Core-on-Core Snaps Back, the Multiple Finally Cracks, and the Member Engine Slows a Fifth Straight Quarter

Results vs. Consensus

Q4 FY2026 Scorecard (16 weeks ended August 30, 2026)

Consensus EPS is shown at $6.54, the most widely cited figure; the cross-source range was $6.52 to $6.55. Revenue consensus ranged $94.85B to $94.97B. Costco does not issue revenue or EPS guidance, so there is no company guide to score against. Every actual above traces to the 8-K Exhibit 99.1 income statement rather than to figures quoted on the call.

Year-Over-Year Comparison

Fiscal Year 2026 vs. Fiscal Year 2025 (52 weeks each)

The four reported quarters of FY26 total revenue ($67,307M, $69,597M, $70,527M, $95,723M) sum to $303,154M, matching the fiscal-year income statement to the dollar; the four quarters of diluted EPS ($4.50, $4.58, $4.93, $6.75) sum to $20.76. Free cash flow is derived from the cash-flow statement as operating cash flow less additions to property and equipment.

Quarter-Over-Quarter Comparison (rates only)

Costco's fiscal calendar runs 12/12/12/16 weeks, so sequential dollar comparisons between FQ3 and FQ4 are not meaningful. The rate metrics are, and they are where the quarter's story lives.

Quality of the Beat

• Revenue: Gasoline price inflation added roughly 3% to the comp and FX subtracted roughly 0.3%, which is the whole gap between the +9.4% reported comp and the +6.7% adjusted figure. Excluding gas sales entirely and adjusting for FX, comps were also +6.7%, so the adjusted number is not an artifact of which gas adjustment you pick. Non-foods was the strongest of the three merchandise categories.
• Margins: The reported 11bps gross-margin decline is a mix outcome, not a pricing outcome. Core-on-core rose 18bps ex-tariff items with all three merchandise categories higher; ancillary margin rose 23bps; LIFO cost 11bps; and the tariff item added 9bps in the "other" line. Reported margin fell only because gas, e-commerce and pharmacy grew faster than core merchandise. Ex-gas inflation, the reported rate was up 20bps.
• EPS: Quality is good but not pristine. Below the operating line, interest income rose to $209M from $169M on a cash balance that grew $6.0B during the year, and the tax rate fell 42bps to 25.2%. Those two together are worth a few cents. The operating line did the work: operating income grew 13.8% against net sales growth of 11.2%, which is genuine leverage.

Segment and Category Performance

Costco does not report profit by segment or merchandise category. The disclosures that matter are the comparable-sales geography split, the merchandise-category comps, and the gross-margin matrix components, all of which the company provides.

Comparable Sales by Geography

U.S. — the gas engine, and the quarter's dispersion

The U.S. carried the quarter on both bases, and the 350bp gap between its reported and adjusted comps is the largest of the three geographies, which is simply where the gas stations are. Record fuel volumes and an all-time-high penetration of U.S. member households buying gas are the mechanical driver. What sits underneath is a clean 7.2% adjusted number, the best of any region.

Assessment: U.S. adjusted comps accelerating to 7.2% from a 6.6% full-year pace is the single most encouraging demand datapoint in the release, and it is the reason the traffic re-acceleration is credible rather than a rounding artifact.

Canada — the laggard, and the one to watch

Canada's 4.6% adjusted comp is well below both its own 6.7% full-year figure and the company's 6.7% total. This is a market with 115 warehouses and five more planned for FY27, so it is not a footnote. Management offered no explanation for the deceleration and no analyst asked for one.

Assessment: An unexplained 210bp sequential-to-annual gap in the second-largest market is the kind of thing that is either macro noise or the start of something. We flag it as an open question for FQ1 rather than a conclusion.

Other International — reported strength, adjusted softness

Other International reported +7.0% but adjusted to +6.2%, and the full-year spread is wider still (+9.8% reported against +6.5% adjusted), which is FX flattering the headline. The forward story here is the pipeline: four European openings in FY27, and a stated FY28 pipeline in Asia and Australia.

Assessment: The international unit-growth runway is intact and arguably strengthening, but the segment is not currently outgrowing the U.S. on a like-for-like basis. The runway is a decade story, not a next-four-quarters story.

Merchandise Categories and Ancillary

Assessment: The category mix inverted versus last quarter. At FQ3 the strength was in ancillary and the discretionary "treat" basket, with core food steady. This quarter non-foods was explicitly the highest-performing category overall. Management tied it directly to assortment recovery after tariff-driven SKU removal earlier in the fiscal year. That reads as a real, if modest, cyclical unlock rather than a mix accident.

Gross Margin Matrix — Q4 FY2026

Assessment: This is the most informative table in the release and the one that changes our thesis. The core-on-core line is the only measure of whether Costco is earning more on the merchandise it actually sells, stripped of the mix distortion created by fast-growing low-margin ancillary businesses. It went from minus 9bps to plus 18bps in one quarter, and it did so with all three merchandise categories contributing. The drivers management named, supply-chain efficiency across all categories and labor productivity in meat, bakery and deli, are the structural kind rather than the price-and-pray kind.

Membership Metrics

Executive penetration of 50.3% is derived as 42.3M executive members divided by 84.1M total paid members; management characterized it as an all-time high without giving the percentage. Sequential FQ3 levels are as disclosed on the FQ3 call. Membership fee income is not comparable sequentially in dollars because FQ3 covers 12 weeks and FQ4 covers 16.

Assessment: The membership table tells two stories that point in opposite directions, and the market has to decide which one is the signal. The count is decelerating, and the fee line is decelerating with it once the September 2024 increase is stripped out. The quality is improving on every metric management tracks: executive penetration is at a record and just crossed half the paid base, both renewal rates ticked up, and the under-40 cohort has grown roughly 60% since COVID to more than a quarter of the base. The deferred-fee balance growing 5.3% against 7.3% of recognized fee income is the quiet number, and it points to fee growth stepping down again in FY27.

Key Topics & Management Commentary

Overall Management Tone: Confident and noticeably more relaxed than at FQ3, where the posture was defensive about a single margin line. This call opened with a fiscal-year victory lap and spent most of its energy on forward capacity rather than backward explanation, including an unprompted joke about the Food Court churro returning. The one place the tone tightened was membership growth, where three separate analysts pressed and management answered every time by redirecting from the count to the quality, never once offering a floor or a timeline for re-acceleration.

1. Core-on-core repairs to +18bps, and it is broad-based

The metric we marked broken in May came back with all three merchandise categories contributing. Management drew a careful distinction between the core component measured against total company sales (down 32bps reported, down 9bps ex-gas inflation, both mix outcomes) and core-on-core measured against core sales, which is the pricing-and-productivity read.

"In terms of core margins on their own sales, our core on core margins were higher by 18 basis points, excluding the impact of tariff refunds and tariff refund reinvestments. The increase in core on core margins was broad-based with fresh, non-foods, and food and sundries all higher year-over-year. Supply chain efficiencies drove margin improvement across all categories. Fresh also benefited from higher labor productivity in meat, bakery and deli." — Gary Millerchip, CFO

The phrasing matters. Last quarter management would not commit to forward expansion and steered every question toward the holistic ex-gas rate, which we read at the time as a team managing expectations down on a line it no longer expected to grow. The FQ4 framing is the opposite: an explicit number, an explicit breadth claim, and two named structural drivers.

Assessment: Our FQ3 read was wrong, and this is the cleanest refutation available. The minus 9bps was the funded price investment management said it was, enabled by the LIFO comparison, and once that investment lapped the underlying expansion reappeared. We are restoring this pillar. The caveat is that management still offered no forward gross-margin framework at all this quarter, and did not repeat even its FQ3 description of the ex-gas rate over the prior 12 to 24 months as "stable with a slight improvement," so the pillar is restored on evidence rather than on guidance.

2. The membership-fee tailwind formally ends

The September 2024 U.S. and Canada fee increase, which contributed roughly half of fee-income growth at its peak and a little more than a quarter at FQ3, fell to under 1% of growth this quarter. Management stated plainly that Q4 was the final quarter carrying any year-over-year benefit from it.

"The September 2024 U.S. and Canada membership fee increase accounted for less than 1% of fee growth. As a reminder, Q4 marks the last quarter in which we will see a year-over-year benefit from the membership fee increase. Excluding the membership fee increase in FX, membership income grew 6.8% year-over-year." — Gary Millerchip, CFO

What remains is a 6.8% underlying annuity driven by base growth and executive mix. The balance sheet suggests even that is optimistic for FY27: deferred membership fees, which is the unrecognized balance that converts into future fee income, grew 5.3% to $3,006M against a member count growing 3.8% and a recognized fee line growing 7.3%.

Assessment: The single largest re-rating catalyst of the past two years is now arithmetically finished, exactly on the schedule we laid out at FQ3. From FQ1 FY27 the fee line grows at whatever the base and the executive mix deliver, and the deferred balance says that is closer to 5-6% than 7%. The only lever that changes this is another fee increase, which was not raised on the call by anyone.

3. Member growth slows a fifth straight quarter, and management reframes the run-rate

Paid members grew 3.8% to 84.1M, against 4.1% at FQ3 and 4.8% at FQ2. The characterization of eight consecutive quarters of deceleration was put to management directly on the call and went uncontested, although on the reported figures the run is five, after growth held at 6.8% in both Q2 and Q3 FY25. The answer was a reframe rather than a rebuttal.

"I think the sort of the growth rate that we've seen in recent quarters is probably more typical of what we'd expect to see based on where we are today with the growth in new members that we're seeing and the sort of normalization of the renewal rate." — Gary Millerchip, CFO

Management named two things that could re-accelerate the headline: rising executive penetration lifting the blended renewal rate over time, and outsized new-market openings in Asia. Neither is in the FY27 plan, whose named geographies are four in Europe, five in Canada and one in Mexico. The Asia and Australia pipeline was pushed to FY28.

Assessment: At FQ3 management described 4% to 5% as "a more normal rate of growth." This quarter the print is below that band and the language shifted to describing whatever is happening as typical. That is not a floor, it is a description. The top of the comp funnel is the least-defended part of this story, and it is the reason we are not upgrading on the margin repair alone.

4. Tariff refunds: $184M in, most of it going back out

Costco received $184M of IEEPA tariff refunds in the quarter, comprising $174M of refunds and $10M of interest, and characterized it as a little more than one third of the total expected. That implies aggregate refunds somewhere in the region of $500M to $550M. A similar amount had already arrived in the first quarter of FY27 as of the call date.

"We received $184 million in the fourth quarter, which was made up of $174 million in refunds and $10 million in interest. This amount represents a little more than 1/3 of the total refunds expected. In the first quarter of fiscal year 2027, we have already received a similar amount of refunds as we did in Q4, and we intend to continue reinvesting the majority of the dollars we receive in increased member values." — Gary Millerchip, CFO

The net retained benefit was 9bps of gross margin, which on $93,873M of net sales is roughly $85M, plus the $10M of interest. Against $174M of refunds received, that implies roughly $90M went straight back out the door in price reductions during the second half of the quarter, spread across produce, meat, beverages, home furnishings and hardware.

Assessment: Handled the way a shareholder should want it handled, and disclosed the way an analyst should want it disclosed. Management committed to reporting the net quarterly impact separately for as long as it runs, which removes the modeling ambiguity. Treat it as a member-value program with an earnings by-product, not as an earnings event: roughly half of what comes in is being given away, and the whole program terminates once the remaining two thirds are collected.

5. The $152M LIFO charge, and what non-foods inflation is actually doing

LIFO cost 11bps of gross margin, with a $152M charge against $43M last year. The full-year charge was slightly north of $200M. The apparent contradiction with management's characterization of inflation as running "in the low single digits" and "relatively stable overall" was pressed in Q&A and explained as a fiscal-year-end true-up artifact.

"It might have seemed a bit contradictory for us to say inflation is relatively consistent, but we had a large LIFO charge during the quarter... Normally, if it was fairly consistent throughout the year, you would spread the charge over each quarter. But because the inflation I mentioned in consumer electronics and in the gas and petroleum-related items happened in the fourth quarter... While we took a $150 million or so charge in the fourth quarter, if you look at the total charge for LIFO for the year, it is slightly north of $200 million. Which if you divide that by the total inventory that we have on hand in the U.S., that would be about 1.5% inflation." — Gary Millerchip, CFO

The named drivers were memory costs in consumer electronics and oil-linked items including gas, motor oil and resins. Beef inflation persists, offset by deflation in eggs, dairy, butter and cheese. Steel, flour and resin were named as additional inflationary commodities.

Assessment: The explanation is arithmetically sound and the 1.5% life-of-inventory figure is a reasonable way to size it. What the explanation does not do is make the inflation go away. Memory pricing and petroleum are the two least forecastable inputs in the basket right now, and both were inflecting upward at the fiscal year end rather than at its start. This is the component of the cost-overhang risk that got worse this quarter, and it will show up in FY27 gross margin if those two curves keep rising.

6. SG&A leverage returns, with healthcare and general liability named as the watch items

Reported SG&A improved 27bps to 8.94%, with the operations component 22bps better and central 5bps better. Excluding gas inflation the total was 2bps better and operations was flat. At FQ3 the operations component had actually deleveraged 3bps ex-gas on healthcare costs and one-time legal items, so this is a reversal of a second FQ3 negative.

"I wouldn't say that there is anything that we'd call out today that would cause us to think there's some major sort of headwind that would impact our ability if we're comping at the rate that you mentioned to be able to create some leverage in the model. I think the one area that we have called out previously in other quarters would be general liability and healthcare as costs that have been growing at a faster rate." — Gary Millerchip, CFO

Management added a detail worth keeping: the employee wage agreements run on a three-year cycle in which year one is the most expensive, and the company is now into year two. That mechanically eases the labor-cost comparison against year one.

Assessment: A direct commitment that a roughly 6% comp produces SG&A leverage, delivered in a quarter that demonstrated it. The wage-agreement cycle detail is new and helpful: it means the labor line gets easier before it gets harder. Healthcare and general liability remain the genuine structural pressures, and management was explicit that a deterioration there would break the leverage commitment.

7. Capex steps to $7.5B and the opening cadence accelerates to 33

FY26 capex was $6.4B. The FY27 plan is approximately $7.5B, a 17% step-up, driven by a growing warehouse pipeline and continued supply-chain investment. FY26 delivered 28 openings including three relocations for 25 net new, slightly below the 26 net guided at FQ3. FY27 plans 33 openings including five relocations, for 28 net new against a long-term target of 30.

"For FY 2027, we are planning approximately $7.5 billion in capital expenditure. This increase is predominantly due to a growth in the pipeline of new warehouses as we target a run rate of 30 net new openings a year... Beyond FY 2027, we would expect to see a slowing in the rate of capital expenditure growth following three years of outsized growth that began in FY 2025." — Gary Millerchip, CFO

Named FY27 geographies are four in Europe, five in Canada and one in Mexico, with new U.S. markets at Buffalo, New York and Lawrence, Kansas. The FY28 pipeline was described as Asia, Australia and other international.

Assessment: The most genuinely bullish disclosure on the call, and the least discussed. Three things land at once: the opening cadence finally accelerates toward the 30-net target after two years of slippage, the capital commitment is sized to deliver it, and management put an explicit end-date on the elevated-capex phase. At $7.5B on roughly $320B of net sales, capex is about 2.3% of sales; the statement that growth slows after FY27 is the first forward free-cash-flow signal this management has given.

8. Third-party delivery goes national, and the younger-member cohort

The Uber Eats partnership expanded from 17 states to the entire U.S., and DoorDash was expanded to include the U.S., alongside the existing Instacart relationship in the U.S. and Canada. Average delivery times across all three platforms are under an hour.

"Members using these new marketplaces are significantly younger than our overall member base... We've observed that these sales through these channels are mostly incremental, with limited impact on our core warehouse grocery business." — Ron Vachris, CEO

This connects to the demographic disclosure: the under-40 member base has grown nearly 60% since COVID and now exceeds a quarter of the total. Management was candid that these members start out spending less and reach peak spend between roughly 40 and 55, when household size and income peak.

Assessment: Incrementality is the whole question and management asserted it without quantifying it, which is the standard tell. What is not in dispute is that Costco is acquiring a younger cohort with a longer maturity curve, and that this mechanically dilutes spend per member today in exchange for a larger base later. That is a good trade for a franchise with a 92.3% U.S./Canada renewal rate, and it is also a coherent explanation for why member count and spend-per-member growth have both been slowing at once.

9. AI search and personalization deliver a second triple-digit quarter

Traffic to costco.com originating from AI search grew triple digits for the second consecutive quarter and continues to convert at the highest rate of any traffic source. Personalized initiatives also grew triple digits, and 10% of all costco.com orders now include a personalized item. Total site and app traffic rose 30%.

"Traffic to our site from AI search grew triple digits for the second consecutive quarter and continues to show the highest conversion rate of all site traffic... Interestingly, the Costco membership is also among the top items originated from AI searches, indicating that there is a strong positive sentiment around the value of our membership coming from these sources." — Gary Millerchip, CFO

Management named the models it sees traffic from (Gemini, Anthropic and OpenAI), noted that Costco does no paid digital advertising, and framed the channel as a neutral environment in which curated assortment and pricing authority surface on their own merits. Separately, the Costco Next extended marketplace was discontinued as immaterial.

Assessment: Still option value, not an FY27 earnings driver, and management is honest that the base is low. The structurally interesting claim is the membership itself ranking among top AI-originated items, which would make LLM search a member-acquisition channel rather than just a merchandising one. That is a genuinely new idea in this coverage and worth tracking, because member acquisition is precisely the metric that has been decelerating for five quarters.

10. Gas: a record year and the loyalty flywheel

Gas comps ran positive mid-30s on both higher price per gallon and record volumes. For the fiscal year, U.S. member-household gas penetration reached an all-time high, and the company estimates it saved members over $3.2B versus average pump prices in its markets. Twenty-six existing high-volume U.S. gas stations were expanded during the year to increase throughput.

"But we do believe as members see the value of gas that we're offering and we've got more new members buying gas for the first time, that we think that's a great indicator of continued spend and growth with members as well." — Gary Millerchip, CFO

Assessment: The FQ3 read holds: gas is a traffic-and-loyalty engine, not a profit pool, and the volume surge is contingent on elevated pump prices persisting. The difference this quarter is that the company invested capital in it, expanding 26 stations, which converts a cyclical windfall into durable throughput capacity. The all-time-high household penetration is the genuinely durable piece; the mid-30s comp is not.

11. Pharmacy and healthcare: nearly +20% for the year, with a GLP-1 price headwind absorbed

Pharmacy sales grew nearly 20% for the fiscal year on double-digit script growth, which management said more than offset the headwind from Medicare maximum-fair-price changes. Digital capability is doing the work: prescription mobile pay-ahead penetration exceeds one third of prescriptions at many U.S. buildings. A new partnership with SCAN Health Plan to develop Medicare Advantage benefits was announced.

"When we see lower prices, we see an opportunity, and we've really taken great advantage of that this year in the pharmacy with the lower GLP-1 pricing, the lower drug prices that we've been out there, how we've been able to capitalize on more unit growth." — Ron Vachris, CEO

Assessment: Pharmacy has now absorbed two consecutive rounds of government-driven price compression and grown nearly 20% anyway, which is the strongest possible evidence that the unit-growth response is real. The SCAN partnership is the first step beyond dispensing into benefit design, and it is the kind of move that turns pharmacy from an ancillary category into a member-acquisition vector for an aging cohort. Asked about a further round of GLP-1 price cuts in calendar 2027, management said only that it expects the underlying headwinds to continue.

Guidance & Outlook

Costco issues no revenue or EPS guidance as a matter of policy, so the outlook is a set of operating commitments rather than a guide. The forward anchors disclosed this quarter:

Implied trajectory: The FY27 bridge is roughly 6.0-6.5% adjusted comps, 28 net new warehouses adding about 3% to the unit base, membership fee income growing 6-7% with no increase assumed, modest core-on-core expansion, SG&A leverage at that comp level, and interest income holding near current run-rates on a $20.2B cash balance. That arithmetic supports low-double-digit EPS growth. It does not support a repeat of FY26's 14.0%, which carried a final quarter of fee tailwind and a one-time tariff benefit.

Street at: Post-print targets came down while ratings held, which is the signature of a Street marking its multiple assumption lower rather than its earnings. The next company datapoint is the September sales release for the five weeks ending October 4, due October 7 after the close.

Guidance style: Characteristically number-light, with one notable improvement and one notable omission. The improvement is the commitment to disclose the net tariff-refund impact each quarter, which removes a modeling unknown. The omission is that the FQ3 gross-margin description, that the ex-gas rate had "generally been stable with a slight improvement, sort of in the mid single digit range" over the prior 12 to 24 months, was not repeated in any form.

Analyst Q&A Highlights

Whether price investment is still moving the sales needle

The opening question was the sharpest of the call and went directly at the return on the tariff-refund reinvestment: if roughly $100M has gone into price and monthly sales have if anything moderated, what does that say about elasticity, and should the Street simply reset to historic comp levels? Management answered the demand half at length and the return-on-investment half not at all, reframing the reinvestment as an obligation to members rather than a lever expected to produce sales.

Q: "It sounds like given the tariff refund, you have invested about $100 million into price to reinforce Costco's value proposition. Yet we really haven't seen a meaningful change in the trajectory of the monthly sales as of yet, and if anything, it has moderated a touch. How are you evaluating the return on those investments? If price is not moving the needle as significantly as maybe it has in the past, should we calibrate our expectations that this year's just going to be a more return to historic levels of same-store sales growth for Costco?"
— Michael Lasser, UBS

A: "On the pricing reinvestments, to your point, really, we view that as very much a giving value back to our members for the tariff refunds that we received, and our goal was to make sure that we spread those as we could to items that would have the most impact for members... Overall, we view that very much as giving a value back to our members for the tariffs that we pay, but we still feel very good about the comparable sales trajectory that we see in our performance and how our members are showing up and spending."
— Gary Millerchip, CFO

Assessment: The reframe is philosophically consistent with how this company has always described price investment, and it sidesteps the actual question. The elasticity point stands unanswered: adjusted comps have sat in a 6-7% band for a full year through periods of heavy price investment and periods of none. That is either remarkable stability or evidence that incremental price investment is no longer buying incremental volume. We lean toward the former, but it is the right question and it did not get an answer.

Whether the narrowing spread between member growth and unit growth matters

A second line of questioning put the structural version of the same concern: paid-member growth is converging toward warehouse-count growth, which means the average new warehouse is adding fewer incremental members. The question offered management an out, asking whether rising spend per member makes the convergence noise. Management took the out and built on it.

Q: "My question is on membership. The spread between membership growth and the unit growth is narrowing, not the MFI, but membership growth. How do you look at that? How should we look at it? I am sure it is a growth driver in some way, but can you make a case that as long as existing member spend continues to climb, then the narrowing spread is just noise?"
— Simeon Gutman, Morgan Stanley

A: "We exit fiscal year 2026 with a record high executive penetration, as I mentioned, record high gas engagement, record high digital engagement. When you look at the things that we think about that are the strongest indicators of loyalty and likelihood of members continuing to spend the biggest share of wallet and spend more top-line sales with us, then we do believe that really bodes well."
— Gary Millerchip, CFO

Assessment: Three record-high engagement metrics is a substantive answer and not a dodge. It is also a different business model than the one the bull case was built on. The historical algorithm was members times spend, with both compounding; the FY27 algorithm is a slower member count times a faster spend per member. That can produce the same revenue growth, but it is a lower-quality version of it because spend-per-member gains are finite and member additions compound.

What would reverse the run of slowing member growth

The most direct challenge of the call put a number on the trend and asked for both a mechanism and a timeline. Management supplied two mechanisms, both of which sit outside the FY27 plan, and declined the timeline.

Q: "I guess the question is with eight quarters in a row where the membership growth has slowed, what do you think it will take to reverse that declining trend, and would you care to speculate when we might be able to see that?"
— Scot Ciccarelli, Truist

A: "I think over time, as you think about the longer-term benefit of executive membership penetration growing, that would certainly be a tailwind in renewal rate that could be an accelerator to total membership growth. As Ron mentioned in some of his opening comments, as we enter some of the Asian markets with some outsized growth in new membership in the future, that could certainly be a catalyst for higher headline membership growth."
— Gary Millerchip, CFO

Assessment: The two named accelerants are executive-penetration-driven renewal gains, which is a slow arithmetic effect on a base already at 92.3% in the U.S. and Canada, and Asian market entry, which the same call placed in the FY28 pipeline. Neither is an FY27 event. The honest translation is that member growth in the high threes is the plan for the coming year, and management would prefer investors judge the franchise on engagement quality instead. That is a defensible position and it is also exactly why a re-rating is not coming from this line.

Reconciling a $152M LIFO charge with "stable" inflation

The apparent contradiction between a tripled LIFO charge and an unchanged inflation characterization drew the most technically satisfying exchange of the call. The question also picked up the Food Court churro line from the prepared remarks, which management took in good humor before delivering a clean accounting explanation.

Q: "I wonder if you could unpack the inflation commentary a little bit more. I think you said that food inflation was running similar to prior quarters, maybe around 1%, and that non-foods was higher but decelerating. I guess as part of that and the inflation outlook, how much do you think bringing the churro back will add to average ticket in the next quarter?"
— Greg Melich, Evercore ISI

A: "It might have seemed a bit contradictory for us to say inflation is relatively consistent, but we had a large LIFO charge during the quarter... Normally, if it was fairly consistent throughout the year, you would spread the charge over each quarter. But because the inflation I mentioned in consumer electronics and in the gas and petroleum-related items happened in the fourth quarter... While it sounds like a big number in the quarter, it is still really only pointing to about 1.5% inflation as the impact of LIFO on the total inventory that we have at the end of the fiscal year."
— Gary Millerchip, CFO

Assessment: A complete and verifiable answer, and the 1.5% life-of-inventory framing is the right way to size it. The part that should not be lost in the accounting explanation is the timing: the inflation that triggered the true-up arrived in the fourth quarter, in memory chips and petroleum derivatives, which means FY27 opens with those cost curves already elevated rather than lapping them.

How Costco is positioned as commerce moves into AI interfaces

A question on agentic commerce drew the most expansive strategic answer of the call. Management gently corrected the framing, distinguishing AI-assisted search from agentic transaction, and then made an unusually specific competitive argument about why a curated, non-advertising retailer benefits from a neutral recommendation layer.

Q: "On the agentic commerce front, clearly an evolving backdrop, just overall, just how do you feel Costco's positioned as the backdrop continues to evolve?"
— Rupesh Parikh, Oppenheimer

A: "We feel it's a great opportunity for us, particularly as today we don't do any paid digital advertising... But AI presents an opportunity in a sort of neutral environment for us to be able to ensure that our value and our quality shows up transparently for our members... There's a lot of work going on in the teams right now to work with those large language models to identify how we can continue to clean up and update our data and our product pages so that our value continues to show up transparently in that environment."
— Gary Millerchip, CFO

Assessment: The strategic logic is sound and specific to this business: a retailer that wins on price and curation, and that buys no search advertising, is structurally advantaged in a channel where recommendations are not auctioned. Whether that holds as AI interfaces monetize is the open question nobody asked. For now the disclosed facts, triple-digit growth for a second quarter and the highest conversion rate of any traffic source, support the argument.

Whether Asian warehouses deliver different member economics

With the opening cadence tilting international, a question sought the underlying member math: how many households does an Asian warehouse acquire relative to a U.S. one? The answer was the most operationally candid of the call and materially qualified the Asia-as-member-growth-catalyst argument management had offered earlier in the call.

Q: "Are you willing to share some metrics on what new member households look like in international countries such as Asia, and how that compares to the U.S. Because it does look like, with this acceleration to a net new 30 units, it does look like that's weighted a little bit more towards international. So can you just help us do some math around that?"
— Kelly Bania, BMO Capital Markets

A: "If you think about some of those warehouses and businesses that we have in, say, Asia, like China, Japan, Korea, in many cases, we're going to see a much higher number of members per warehouse. But they often visit less frequently... In a U.S. market, we might see less members per warehouse, but they're spending significantly more per member, and we get to a similar strong economic outcome... but that might also mean when you look at our total metrics, it might be less impressive from a sales growth per member because it brings down that overall average."
— Gary Millerchip, CFO

Assessment: The most useful disclosure in the Q&A, and it cuts both ways. Asian entry would lift headline member growth, which is the metric under pressure, but management pre-emptively warned that it dilutes spend per member and that some of those first-year members are tourists rather than annuities. That is honest, and it means the eventual Asia catalyst will improve one bear point while worsening another. Investors should not expect the FY28 openings to be a clean fix.

Whether grocery inflation is about to inflect higher

A layered question asked management to look forward on grocery inflation across the industry: egg deflation is lapping, tariff-rebate reinvestment is ending elsewhere, and oil has been persistent, so does the low-single-digit print inflect? The answer declined to forecast and named the two variables that make forecasting impossible.

Q: "Just want to get the overall view on how you think grocery inflation proceeds over the next three to six months and whether the industry is sort of strong enough and rational enough for those prices to actually go up."
— Chris Horvers, JPMorgan

A: "I think for us, as you might expect, we think it's difficult to predict what's going to happen with inflation. I think especially because of the uncertainty around the situation with the Middle East and how long that might impact and what you might see in terms of volatility in prices on oil and oil-related items. Then, of course, there's still uncertainty around potentially where exactly we land on the tariffs that exist today or could exist in the future."
— Gary Millerchip, CFO

Assessment: A genuine non-answer, but an honest one, and it names the right two variables. The more telling part is what followed: management redirected to finding cost savings that offset commodity inflation, which is the posture of a team planning to eat inflation rather than pass it through. That is good for share and for the value perception that drives renewals, and it caps gross-margin upside if the oil curve stays where it is.

Where SG&A pressure is building, and the next GLP-1 price step-down

The final question of the call paired the leverage question with a specific forward risk: whether the next round of GLP-1 price cuts in calendar 2027 becomes a larger headwind than the first. Management gave the clearest leverage commitment of the call and then reframed the drug-pricing question as a volume opportunity.

Q: "I was wondering if there's any specific cost buckets within SG&A where you're starting to see more pressure relative to the last few quarters, and if there's any reason you wouldn't be able to continue leveraging SG&A if you're comping in this 6% range in the foreseeable future... Just curious if you're starting to see an impact to sales from some of these recent price changes, and if not, could we see a larger headwind next year in calendar 2027 from the next round of pricing cuts?"
— Christopher Nardone, Bank of America

A: "Q4 would be a good example where we were able to continue to achieve a little bit of SG&A leverage, even with continued cost increases in those areas. But certainly if that trend were to deteriorate, that would be a factor that could influence our ability to do that. But when we look at our employee agreements, I think those are factored in for a three-year period, and year one is always the most expensive of a three-year agreement there. So we're into year two now with that."
— Gary Millerchip, CFO

Assessment: The wage-agreement cycle detail is new and it is the reason to believe the leverage commitment. Year one of a three-year agreement is the most expensive and the company is now into year two, which gives SG&A an easier wage comparison from here than it had in year one. The GLP-1 half was answered with an argument from precedent rather than from arithmetic: prices fell before, scripts rose enough to offset, therefore it should hold. That is a reasonable base case and it is not a forecast.

What They're NOT Saying

1. Capital allocation was absent from a fiscal year-end call. The words dividend, buyback, repurchase and special dividend do not appear anywhere in the transcript. Cash grew $6.0B during FY26 to $20,207M, roughly $45.60 per share, and the company returned only $3.3B across dividends and buybacks against $9.4B of free cash flow. At FQ3 the special-dividend question was live enough to consume a full exchange. This quarter no one asked and management volunteered nothing, on the one call of the year where a year-end capital plan would be conventional.
2. The tariff-refund lawsuit disclosed at FQ3 was not mentioned once. Management told the May call that the amount and timing of member returns depended partly on "developments in the lawsuit filed against the company regarding the return process." Four months later, with $184M collected and a similar amount already in hand for FQ1, the litigation does not appear in the transcript at all. Silence is not resolution.
3. No forward gross-margin framework at all. At FQ3 management described the ex-gas rate over the prior 12 to 24 months as "stable with a slight improvement, sort of in the mid single digit range." This quarter it reported an excellent core-on-core number and gave no framing of any kind, backward or forward. Reporting a good print and not repeating even that description is a conspicuous combination.
4. No timeline, and no question, on the next membership fee increase. Management confirmed Q4 was the final quarter of benefit from the September 2024 increase. The obvious follow-up, when the next one comes, was neither volunteered nor asked. It is the single largest available lever for re-accelerating the fee line and it went unaddressed on the call where the old one expired.
5. No floor on member growth. The FQ3 formulation was that 4% to 5% is "a more normal rate of growth." The FQ4 print of 3.8% is below that band, and rather than defend the band management described the recent rate as "more typical of what we'd expect." A description of the present replaced a commitment about the future, and nobody pushed back on the substitution.
6. Canada's deceleration went unexplained. Canadian adjusted comps of 4.6% sit 210bps below the country's own full-year figure and 210bps below the company total, in the second-largest market with five FY27 openings planned. The number is in the press release table. It was not discussed in the prepared remarks and no analyst raised it.
7. The remaining tariff refunds were not sized in earnings terms. Management committed to reporting the net quarterly impact going forward, which is a real improvement, but declined to quantify the aggregate. "A little more than one third" of the total is the only anchor given, and the reinvestment share was described only as "the majority."

Market Reaction

• Pre-print setup: COST closed at $896.48 on September 24, up 4.0% year-to-date against the S&P 500's +12.5%, down 6.6% over the trailing 30 days and down 5.2% over the trailing twelve months. The 52-week closing range entering the print was $850.00 to $1,094.32, putting the shares roughly 18% below the twelve-month high and about 5% above the low.
• After-hours move: Essentially nothing. The shares ticked up around 0.3% after the release and were marginally lower in Friday premarket. The tape did not take a view overnight.
• Next-day session: The stock opened at $887.00, a 1.1% gap down, traded as low as $883.10, then reversed through the day to close at $922.77, up 2.9% and $26.29 on the session. The intraday range spanned $883.10 to $925.14, a swing of roughly 4.8% from low to high, and the close was within $2.40 of the high.
• Volume: 4.6M shares against a 2.0M 30-day average, 2.3 times normal.
• Relative: The S&P 500 rose 0.5% on the same session, so COST outperformed by roughly 240bps.

The shape of the move is more informative than its size. A gap down at the open, a reversal through the session, and a close near the high on 2.3 times normal volume is the signature of a market that read the headline first and the gross-margin matrix second. The headline said reported gross margin fell 11bps, which is what a fast reader takes from the press release. The matrix said core-on-core was up 18bps and broad-based, which requires the call to find, because the press release does not contain it.

The timing says the market was slow to price the call, not slow to hear it. The call, and the core-on-core figure with it, was public from 5:00 PM ET on September 24, the evening before the session, yet the premarket was still marginally lower and the stock opened with a gap down. The regular session then bought the stock 4.5% off its low. Our read is that the buyers were paying for the margin repair, not for the EPS beat, most of which the release itself identified as a non-recurring tariff item.

The setup matters too. This was a stock that entered its fiscal year-end print having fallen 5.2% over twelve months while growing earnings 14.0%, and 6.6% in the preceding thirty days alone. Sentiment coming in was poor enough that a quarter with one genuinely repaired pillar was sufficient to produce a 240bp day of relative outperformance. That asymmetry, where modest good news gets paid for, is itself evidence that the crowding we flagged at FQ3 has substantially cleared.

Street Perspective

Debate: Does the core-on-core repair prove the FQ3 dip was a choice, or was FQ4 flattered?

Bull view: The +18bps is stated explicitly as excluding tariff refunds and their reinvestment, so the flattering item is already stripped, and the breadth claim covers all three merchandise categories. The drivers named are supply-chain efficiency and labor productivity, which are structural and repeatable.

Bear view: One quarter does not restore a streak that was broken, and the company withdrew its forward margin framework in the same breath as reporting the number. Non-foods, the strongest category, was also the category recovering from tariff-driven SKU disruption, so part of the expansion is a recovery comparison rather than a new run-rate.

Our take: The bull has the better argument and we are conceding the point. The exclusion is explicit, the breadth is explicit, and the mechanism is named. The bear's forward-framework observation is fair and is why we restore the pillar without also restoring the rating.

Debate: Is decelerating member growth a warning on the comp funnel, or noise while spend compounds?

Bull view: Every engagement metric that predicts future spend hit a record: executive penetration, gas penetration, digital engagement. Renewal rose on both measures. A slower-growing base of higher-quality members produces the same revenue with better economics, and the under-40 cohort is a decade-long annuity being acquired at today's prices.

Bear view: Five consecutive quarters of deceleration is a trend, not a quarter, and management named only two accelerants, both outside the FY27 plan. Spend-per-member gains are bounded by household budgets; member additions compound. Substituting the second for the first lowers the terminal growth rate, which is precisely what a 40-plus multiple is discounting against.

Our take: The bear has the better forward argument, and management's own answer on Asian warehouse economics strengthens it: the eventual member-growth catalyst comes with a spend-per-member dilution attached. This is the live bear point and it got worse this quarter.

Debate: Has the de-rating created an entry point at 44x trailing?

Bull view: The stock fell 5.2% over twelve months while earnings grew 14.0%, which is roughly 17% of multiple compression absorbed without any deterioration in the business. Free cash flow grew 19.8% to $9.4B, the balance sheet holds $20.2B of cash, and the capex peak now has a stated end. Buying a franchise of this quality after a year of pure de-rating is how the last upgrade worked.

Bear view: 44x trailing and roughly 40x forward for a low-double-digit grower is still a PEG above 3.5x. The de-rating from 50x to 40x removed the most egregious part of the premium; it did not create value. And the earnings growth rate that multiple is being paid against is itself stepping down as the fee tailwind ends.

Our take: Both are right about different things, which is the definition of a Hold. The compression is real and was the most credible argument for an upgrade in this note. It is not enough on its own, because what compressed was the multiple on a growth rate that is also compressing.

Debate: Are the national Uber Eats and DoorDash rollouts incremental, or a margin-diluting concession?

Bull view: Management says the sales are mostly incremental with limited impact on core warehouse grocery, that the members using them are significantly younger, and that members engaging across channels shop more frequently in the warehouse as well. This is member acquisition and share of wallet in a cohort Costco has historically under-indexed to.

Bear view: Third-party marketplace economics are worse than warehouse economics, the incrementality claim was asserted without a number, and a delivery-first relationship with a young member may never convert into the high-frequency warehouse habit that makes the model work. The Costco Next shutdown in the same quarter is a reminder that not every digital extension survives.

Our take: Incremental until proven otherwise, but the margin question is real and unquantified. The relevant test is not delivery volume, it is whether under-40 renewal rates hold as that cohort's share of the base rises past a quarter. Management has not disclosed cohort renewal and should be asked for it.

Model Update Needed

Valuation impact: At the $922.77 reaction close, COST trades at 44.4x FY26 diluted EPS of $20.76, 44.8x the $20.61 ex-tariff-item figure, and roughly 40x the midpoint of our FY27 estimate. On free cash flow the multiple is 43.6x on $9,390M, a 2.3% yield. Market capitalization is roughly $409B against $21.3B of cash and short-term investments and $6.2B of total debt. Our FY27 estimate is new in this note: against the post-FQ3 driver assumptions it takes the margin repair and the lower tax rate up and the slower fee line down. The larger change is not the earnings path but the price paid for it, and roughly 40x forward for a low-double-digit grower is a fair price rather than a cheap one.

The conditions that would take us back to Outperform are specific: core-on-core expansion sustained for a second and third consecutive quarter; paid-member growth stabilizing at or above 4% rather than continuing to step down; an announced membership fee increase, which would restore the annuity's growth rate outright; or the shares trading into the lower half of the 52-week range without a corresponding deterioration in the operating metrics. Any two of those four would do it. The last is not met today: the shares are in the lower half of the range, at $922.77 against a $972.16 midpoint, but paid-member growth slowed for a fifth straight quarter, which is the deterioration that condition excludes.

Thesis Scorecard Post-Earnings

Overall: The thesis is modestly stronger than at FQ3 but the rating conclusion is unchanged. Two of the three legs of our May downgrade have moved against us and we are marking both: core-on-core has re-expanded broadly, and a year of de-rating has narrowed the valuation objection. The third leg holds and in fact strengthened, because the membership-fee tailwind ended exactly as forecast while the member count decelerated for a fifth consecutive quarter with no floor committed. Net, one bull pillar restored, one bull pillar retired, one bear point eased and one bear point escalated.

Action: Hold. Own it, do not add here. This is a franchise compounding earnings at low double digits with best-in-class renewal, a genuinely accelerating unit pipeline and a stated end to the elevated-capex phase, trading at roughly 40x forward. That combination is fairly priced rather than mispriced. We have named four specific conditions for an upgrade above; the price condition is not met while member growth keeps slowing, and we would expect the fastest route to be a membership fee increase, which would repair the pillar that ended this quarter in a single announcement. Absent one, the algorithm is a good business at a full price, which is a Hold.