added $3.5B, or 1.3 points, to sales growth. Currency added another half point.³ Without them, comparable sales rose 6.6%. The company says comparable sales came from a 5% rise in the average ticket and 3% more trips.³ The ticket includes inflation, which the CFO put in the low single digits in the fourth quarter.⁹ So the real gain in units is smaller than the 8.4% reported comparable figure, though still good for a business this size.

Kirkland Signature, the house brand, is the other quiet engine. The company says it saves members at least 15% to 20% against the national brand at equal or better quality, and that private label generally earns higher margins.⁹³ Costco doesn’t publish its Kirkland share of sales, so I can’t measure it.

Where the money comes from

The US and Puerto Rico earned $7.6B of the $11.7B operating income in fiscal 2026, Canada $2.1B and Other International $1.9B.³ The US and Canada were 86% of net sales and 83% of operating income. California alone was 26% of US net sales.³ That last figure is a concentration I would not have guessed. One state’s economy, wages and rules matter a lot to this company.

Abroad, Costco runs 43 warehouses in Mexico, 37 in Japan, 29 in the UK, 20 in Korea, 15 in Australia, 14 in Taiwan, seven in China, five in Spain, three in France, two in Sweden and one each in Iceland and New Zealand.³ Other International had the fastest sales growth in fiscal 2026, at 12%.³ Management’s comment on the call is useful: in Asia, warehouses have many more members who visit less often; in the US, fewer members spend much more each.⁹

This chart is the single most important fact on the page, so I will spell it out. In fiscal 2016 the fee was $2.65B and operating income $3.67B. The fee was 72% of the profit, and the merchandise itself earned 0.95% after its costs C (calculated).⁴⁶ In fiscal 2026 the fee was $5.91B, still growing, but merchandise earned $5.78B, or 1.94% of net sales C (calculated).³ The margin on goods roughly doubled. The common story says Costco sells goods at cost and lives on the fee. That was close to true ten years ago. It is half true now.

I don’t read that as greed. The gross margin barely moved, from 11.35% in fiscal 2016 to 11.09% in 2026 C (calculated).⁴³ The gain came from costs: SG&A fell from 10.40% of net sales to 9.15% C (calculated).⁴³ Each warehouse sells much more, and a building’s rent, staff and power don’t rise as fast as its sales. The question for the valuation is how much more of that saving Costco keeps, and how much it gives back. Its whole model says give most of it back.

Warehouses, and what each one sells

Costco ended fiscal 2026 with 939 warehouses, 25 more than a year before, from 28 openings including three relocations.³ It owns the land and building at 755 of them, and owns the building on land it leases at 136 more.³ It plans up to 33 openings in fiscal 2027, five of them relocations, and says it is building toward 30 net new warehouses a year.³⁹ The CFO said on the call there is a path for that pace “for the next 5 to 10 years”.⁹

The table behind this chart is in the 10-K every year, and I think it is underused. It shows that new warehouses start at about two thirds of the chain average and keep climbing for years. The class of 2020 opened 13 buildings at an annualized $132M each and did $268M in 2026.³ Warehouses opened before 2018 averaged $306M.³ So new units don’t dilute the chain for long, and the old units are still growing. The average went from $163M in fiscal 2017 to $292M in 2026, about 6.7% a year C (calculated).³ Some of that is inflation and gasoline. Most of it is more members spending more.

Members

Membership is where the recent news is less good. Paid members grew 6.3% in fiscal 2025 and 3.8% in 2026.³ An analyst on the September call said membership growth had slowed for eight quarters in a row. The CFO’s answer was that the recent rate is “more typical” of what to expect, and that a few Asian openings could lift it for a while.⁹ Warehouses grew 2.7% in the year, so members per warehouse still rose, just slowly C (calculated).³

The renewal rate tells the same story. In the US and Canada it was 92.9% at the end of fiscal 2024 and 92.3% at the end of 2025 and 2026. Worldwide it went from 90.5% to 89.8%.¹⁵³ The company says members who sign up online renew at a slightly lower rate, and so do new international members.³ In the fourth quarter both rates rose 10 basis points from the third.⁹ That is a small move, but it was the first rise in a while.

The offsets are real. Executive members rose to 42.3M from 38.7M, and they made up about 75.2% of worldwide net sales, against 73.6% a year earlier.³¹⁰ Members under 40 have grown nearly 60% since the pandemic and are now more than a quarter of the base, management says.⁹ Younger members spend less at first and more as their households grow. Costco is trading slower headcount growth for richer members. That can work for years. It can’t work forever.

Digital, delivery and AI search

Digitally enabled sales, which include orders picked up in the warehouse and Costco Travel, were 11% of net sales in fiscal 2026, and comparable digital sales rose 21%.³ Management put the total above $33B on the call.⁹ It is widening delivery through Instacart, Uber Eats and DoorDash, and says those orders are mostly extra trips that don’t take from warehouse grocery.⁹ It closed Costco Next, its extended online marketplace.⁹

The CFO said traffic to the website from AI search tools grew at triple digits for a second quarter, converts better than any other source, and that the membership itself is among the top items those searches lead to.⁹ He also said Costco buys no paid digital advertising.⁹ I take two things from that. A company that wins on price is likely to do well when software compares prices for the shopper. And a company with no ad business at all has an option others already use: Walmart sells advertising inside its stores and apps. Costco has said nothing about doing so, and I put nothing for it in my numbers.

Tariffs, and the refund

Costco paid about $500M of tariffs under the International Emergency Economic Powers Act from February 2025. The Supreme Court ruled them unlawful in February 2026, and refunds began.³ It received $184M in the fourth quarter, $174M of it booked against merchandise costs, and $155M more after the year ended.³ It says it will spend most of the refunds on lower prices for members. In the fourth quarter the net effect was worth $0.15 a share.² I take the whole after-tax refund out of my starting earnings, $0.31 a share, not the $0.15 net, which is the more cautious reading; on the net figure the multiple would be 49.1 times instead of 49.5 C (calculated).² I treat the rest as passing through to members.

IIThe moat

A durable advantage is a cost or a habit that a rival with money and talent can’t copy in a reasonable time. Costco has both. What makes it unusual is that the two feed each other, and that the company has chosen for forty years to spend the gains on its customers.

The loop works like this. Low prices bring members. Members bring volume. Volume lowers the cost per item, in buying and in running the building. Costco passes the saving on, prices fall further, and more members come. The fee is the toll on the loop, and the renewal rate is the measure of whether members think the toll is worth it.

The test of any such advantage is whether it has widened or narrowed. Here is the evidence both ways.

Evidence it is widening

• Sales per warehouse rose from $163M to $292M in nine years, and each new class of warehouses climbs toward the old ones.³
• The merchandise margin doubled while the gross margin was flat, which means the cost advantage grew and members still paid the same margin on goods C (calculated).⁴³
• Executive members, who pay double and spend more, went from 38% of paid members in fiscal 2017 to 50% in 2026, and to 75.2% of sales C (calculated).¹⁴³
• Suppliers now fund more than all of the inventory. They accept that because Costco sells their goods fast and in volume.³
• Gasoline penetration among US member households reached a record in fiscal 2026, management says, and the company estimates it saved members over $3.2B against the average pump price.⁹

Evidence it is narrowing

• Paid member growth fell from 6.3% to 3.8% in a year, and renewal rates are lower than in 2024.³¹⁵
• Sam’s Club’s comparable sales rose 8.6% in its latest quarter, 4.4% without fuel, so the main rival is not standing still.¹⁶
• The last fee rise, in September 2024, accounted for about 40% of fee growth in 2025 and 30% in 2026; in the fourth quarter its effect ran out.³⁹ Underlying fee growth, without the rise and currency, was 6.8% in the fourth quarter.⁹
• Delivery through Instacart, Uber Eats and DoorDash means members shop Costco through someone else’s app. That is good for sales today. Over time it weakens the reason to walk into the building, which is where the treasure-hunt habit forms.⁹

On balance I think the advantage is wide and still widening in cost, and flat to slightly narrowing in membership. The cost side is the part that compounds. The membership side is the part the market prices.

IIIThe competitive advantage

The cleanest comparison is with the other two US warehouse clubs, because they sell the same kind of thing in the same kind of box. Costco’s fiscal year ends in August; Walmart’s and BJ’s end in January, so the dates differ by five months.

The table says something I didn’t expect. Costco does not take a lower gross margin than Sam’s. Its gross margin is a fraction lower, inside the noise of how each company defines it. Where it wins is the cost of running the building. Sam’s spends about 11.4 cents of every sales dollar on operations; Costco spends about 9.2 C (calculated).⁷³ That gap of about two cents is the whole advantage, and it comes from selling nearly twice as much in each box.

BJ’s shows the other side. Its operating margin is close to Costco’s, but most of BJ’s profit is the fee, and it sells about $80M a club C (calculated).¹⁷ It also trades at 23 times earnings against Costco’s 50 C (calculated).¹⁸¹ The market pays more than double for Costco’s earnings. Some of that is quality. Some of it is fashion.

Against general retailers the comparison changes. Walmart is far larger, with $713.2B of revenue, and trades at about 45 times trailing earnings C (calculated).⁷¹⁶¹⁸ Amazon’s Prime membership is the most direct threat to Costco’s fee, and Prime bundles delivery, video and music. I cover Amazon on its own page, where my base value is below the price as well.

Where Costco wins

• Bulk staples, fresh meat and bakery, gasoline and pharmacy, where volume per item matters most and the treasure hunt brings people in.⁹
• Big-ticket items such as appliances, televisions, jewellery and travel, where the return policy and the trust behind it are worth a lot.³⁹
• Suburban households with cars, storage space and children, the members who renew most.⁹

Where it loses

• Small households in dense cities, who don’t want 48 rolls of anything.
• Speed and range online. Costco carries 9,000 to 11,000 items online, against many millions at Amazon.³
• Markets where the model needs to be taught. Asian members visit less often, and the company is still working on frequency there.⁹

IVManagement

Costco promotes from inside. Ron Vachris took over as chief executive in January 2024 from Craig Jelinek. Vachris was president and chief operating officer for the two years before that, head of merchandising from 2016 to 2022, and before that ran real estate and the Northwest region.³⁸ The 10-K says most officers have more than 25 years with the company.³

The CFO is the notable outsider. Millerchip came from Kroger in 2024, when Richard Galanti, the long-serving CFO, handed over the job in March of that year.⁸ The bench below him is home-grown. Two of the most senior roles, merchandising and US operations, changed hands in 2026 within the same culture.³

Pay, and what it pays them to do

Vachris’s reported pay for fiscal 2025 was $13.9M. A salary of $1.2M, a cash bonus of $600,000 and performance stock with a grant value of $12.0M made up nearly all of it.⁸ He got 13,834 stock units, 19% more than the year before, which the committee justified by his performance and by what peer chief executives earn.⁸

The cash bonus is tied to the budget. For fiscal 2025 the pre-tax income goal was $10.73B and the company made $10.87B on its adjusted measure; the sales goal was $271.02B and it made $271.78B. Both paid in full.⁸ For fiscal 2026 the bonus could reach $1.12M, and for fiscal 2027 the base amount is $1.2M, split $500,000 on sales, $500,000 on pre-tax income and $200,000 on environmental and social measures.²⁰²¹ That is a large rise in percentage terms and small in dollars.

The stock is what matters. To earn the units for fiscal 2025, Costco had to grow net sales 3% or pre-tax income 2%, at constant currency.⁸ Either one would do. In a year the company grew sales about 8%, that bar is low enough to be close to a formality. After that the units vest 20% a year for five years. For most officers, long service speeds that up: a third vests at once after 25 years, two thirds after 30 and all of it after 35.⁸ His fiscal 2025 grant vests over the full five years.⁸

So follow the money. The structure pays executives mainly for the share price over the next five years and for staying. It barely pays them for any single year’s numbers. I think that fits this company. Paying a management that wants low prices and patient growth to hit a margin target would work against it. The weak point is the other side of the same coin: the pay doesn’t reward discipline on the price paid for the company’s own shares, or on capital spending, and nothing in it punishes a slow erosion of membership.

Some smaller facts. There are no options and no pension. The only change-in-control terms are those in the stock plan that apply to every holder of units.⁸ The proxy notes that executives get a free Executive membership card. The pay ratio was 283 to 1, or 210 to 1 against full-time staff only.⁸ At the January 2026 meeting, 88% of votes cast backed the pay plan, down from 94.7% the year before C (calculated).²²⁸ In May 2025 the committee let staff choose a three-year vesting schedule without the long-service speed-up; the CFO and the CIO took it.⁸

Their own money

Vachris owned 52,258 shares and units in November 2025, worth about $53M at $1,012.28 C (calculated).⁸¹ That is close to four years of his reported pay. All directors and officers together held 460,270, or about 0.1% of the company, worth about $466M C (calculated).⁸ Directors must hold at least $1M of stock within five years.⁸ Nobody running Costco owns enough to steer it. The two largest holders are index managers, Vanguard with 9.79% and BlackRock with 7.83%.⁸ The culture, not an owner, is what keeps management honest here, and culture is harder to check than a share count.

What they did with the money, over a full cycle

I take fiscal 2017 to 2026 as the cycle. It covers a tax cut, a pandemic, a burst of inflation and a tariff shock. Over those ten years Costco generated $95.6B of operating cash. It spent $39.7B on warehouses, depots and systems, paid $29.3B in dividends and bought back $5.3B of stock C (calculated).⁶³ Debt was $5.2B at the start and $6.2B at the end.³⁴ Cash and short-term investments rose from $5.8B to $21.3B.³¹⁴

Three things stand out. First, nearly all reinvestment went into the core: more warehouses, more depots, more gas stations. There were no large acquisitions. Second, the specials. When cash piles up, Costco pays it out in one go rather than buying stock at any price, which I respect. Third, the buybacks are small, and at an average of $950.90 a share in fiscal 2026 they bought stock at nearly twice my base value.³ At $848M that does little harm. But it is a hint that the board does not ask what the stock is worth before buying it.

The return on that capital has risen. On my measure, operating profit after tax over equity and debt less cash, it went from 19% in fiscal 2016 to 43% in 2026 C (calculated).⁶³ Part of that is the lower tax rate. Most of it is that suppliers and members now fund so much of the working capital. A business that earns over 30% on its capital and can reinvest at that rate for decades is rare. That is the bull case in one sentence. The problem is only that everyone knows it.

Cash is now $21.3B, the most in its history.³ The $4B buyback authorization runs out in January 2027 with $1.1B unused.³ The last three specials came three to four years apart. I don’t forecast one, because it changes nothing about value: it moves cash from the company to owners who already own it.

VFinancials

Costco’s accounts are simple and conservative. There is one class of stock, no large acquisitions, little debt, and the largest judgement in the numbers is the self-insurance reserve for health care and liability claims, which the company says has historically needed only small adjustments.³

Margins

The operating margin rose from 3.09% of revenue in fiscal 2016 to 3.85% in 2026 C (calculated).⁶³ That looks tiny. It is the right number to watch, though, because at these margins small moves are large. One tenth of a point of net sales in fiscal 2026 was $297M, or 2.5% of operating income C (calculated).³ The 10-K says the same: because the business runs on very low margins, modest changes in costs can have substantial effects on net income.³

In fiscal 2026 the gross margin fell three basis points, but rose ten without the effect of higher gasoline prices. SG&A fell ten basis points and rose one without gasoline.³ Core merchandise margins on their own sales rose 15 basis points, in every category.³ A LIFO charge of $152M in the fourth quarter, from memory chips in electronics and oil-linked goods, took 11 basis points off that quarter’s gross margin.⁹ In short, the year was steady underneath, with gasoline and inflation moving the reported numbers about.

Earnings quality

Net income grew 14.7% a year over the decade, faster than operating income at 12.3% C (calculated).⁶ The gap is mostly tax. The rate fell from 34.3% in fiscal 2016 to 24.7% in 2026, mainly because of the 2017 Tax Act C (calculated).⁴³ That was a one-time gift. I don’t assume another one.

Interest income was $601M in fiscal 2026, about 5% of pre-tax income, earned on the cash pile.³ If rates fall it falls too. Stock pay was $924M, about 10% of net income, and it is already counted as a cost.³ Costco issues new shares to cover vested units, and buys back roughly enough to keep the count flat.³

From profit to cash

Operating cash flow was $15.8B in fiscal 2026 against $9.2B of net income.³ The gap is depreciation of $2.7B and a $2.3B release of working capital: inventories rose $1.2B and payables rose $2.8B C (calculated).³ The 10-K credits faster inventory turns and better payment terms with suppliers.³

This is a feature of the model, and it deserves care. A retailer that sells its stock before paying for it gets cash from growth. As long as sales rise, suppliers fund more of the shelves. The day sales stop rising, that source stops too. So I count only a modest share of it in my base case, 1.5% of each year’s added net sales, about a fifth of the ratio implied by fiscal 2026 E (estimate).

What an owner could take out

Here is the number I care most about. If Costco stopped opening warehouses and simply kept the ones it has in good order, how much cash could it hand to its owners each year? I start from net income, take out the tariff refund, add back depreciation, and deduct what it would cost to stand still.

The answer is about $8.7B, or $19.5 a share E (estimate). That is about 6% below reported net income, so for Costco the accounting profit is a fair guide to owner cash. I can’t pin it down more tightly, because Costco doesn’t say how much of its capital spending keeps old buildings fresh and how much builds new ones. Reasonable splits give $8.3B to $9.1B E (estimate). Either way, an owner buying at $1,012.28 gets about 1.9% of the price back each year from today’s business, against 4.94% on a ten-year Treasury.²⁴ Everything above that has to come from growth.

The balance sheet

At 30 August 2026 Costco held $20.2B of cash and $1.1B of short-term investments, against $6.2B of debt and $4.0B of lease liabilities.³ $2.25B of notes fall due in fiscal 2027.³ It has signed $1.7B of leases that have not started.³ It doesn’t need the debt, and could repay all of it from cash.

Some of the cash isn’t free, though. Members have prepaid $3.0B of fees for the coming year, Executive members are owed $3.0B of rewards, and payables exceed inventory by $3.3B.³ That money is part of how the business runs. In my valuation I count cash less debt and finance leases, less the November dividend, as $13.1B of net cash C (calculated). If I held back $6B of it as working float, my values would fall by about $13 a share C (calculated).

VIValuation

Start with what would go wrong

Before any numbers, I ask what would most likely make this a bad investment from here, and work back from that. For Costco the business is not the most likely problem. The price is. Here are the three ways it goes badly wrong, in the order I think most likely.

1. The business does fine and the multiple falls. Costco earns about 2.3 times as much per share in ten years, as in my base case, but the market pays 25 times earnings instead of 50. The return on $1,012.28 would then be about 2.3% a year, with dividends, less than a Treasury E (estimate). Nothing has to break for this to happen. In October 2017 the market paid 26.5 times C (calculated).¹⁶
2. The promise on price slips. The model works only if members believe Costco is always cheapest. A management that took margin faster, or a fee rise that members resent, would show up first in the renewal rate. Each point of renewal is roughly 0.8M paid members and about $60M of fees, before the lost sales E (estimate).³
3. A rival with deeper pockets cuts prices. Costco’s operating margin is 3.85%. A one-point cut in gross margin to match a rival would cost about $3.0B, a quarter of operating income C (calculated).³ Walmart, which owns Sam’s Club, earned $29.8B of operating income last year and can afford a long fight.⁷

I can’t put honest odds on the second and third, so I don’t try. I can show what the first one does, and I do below.

What the price implies

On consensus, Costco earns $22.61 a share in fiscal 2027, up about 11%.²⁵ My base case lands on the same figure E (estimate). So I am not more cautious than the market about next year. The gap is about the ten years after it, and about what someone will pay in 2036.

Over ten years the price rose 6.85 times while earnings per share rose 3.90 times C (calculated).¹⁶ About three tenths of the gain, measured in logs, came from the market paying more for each dollar of profit C (calculated). That part of the record can’t repeat from here without the multiple going higher still.

Reverse DCF: what the price needs

Held at my base margins, the price needs net sales to grow about 20% a year for ten years, to about $1.9T of revenue in fiscal 2036 C (calculated). The last ten years gave 9.9%, with a pandemic, inflation and higher pump prices helping.⁶ Even with my bull margins it needs about 18% C (calculated). I don’t think any reasonable path gets there.

There are two other ways to read the same price. One is that it equals my base cash flows at a discount rate of about 6.2% C (calculated). That is the return the price offers if my base is right, 1.3 points above the Treasury. The other is to ask what a buyer in 2036 pays. If my base earnings come through and the market still pays 50 times them in 2036, the return from $1,012.28 is about 9.1% a year E (estimate). That is the real bet in this price: not a faster business, but a multiple that never comes down.

The discount rate

I discount at 8.5%. The capital asset pricing model with the 4.94% ten-year Treasury and Damodaran’s 4.20% implied equity risk premium gives 8.1% on Costco’s beta against the market over 119 months, 8.6% over 60 and 7.8% over 36 C (calculated).²⁴²⁶²⁷ On my Mastercard page I took the ten-year window. Here I take the highest of the three, because a business priced at 50 times earnings carries more risk from its multiple than its low beta shows. The betas use price returns with the three special dividends added back.²⁷

At 9.0%, the rate on my Visa and Mastercard pages, the base is $479; at 9.5%, the rate on my American Express page, $437 E (estimate). No row of the table reaches the price.

Is the base too harsh? A little, and I would rather say where. The 8.5% rate sits near the top of my three readings, 7.8% to 8.6%; at the 8.1% of the ten-year beta the base is $579. The base has no fee rise, though Costco has raised the fee twice in ten years, and 3.5 points of extra fee growth in each of fiscal 2030 and 2031 adds $14. Growth also steps down from 6.2% in 2036 to 3.5% after it, and at 4.0% the step is smaller. Put the three together and the base is about $636 E (estimate).³ The price is still 59% above that. The caution in my margin path matters as much: six basis points a year against ten in the last decade, and my bull case shows what ten would do. None of this is the difference between $530 and $1,012.

Three cases

My base case is Costco carrying on much as it has, a bit slower. Adjusted comparable sales of 6.5% in fiscal 2027, against 6.6% in 2026 and 7.6% in September, which a Labor Day shift lifted by a little over half a point, easing to 4% by the end E (estimate).³²⁸ Thirty net openings a year, management’s own target. A merchandise margin that keeps rising, but at about six basis points a year against ten in the last decade, because a company that shares its gains with members won’t keep them all E (estimate). That gives revenue of $640B and operating income of $27.1B in 2036, about 8.8% a year against 12.3% in the last ten C (calculated).

The bull case has Costco beat its own record for a decade from a much larger base: 33 openings a year, comparable sales near 7% for five years, and a merchandise margin of 3.0% by 2036 E (estimate). The bear case has membership growth keep fading, openings slow to 22 a year, and the margin give back some of its gains to defend price E (estimate). Even the bear is a good business. It is only a poor purchase at $1,012.28.

What I can’t judge

Some inputs I can’t know, and I would rather say so than pretend. I don’t know how much of the capital spending is upkeep; I show a range. I don’t know how much of the cost gains management will choose to keep; their whole philosophy says little. I don’t know whether younger, digital members will renew like their parents; the early numbers say a bit less. And I can’t know what multiple the market will pay in 2036. That last one matters more than all the others together, which is why I show the returns at several multiples instead of picking one.

Multiply it out

My base needs several things at once. Adjusted comparable sales averaging 5% or more for five years: I would give that 55%. About 30 net openings a year: 75%. A merchandise margin rising about six basis points a year: 60%. Renewal in the US and Canada staying at 90% or better: 85%. No lasting price war: 85%. Multiplied, that is about 18% E (estimate). They overlap, since good comparable sales help the margin and the renewal rate, so the true figure is higher, perhaps a third. And missing one doesn’t ruin the case. But the price needs more than my base. It needs my base and a multiple near today’s in 2036. If I give that a one-in-five chance, the joint odds are about 4% before overlap E (estimate). The argument for the price is not simple. It needs many things to go right together.

Quality and price are two questions

How good is the business? Among the best I have studied: high returns on capital, a habit that renews every year, a culture that turns size into lower prices, and almost no debt. How good is the price? Poor, on my numbers. At $1,012.28 the price is above my bull case. A great business can be a poor purchase at the wrong price, and on this evidence that is what this looks like.

It helps to compare it with the next best use of the same dollar. A ten-year Treasury pays 4.94% with no risk to the dollars promised.²⁴ Costco returns about 6.2% a year on my base, one and a quarter points more, with all the risk of a share. On this site, Visa returns 8.2% on that page’s base case, and American Express, which also lives on fees members pay up front, is priced level with my base value at a 9.5% rate. BJ’s, a smaller and weaker club with no page here, trades at 23 times earnings against Costco’s 50 C (calculated).¹⁸¹⁷¹⁹ Of the businesses I cover, Costco is one of the best and one of the most fully priced.

The margin of safety

My margin-of-safety price is $371, 30% below my base value. At $371 even my bear case returns about 7.8% a year, close to the 8.5% I require, and the base returns about 10.5% C (calculated). A 30% cushion is what I use for a business this predictable. It would be wider for one I understood less well. The price is 173% above it.

Where this could be in 5 and 10 years

My base value rises to $654 a share by August 2031 and $813 by August 2036 E (estimate). Add $260 of owner cash paid out along the way and the yearly return from today’s price is 0.7%. The bear is −5.3% a year and the bull 4.5%.

For the base to be right at five years, I need revenue growing about 8.6% a year to $457B and the operating margin near 4.0%, from 3.9% now. That gives owner cash of $9.7B. At ten years, the margin on goods has to reach 2.55%, from 1.94%. Revenue is then $640B and owner cash $15.1B.

The bear and the bull differ on three numbers. The bear has comparable sales fading to 2.5%, 22 net new warehouses a year and the margin on goods falling to 1.75%. The bull has 33 warehouses and 3.0%. The 8.5% rate matters as much, because most of each year-10 value is what the business is worth after 2036. That is why the range widens with time.

The most likely way my base is wrong is the margin on goods. Costco added about ten basis points a year for a decade, and I assume six. If it stalls at 1.9%, my year-10 base value is $699, not $813 E (estimate).

These are estimates of value, not forecasts of the share price. The market price can sit above or below value for years.

VIIThe case for the price

What most people miss

First, the fee is no longer most of the profit. It was 72% of operating income in fiscal 2016 and 51% in 2026, because the margin on goods rose from 0.95% to 1.94% while the gross margin stayed flat C (calculated).⁴³ Costco now earns real money on what it sells, from costs alone. Second, suppliers fund more than all of the stock: payables were 117% of inventory at the end of fiscal 2026, against 85% ten years earlier C (calculated).³⁴ Third, Executive members are half of paid members but 75.2% of sales, up from 73.6% in a year.³¹⁰ Each of those makes growth cheaper to fund than it looks.

The bull case

In plain words: Costco is a machine that turns size into lower prices, lower prices into loyalty, and loyalty into more size, and it can keep doing this for decades. The clearest written version is from Nick Sleep and Qais Zakaria of the Nomad Investment Partnership. In their December 2004 letter they called Costco the best example they could find of “scale efficiencies shared”. They added: “Most companies pursue scale efficiencies, but few share them.”²⁹ Their average cost then was about $39 a share, before the splits and specials since.²⁹ The partnership closed in 2014, so this is an old argument, but it is the argument.

Management’s case today is the same idea with numbers: 30 net new warehouses a year, a path for that pace for five to ten years, younger members who grow into bigger spenders, and AI search sending shoppers who convert well.⁹ I found few named investors with a current, public case. Of the funds I checked, Markel held 33,748 shares at 30 September 2026, down from 39,150 at 30 June, and Baron held 5,569 at 30 June.³⁰³¹ None of Altimeter, Atreides, Duquesne, Coatue, Tiger Global, Pershing Square, ARK, Akre or Fundsmith reported any at 30 June.³¹ Costco’s holders are mostly index funds.⁸

Against reality

Much of the bull case is true today. Openings, the vintage table, the member mix and the returns on capital all back it.³ My bull case gives it all of that: 33 openings a year, comparable sales near 7% for five years and a merchandise margin of 3.0% by 2036. That is worth $731 E (estimate). So if Costco does all that and hits its targets, it is a better business in 2036 than today by a wide margin. It still doesn’t reach $1,012.28 at 8.5%. The price needs net sales growth of about 20% a year at my base margins, against 9.9% in the last decade, or a buyer in 2036 who pays about 50 times earnings again C (calculated).

So the honest version of the bull case at this price is about the multiple, not the business. Costco has been priced higher each year for ten years partly because the market kept paying more: the multiple went from 27.8 to 49.5 C (calculated).¹⁶ That can continue. Nothing in the filings tells me it will.

What a 10x would take. Ten times $1,012.28 is $10,123 a share, about $4.5T of market value on today’s share count C (calculated). At 30 times earnings that needs about $150B of net income, 16 times fiscal 2026. At a 3.5% net margin, above the 3.0% of 2026, it needs about $4.3T of revenue C (calculated). Walmart had $713.2B last year.⁷ To do it in ten years Costco would need revenue growth near 30% a year with no rise in the multiple. Costco itself rose 6.8 times in the last ten years from a $65B company, a very good decade for a company that size C (calculated).¹ From $450B, tenfold is close to impossible, and I would not build a case on it.

Where I come down

The vintage table and the supplier terms moved me: growth here costs less capital than I assumed, and that is why my base has returns on new capital of 25% for ever. The bull case didn’t move me on value, because even my bull falls short of the price. What would raise my values most is the merchandise margin rising faster than six basis points a year with renewal rates rising too, which would show the cost gains and member loyalty growing together.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• California and labour costs. California is 26% of US net sales, and wages and benefits are the largest cost after goods.³ The company raised the top of its wage scales by $1 an hour in March 2026.³ The CFO named health care and general liability as costs growing faster than sales.⁹
• Membership quality. Online sign-ups and new international members renew at a slightly lower rate.³ If the mix keeps shifting that way, the renewal rate keeps easing even when nothing is wrong.
• Gasoline. It is 11% of net sales and a reason to visit.³ Electric cars cut both, slowly. Higher pump prices flatter sales, and a fall would do the opposite.³
• Tariffs and inflation. The company paid about $500M of tariffs it is now getting back, and says higher tariffs are more likely to hurt than help.³ The fourth-quarter LIFO charge showed inflation in electronics and oil-linked goods.⁹
• Pharmacy policy. Pharmacy grew nearly 20%, helped by weight-loss drugs, but Medicare’s new maximum prices lower revenue per prescription, and an analyst on the call asked about the next round of cuts.⁹
• Leadership and culture. The culture is the asset. Two of the most senior roles changed hands in 2026, and the next generation has to keep the habit of giving gains away when the market pays for keeping them.³
• Legal claims. The 10-K lists wage and hour class actions in California, among others, with no estimate of possible loss.³

What would raise my base value: the fiscal 2027 first quarter, expected in December, showing adjusted comparable sales of 6% or more, renewal rates rising again, and the merchandise margin above 2.0%. What would lower it: paid member growth below 3%, a renewal rate below 92% in the US and Canada, or gross margin given up to match a rival.³