Costco, two engines
The question is not whether $902 is cheap or expensive. It is: if you buy at 45 times earnings, which assumptions are you paying for? Believers buy, sceptics pass, but either way you should know what you are believing.
The approach: one company, two completely different businesses. Membership fees behave like an annuity; merchandise is the traffic engine. Look at the economics of each on its own, then put them back together and ask: which one does the 45× multiple buy, and which year is the market pricing?
Disclaimer · This material is for personal research and educational discussion only. It is compiled from public information and does not constitute investment advice, an offer, or a recommendation, nor a basis for buying or selling any security. The presenter holds no position in COST but is a paying Costco member. Investing involves risk; make your own judgement and bear the consequences of your own decisions.
One warehouse, two engines
Founded 1983 · HQ Issaquah, Washington · 931 warehouses · 341,000 employees · FY25 total revenue $275.2B · net income $8.1B
How the two engines relate: merchandise is sold close to cost so that members feel the fee is worth it; members who feel it is worth it renew, and the renewal rate is the merchandise engine's only KPI. So you cannot judge Costco by its gross margin. The low margin is the design, not the problem.
From $1,094 to $902, back to last September's price
| Date | Event | Close | Next-day move |
|---|---|---|---|
| 2025-09-25 | Q4 FY25 earnings | 943 | −2.9% |
| 2025-12-11 | Q1 FY26 earnings | 884 | 0.0% |
| 2025-12-22 | One-year low | 850 | — |
| 2026-03-05 | Q2 FY26 earnings | 983 | +1.6% |
| 2026-05-19 | One-year high | 1,094 | — |
| 2026-05-28 | Q3 FY26 earnings | 995 | −3.9% |
| 2026-09-24 | Q4 FY26 earnings (full year) | — | TBD |
- Down 17.7% from the high, but only 6% over a year, and up 4.9% year to date. This is not a crash. It is giving back the run to $1,094 in May.
- After the Q3 report on 28 May the stock drifted from $995 to $902. Only one day in that stretch fell more than 4% (7/9, −4.2%); the rest was a slow grind.
- Sell side, 59 analysts: 39 buy / 19 hold / 1 sell. Consensus target $1,102, range $1,000–1,275. Even the lowest target is above the current price, which is a sentiment reading, not a margin of safety.
- Options implied volatility 24.4%, 30-day realised 22.1%, IV at the 67th percentile of its 52-week range. Two weeks before earnings, premium is not cheap.
Revenue +11.6%, fees +10.7%, gross margin −21 bps
Seven lines to watch: revenue, membership fees, merchandise gross margin, SG&A, net income, comparable sales (ex gas and FX), e-commerce.
| Q1'25 | Q2'25 | Q3'25 | Q4'25* | Q1'26 | Q2'26 | Q3'26 | Q3 YoY | |
|---|---|---|---|---|---|---|---|---|
| Total revenue $M | 62,151 | 63,723 | 63,205 | 86,156 | 67,307 | 69,597 | 70,527 | +11.6% |
| Membership fees $M | 1,166 | 1,193 | 1,240 | 1,724 | 1,329 | 1,355 | 1,373 | +10.7% |
| Merchandise gross margin | 11.27% | 10.85% | 11.25% | 11.13% | 11.32% | 11.02% | 11.04% | −21 bps |
| SG&A ÷ net sales | 9.59% | 9.06% | 9.16% | 9.21% | 9.60% | 9.19% | 8.96% | −20 bps |
| Net income $M | 1,798 | 1,788 | 1,903 | 2,610 | 2,001 | 2,035 | 2,192 | +15.2% |
| Diluted EPS | 4.04 | 4.02 | 4.28 | 5.87 | 4.50 | 4.58 | 4.93 | +15.2% |
| Comparable sales · ex gas and FX | — | — | — | 6.4% | 6.4% | 6.7% | 6.6% | US 6.8 · Canada 6.2 · international 5.9 |
| E-commerce · ex FX | — | — | — | 13.5% | 20.5% | 21.7% | 20.8% | Site and app traffic +37% |
* Q4 is a 16-week quarter. Merchandise gross margin = (net sales − cost of goods) ÷ net sales, derived from the press-release figures and consistent with the call (Q3 11.04%).
Q3 call highlights: gross margin ex gas +1 bp, core-on-core −9 bps (fresh and food slightly lower); SG&A improvement mostly from leverage; traffic +2.4%, ticket +4.2% (ex gas and FX); about a quarter of the fee growth comes from the September 2024 fee increase; a tariff refund has been filed with customs and is expected within 2–3 months (one-off).
82.9 million people paying $70 a year, 92% of them renewing
- This is an annuity. No cost of goods, deferred and recognised over 12 months, and 92% of 82.9 million people will pay again next year. FY25 fees of $5.3B equal 51% of operating income.
- The fee-increase cycle is a hidden second source of growth. US and Canada fees were raised in 2011-11, 2017-06 and 2024-09, 5.5 to 7 years apart. The latest took Gold Star from $60 to $65 and Executive from $120 to $130, and lifted the Executive reward cap from $1,000 to $1,250.
- The current increase accounts for about a quarter of fee growth today and will have mostly lapped by Q4 FY26, the 9/24 report. From FY27, fee growth comes down to two factors: member count (+4%) × Executive upgrades (+9.6%).
- Numbers to watch: renewal rate (every quarterly call), Executive share, and fee per paid member (whether $70 holds once the increase has lapped).
A $270 billion business that earns 1.9%, by design
- The markup cap is self-imposed: management has said for years that branded goods are marked up no more than 14% and Kirkland no more than 15%. The profit given up goes to members in exchange for renewals.
- Comparable sales 6.6% = traffic 2.4% × ticket 4.2% (Q3, ex gas and FX). Traffic means more people coming; ticket means each visit buys more. Both rising says the growth is not being propped up by price.
- E-commerce +20.8%, about 7% of net sales (about 10% including digitally enabled transactions). Site and app traffic +37%. A small base, but the right direction.
- Gas is about 10% of net sales, and fuel prices distort reported comps, so always read comps ex gas and FX.
- Openings: FY26 target 26, with 14 more (US and Canada) in October and November. Openings are the only growth that needs capital: capex of about $6.5B, 40–50% of operating cash flow.
The merchandise engine's margin is deliberately held at around 2%: the gross margin cap is fixed, SG&A dilutes with scale, and any surplus is handed back to members as lower prices. So the merchandise engine's output is not profit, it is renewals and traffic, and the fee engine turns those into profit.
That is why "Costco's gross margin fell" is almost never bad news, while "the renewal rate fell" is.
Revenue 9% compounded, EPS 13% compounded, share count unchanged
| Fiscal year | Revenue $B | Net income $B | Diluted EPS | ROIC | Year-end P/E |
|---|---|---|---|---|---|
| FY15 | 116.2 | 2.38 | 5.37 | 13.6% | 25.9× |
| FY18 | 141.6 | 3.13 | 7.09 | 15.3% | 32.6× |
| FY20 | 166.8 | 4.00 | 9.02 | 13.2% | 38.4× |
| FY22 | 227.0 | 5.84 | 13.14 | 18.0% | 40.3× |
| FY24 | 254.5 | 7.37 | 16.56 | 20.2% | 53.8× |
| FY25 | 275.2 | 8.10 | 18.21 | 19.1% | 51.7× |
| TTM | 293.6 | 8.84 | 19.88 | 19.0% | 45.4× |
- Ten-year CAGR: revenue 9.0%, net income 13.0%, EPS 13.0%. Diluted shares went from 443M to 444M, so almost no buybacks and no dilution: all of the EPS growth came from earnings.
- ROIC rose from 13.6% to 19%. SG&A diluting with scale plus a rising fee share is where the slow margin expansion comes from (operating margin 3.1% → 3.8%).
- Net cash $10.7B (Q3 FY26, including lease liabilities). Capital returns are a $1.47 quarterly dividend (0.6% yield) plus occasional specials: $15 per share in 2024-01, and $10 per share in 2020-12 before that.
- Working capital is close to zero: inventory turns in 28 days, payables in 32, so suppliers are funding it. Cash conversion cycle 0.5 days.
45×: down from 54×, still above the ten-year 75th percentile
| FY15 year-end → now | Start | End | Multiple |
|---|---|---|---|
| Share price | $139 | $902 | 6.5× |
| EPS (FY15 → TTM) | $5.37 | $19.88 | 3.7× |
| P/E | 25.9× | 45.4× | 1.75× |
| 6.5 ≈ 3.7 × 1.75. Of the eleven-year gain, earnings contributed about two thirds and multiple expansion about one third. | |||
- The pullback removed the FY24 leg of multiple expansion (53.8× → 45.4×), not the ten-year norm. The ten-year range is 25.7–53.8×, median 36.9×, 75th percentile 40.3×.
- With EPS unchanged, a return to 40× means another −12%; to the 37× median, −19%; to 35×, −23%. Multiple compression on its own is the biggest source of downside.
- The other way round: in twelve fiscal years only FY14–FY17 sat near or below 30×, and it has never gone back. The market's premium for Costco is not a recent thing.
- So the honest answer to "has the pullback gone far enough" is: the price is back to a year ago, the valuation is not back to a decade ago. It is cheap relative to May, not relative to history.
BJ's, the same model, trades at 19×; Costco at 45×
| Company | Market cap $B | P/E | EV/EBITDA | ROIC | Latest FY revenue growth | Membership model |
|---|---|---|---|---|---|---|
| COST Costco | 401 | 45.4× | 27.0× | 19.0% | +8.2% | Yes · fees are 51% of operating income |
| WMT Walmart | 845 | 38.3× | 21.2× | 12.7% | +4.7% | Partly · Sam's Club + Walmart+ |
| BJ BJ's Wholesale | 11.5 | 19.3× | 12.1× | 11.9% | +4.7% | Yes · same model, regional in the eastern US |
| PSMT PriceSmart | 5.3 | 32.5× | 15.7× | 10.3% | +7.2% | Yes · Latin America and the Caribbean |
| AMZN Amazon | 2,753 | 20.3× | 11.4× | 8.5% | +12.4% | Prime · a different kind of membership fee |
| TGT Target | 71 | 16.2× | 9.9× | 11.3% | −1.7% | No |
| DG Dollar General | 27.5 | 16.2× | 12.8× | 7.0% | +5.2% | No |
| NASDAQ consumer staples sector P/E 34.0× (2026-09-10). P/E = 1 ÷ earnings yield; EV/EBITDA and ROIC are FMP TTM figures; growth is each company's latest full fiscal year. | ||||||
- Among the three same-model companies, Costco's ROIC is 1.6× BJ's, its growth 1.7×, and its P/E 2.4×. Half the premium is explained by quality; the other half is the price the market puts on certainty.
- Walmart at 38× is the closest reference: also priced as a "growth stock" within consumer staples. Both are dearer than the sector's 34×.
- The peer table is a coordinate system, not something to add up: nothing among the peers has a fee engine of comparable scale.
$902 = 40× FY27 consensus, or 36× FY28 consensus
| Consensus EPS | FY26E | FY27E | FY28E | FY30E |
|---|---|---|---|---|
| EPS | $20.52 | $22.67 | $24.95 | $28.49 |
| YoY | +12.7% | +10.5% | +10.1% | CAGR 8.5% |
| Price ÷ EPS | 44.0× | 39.8× | 36.2× | 31.7× |
| Analysts | 24 | 27 | 21 | 3 |
| Reverse DCF · discount rate | 8% | 9% | 10% |
|---|---|---|---|
| Ten-year FCF growth implied by the market cap | 12.7% | 15.4% | 17.9% |
| Starting from TTM free cash flow of $8.8B, enterprise value $390.2B, terminal growth 3%. Discount rate = 10-year Treasury 4.95% + a 3–5% risk premium. | |||
- On consensus multiples: the current price sits between "FY27 consensus × 40" and "FY28 consensus × 36". The market is paying today's ten-year 75th-percentile multiple for profits one to two years out.
- On cash flow: to justify a $390.2B enterprise value, free cash flow has to grow 13–18% a year for the next ten years. Net income compounded at 13% over the past ten; sell-side consensus EPS for FY26–30 compounds at only 8.5%.
- Both lines say the same thing: buying at $902 is a bet that the past decade's growth repeats for another decade, and that the multiple does not compress. Consensus itself does not support the first part.
Three verifiable assumptions, and what would falsify them
① Renewal rates hold at 92% / 90%
The principal of the annuity. FY25 year-end 92.3% / 89.8%, Q3 FY26 92.2% / 89.7%. Not slipping a year after the fee increase is the single most important piece of evidence in this cycle.
Watch: US and Canada and worldwide renewal rates disclosed on each quarterly call.
② The member mix keeps upgrading
Paid members +4%, Executive +9.6%, Executive at 75% of sales. Once the fee increase has lapped, these two factors have to carry fee growth of 7–8%.
Watch: paid-member growth, Executive share, fee per paid member.
③ Comparable sales stay at 5–7% (ex gas and FX)
The output of the merchandise engine. The last four quarters: 6.4 / 6.4 / 6.7 / 6.6%, with traffic and ticket both positive. This number leads the renewal rate.
Watch: comps, traffic and ticket together every quarter.
If all three hold, Costco is a machine that grows EPS 10–13% a year, and whether 45× holds depends on rates and sentiment. If any one of them breaks, the multiple will de-rate before earnings do. That is what 45× buys.
The biggest risk is the valuation, not the business
| Risk | Number | What to watch next quarter |
|---|---|---|
| Multiple compression | 45.4× → 40× = −12% · → 35× = −23%, with EPS unchanged | Rates, sector rotation; can happen with no company-level trigger |
| Merchandise gross margin | Q3 core-on-core −9 bps, fresh and food slightly lower | Whether Q4 is a second consecutive weaker quarter; whether management frames it as investment in price |
| One-off tariff refund | Filed, expected within 2–3 months (Q4 FY26 / Q1 FY27) | Amount and how it is booked; EPS growth excluding it |
| Fee-increase benefit fading | About a quarter of fee growth today, lapped by Q4 FY26 | Whether Q1 FY27 fee growth holds at 7–8% |
| Opening pace | FY26 guidance 28 → 26, two pushed to FY27 | FY27 opening and capex guidance |
| Consumer narrative | 9/8 Cramer: "Costco has lost its edge, go to the dollar stores" · 9/9 the company announced 14 more openings in Oct–Nov | Whether Q4 traffic is still positive; any data on lower-income members leaving |
| Crowded sell side | 39 buy / 19 hold / 1 sell · lowest target $1,000 still above the price | How many targets get cut after the report |
| Insiders · Congress | No open-market buys in 12 months; sales all ≤2,400 shares each, mostly tax withholding and gifts. Both chambers of Congress have bought and sold in the past year, each trade ≤$50k | Not a signal, noted for the record |
- Whether Q4 fee growth is still ≥10% (a 16-week quarter with the last of the fee-increase benefit); full-year fees around $5.7–5.8B
- Paid members, Executive members, renewal rate, on a full-year basis, against FY25 year-end 81.0M / 38.7M / 92.3%
- Q4 comps ex gas and FX, and the traffic / ticket split
- Merchandise gross margin core-on-core, whether negative for a second consecutive quarter
- FY27 opening count and capex guidance; the tariff refund amount
- Anything said about fees, even just "no plans"
- Consensus Q4 EPS $6.55 (vs $5.87 a year ago, +11.6%), revenue $94.8B.
- The typical next-day move after earnings is ±3–4% (last four: −2.9 / 0.0 / +1.6 / −3.9%).
- This was prepared before the report, so every number above will be updated within two weeks. The direction of the update matters more than the numbers themselves.
Where every number comes from
Primary sources
· 10-K FY2025 (2025-10-08): paid members 81.0M, cardholders 145.2M, Executive 38.7M at 73.6% of sales, renewal 92.3% / 89.8%, 914 warehouses, 341,000 employees, gas about 10%, e-commerce about 7%
· 8-K quarterly press releases: 2025-09-25 (Q4 FY25), 2025-12-11 (Q1 FY26), 2026-03-05 (Q2), 2026-05-28 (Q3): revenue, fees, net income, EPS, comps, warehouse count
· Q3 FY26 call transcript (Motley Fool, 2026-05-28): members 82.9M, Executive 41.2M / 75%, renewal 92.2% / 89.7%, gross margin and SG&A breakdown, traffic and ticket, opening and capex guidance, tariff refund
· 8-K 2026-07-08: quarterly dividend $1.47
· Price: IBKR real-time 2026-09-11 09:43 ET · daily prices, financial statements, peer metrics, sell-side consensus, insider and congressional trades, sector P/E: FMP
· Treasury yields: FMP treasury rates 2026-09-10 (10-year 4.95%)
Secondary sources
· Fee-increase history (2011-11, 2017-06, 2024-09 and amounts): company announcement of 2024-07-10 and public reporting
· Markup caps of 14% / 15%: management's public statements over the years, not a figure from this report
· 2020-12 special dividend of $10: company announcement
· Cramer's comments and the 14 new warehouses: as reported by 24/7 Wall St (9/8) and Fox Business (9/9)
· Peer gross margins (Walmart about 25%, Target about 28%, BJ's about 18%): approximate figures from each company's latest fiscal year, not pulled individually for this piece
Derived figures
· TTM = sum of the last four quarters (Q4 FY25 to Q3 FY26)
· Merchandise gross margin = (total revenue − fees − cost of goods) ÷ (total revenue − fees)
· Merchandise-only operating income = operating income − fees (treating fees as 100% profit, ignoring member-service costs)
· Year-end P/E = 1 ÷ FMP earnings yield (year-end market cap ÷ that year's net income)
· Return attribution: price multiple ≈ EPS multiple × P/E multiple
· Reverse DCF: ten years of constant growth + 3% terminal, solving for the growth rate at which present value = enterprise value
· Scenario table = sell-side consensus EPS × 35 / 40 / 45
· Fee per paid member = TTM fees ÷ paid members
A note on method
· Every multiple is a judgement. The ranges show where history and peers sit, not what the number should be.
· No price target is given.
· Not obtained: 13F institutional holdings, the original call transcript (not available on the FMP Starter tier; the transcript is from a third party), monthly sales data.
· Prepared before the 9/24 report; all figures should be updated from the report.
Disclaimer · This material is for personal research and educational discussion only. It is compiled from public information and does not constitute investment advice, an offer, or a recommendation, nor a basis for buying or selling any security. The presenter holds no position in COST but is a paying Costco member. Investing involves risk; make your own judgement and bear the consequences of your own decisions.