The Margin Trough Reversed in One Quarter. Same-Store Sales Got Worse. The First Fact Matters More.

Results vs. Consensus

Luckin reports in renminbi with a US dollar convenience translation at RMB6.7980 to US$1.00. We keep renminbi as the currency of record. The Street quotes this name on non-GAAP earnings per ADS, which strips share-based compensation and its tax effect and nothing else, so the comparison below is like-for-like. Consensus was firm this quarter: two independent providers agreed to within US$0.003 on earnings and US$3M on revenue.

Year-over-Year Comparison

Note on comparatives. Beginning in Q1 2026 the company revised its definition of same-store sales growth and restated prior periods to conform. The revision reaches past the operating metrics into the income statement. Same-store sales, reported at the time as +13.4%, are now +13.8%. Store-level operating margin, 21.0% on the basis published in July 2025, is now 21.5%. GAAP operating income, then RMB1,700.1M, is now RMB1,739.7M. GAAP net income, then RMB1,250.5M, is now RMB1,280.3M. Cost of materials, then RMB4,600.5M, is now RMB4,560.8M. Every comparative in this report is taken from the current filing rather than from the 2025 releases, so the year-over-year comparison is drawn on one basis throughout. The self-operated revenue base is unaffected. Readers cross-checking against our own 2025 coverage will find the superseded figures there.

Sequential Comparison (vs. Q1 2026)

Luckin is a seasonal business and Q2 is its strongest quarter, so the sequential move overstates the underlying improvement. It is shown here because the size of the step is the quarter's central fact, and because a seasonal benchmark exists: last year the same Q1-to-Q2 step in store-level margin was 400 to 450bps, depending on whether the prior-year Q2 is taken as published at the time (21.0%) or as restated in the current release (21.5%).

Quality of Beat/Miss

Revenue. A clean 3.1% beat, and the composition is better than the headline. Growth of 28.5% is a deceleration from Q1's 35.3%, which is what a subsidy lap produces, but it is entirely unit-driven and entirely organic: no acquisitions, no accounting change, no currency effect (the company reports in its functional currency). Product sales grew 28.7% and partnership store revenue 27.9%, so the two engines grew in line with each other for the first time in five quarters. The one soft line is "net revenues from others," which fell 17.0% to RMB171.1M, but at 1.1% of the total it is noise.

Margins. This is where the quarter was won, and the mechanism is specific. Delivery expense fell 3.1% in absolute renminbi, to RMB1,617.9M from RMB1,669.7M, while revenue grew 28.5%. That is a 333bp reduction in the delivery line as a share of revenue, from 13.5% to 10.2%, and it is worth more than the entire 71bp decline in operating margin. Every cost line that matters went the wrong way: materials rose 165bps to 38.6% on green-bean prices and selling-price mix, store rental and other operating costs rose 119bps to 22.8% despite peak season, and sales and marketing rose 103bps to 5.8% on platform commissions. Only depreciation and store preopening improved among the lines that carry weight, by 6bps and 2bps. Strip the delivery line out and this was a margin-compression quarter. Include it and operating income grew 22.0%. The recovery is real, and it is currently a one-legged stool.

EPS. The cleanest line in the print. Non-GAAP net income grew 22.8% and non-GAAP EPADS 21.4%, so the per-share figure is tracking the operating result rather than being flattered by the buyback (the share count reduction is real but still small, roughly 15.7M weighted-average shares sequentially against a 2.57 billion base). GAAP net income grew a slower 16.1%, and the gap is explained by two things worth naming: share-based compensation rose 76.7% to RMB272.9M, which is 1.7% of revenue against 1.2% a year ago, and the effective tax rate ticked up to 30.3% from 29.7%. Below the operating line, interest and investment income fell to RMB30.5M from RMB42.6M and the company booked RMB18.0M of net foreign-exchange loss. None of it changes the shape of the quarter, but the non-GAAP-to-GAAP gap is widening, and it is widening because of stock compensation.

Cost Structure and the Anatomy of the Margin Move

The single most useful table in this print is the operating-expense breakdown as a share of revenue, because it shows that the margin recovery had exactly one source. Every line is drawn from the income statement in the 6-K exhibit; the components sum to the reported total exactly.

Read down that table and the story is unambiguous. Delivery gave back 333bps. Materials, rent, marketing and administration took 411bps. The four residual lines, depreciation, store preopening, impairment and the restructuring item, net to 7bps of relief, leaving operating margin down 71bps year over year. The business still grew operating income 22.0%, because the revenue base grew 28.5%. Against Q1 2026 the picture is far more flattering, because Q1 carried the winter off-peak: materials fell 191bps sequentially and store rental fell 276bps purely on volume leverage.

The forward question is how much delivery reversion is left. The line ran at roughly 7% of revenue in Q2 2024, before the platform subsidy war, peaked at 18.9% in Q3 2025, and now sits at 10.2%. On that arithmetic there is something on the order of 300bps of further normalization available if the mix returns to its pre-war shape, which would be enough to offset the material and rent pressure and then some. That is the bull case in one sentence, and it is testable within two quarters.

Segment Performance

The two-engine model converged this quarter. For five quarters the asset-light partnership channel had grown materially faster than the self-operated base; in Q2 the two grew within 130bps of each other. That convergence is worth noticing, because partnership growth had been the line that flattered the consolidated number while self-operated margins were compressing.

Self-Operated Stores

Self-operated revenue grew 26.6% to RMB11,564.1M and store-level operating profit grew 25.9% to RMB2,468.6M, holding the store-level margin at 21.3% against a restated 21.5% a year ago. That margin is the recovery: one quarter ago it printed 13.6%, and the 770bp sequential step compares with a 400 to 450bp seasonal step in the same quarters last year. Between 320 and 370bps of the move is therefore genuine recovery rather than summer. We cannot narrow the range further, because the prior-year Q2 figure was restated in this release while the prior-year Q1 figure was not republished.

"Store-level operating profit increased by 26% year-over-year to RMB 2.5 billion, with self-operated store level operating margin of 21.3%." — Jing An, CFO

Underneath that headline sits the quarter's least-discussed number. Self-operated stores grew from 16,968 to 23,734, up 39.9%, while self-operated revenue grew 26.6%. Measured against the average store count in each period, revenue per self-operated store fell roughly 9.5% year over year, from about RMB561k to about RMB508k. That decline is nearly twice the -5.3% same-store number, and the gap is definitional: same-store sales measure only stores open through both comparable periods, so the additional 4 points of dilution is the drag from 6,766 net new stores opening below system average. Neither the release nor the call addressed average unit volume.

Assessment: Holding a 21.3% store-level margin while average unit volume falls 9.5% is a genuinely impressive operating result, and it is the strongest evidence yet that the cost structure is flexible rather than fixed. It is also the quarter's most fragile fact, because the offset came almost entirely from the delivery line, which is a finite, one-time reversion. When delivery normalization is exhausted, the AUV dilution will land on the margin unfiltered unless new-store productivity improves. That is the single number we will be watching for the next four quarters.

Partnership Stores

Partnership revenue grew 27.9% to RMB3,667.8M, holding 23.1% of the total against 23.2% a year ago, with the partnership base at 12,576 stores, up 36.1%. The company disclosed the composition for the first time in a while: sales of materials RMB2,402.2M, delivery service fees RMB542.1M, profit sharing and royalty fees RMB492.1M, equipment sales RMB203.4M, and franchise and other service fees RMB28.0M. Materials are 65.5% of the channel's revenue, which is a useful reminder that this is a wholesale-supply relationship with a royalty attached, not a classic high-margin franchising stream.

Assessment: The channel is doing its job of extending the footprint at low capital intensity, but the narrowing growth premium over self-operated stores means it is no longer masking anything. The disclosure detail is welcome and slightly deflationary to the "asset-light franchise" framing: at two-thirds materials pass-through, partnership revenue carries a very different margin profile from the royalty line investors sometimes assume.

Store Network and International

Luckin opened 2,714 net new stores in the quarter, an 8.1% sequential increase in the unit base, and more than 5,000 in the first half. It did so in the same quarter that same-store sales fell 5.3%, which is the strategic choice the Q&A focused on. The international footprint remains small but is now present in four markets.

Key KPIs

The same-store sales series on the company's restated definition now reads: -2.0% (Q4 2024), +9.2% (Q1 2025), +13.8% (Q2 2025), +14.3% (Q3 2025), +1.3% (Q4 2025), -0.1% (Q1 2026), -5.3% (Q2 2026). Read as a curve, the subsidy war is visible as a single hump and the current decline is its mirror image. Q3 2025 was the peak at +14.3%, which is why management is guiding to continued pressure through Q3 2026.

Key Topics & Management Commentary

Overall Management Tone: Confident and, unusually for this cycle, willing to be specific about the shape of the problem rather than only the shape of the recovery. Management volunteered that subsidies were withdrawn faster than it had expected at the start of the year, named Q3 as the quarter where the high base still bites, and closed on "cautiously optimistic" rather than reaching for a stronger word. The posture on expansion was the most assertive of the past four calls, and the least evidenced: the defence of opening 2,714 stores into a same-store decline was made entirely at the level of strategy and market runway, with no unit-economics data offered.

1. The Margin Trough Is Confirmed and Behind Them

Three quarters ago the delivery-expense line was consuming 18.9% of revenue and this franchise's operating margin had halved. We wrote at Q1 that the trough was forming rather than deepening and that delivery-mix reversion was the mechanism by which margin would recover. Q2 settles it. Operating income grew year over year for the first time in three quarters, store-level margin returned to within 20bps of the prior year, and the CFO led with recovery rather than defence.

"In the second quarter, we delivered solid top-line growth and continued profitability recovery, reflecting our operational resilience and the strong execution of our growth strategy." — Jing An, CFO

The interpreted CEO remarks made the same point and tied it explicitly to the competitive environment rather than to anything Luckin did: profitability recovered because "food delivery competition gradually returned to more rational levels."

Assessment: The single most important confirmation in this print, and it validates the specific mechanism the Q1 upgrade rested on rather than merely the direction. The caveat is that management attributes the recovery to industry rationalization, which is an external variable outside its control and one that could reverse if a platform reopens subsidies. The recovery is real; the credit belongs at least partly to the market.

2. Same-Store Sales Deteriorate to -5.3%, and Q3 Will Not Be Better

Self-operated same-store sales fell 5.3%, against -0.1% in Q1 and +13.8% in the year-ago quarter. Management framed it as the arithmetic consequence of lapping the 2025 subsidy surge, and, importantly, said the quarter came in as planned.

"Same-store sales for our self-operated stores declined 5.3% year-over-year in the second quarter, primarily due to the temporary high comparison base raised by elevated food delivery subsidies in the prior-year period. This performance was fully consistent with our previous expectations for the trajectory of business operations." — Jinyi Guo, Co-founder & CEO (via interpreter)

On the call management went further than the release and gave the Street a forward marker, which for a company that publishes no numeric guidance is the closest thing to an outlook it offers.

"From an external perspective food delivery subsidies have been scaled back more quickly than we initially expected at the beginning of the year. The high base pressure created by last year's elevated subsidies has also become more apparent, and it's already reflected in our second quarter same-store sales performance. As for the third quarter, as platform subsidies were particularly intense during July and August of last year. We expect the resulting high base effect to persist." — Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: The claim that this was "fully consistent with our previous expectations" is defensible on direction and generous on magnitude. Two calls ago management described 2026 as choppy; a move from -0.1% to -5.3% in one quarter is more than choppy. The offsetting fact is that Q3 2025 was the peak comparison at +14.3%, so the worst of the lap is now dated and self-correcting: by Q4 the base is +1.3% and by Q1 2027 it is -0.1%. The optics stay bad for one more quarter and then mechanically improve.

3. The Number Nobody Discussed: Average Unit Volume

Same-store sales measure a comparable-store cohort. They do not capture what happens to the average store when the denominator grows 39.9% in a year. Self-operated revenue per store, measured on average store counts, fell roughly 9.5% year over year. The gap between that figure and the -5.3% same-store number is the dilution from new stores opening below system average, and it went unmentioned in the release, the prepared remarks and the Q&A.

This matters because it separates two very different bear cases. If the whole decline were the subsidy lap, it reverses on its own when the base resets. If a meaningful part is density and cannibalization, it does not. Management's own framing implicitly acknowledges the second possibility without engaging it: the defence of the expansion pace rested on total-market runway, never on new-store payback or new-store revenue relative to the base.

"So for Luckin, store expansion is not simply about adding more stores. It's actually more about leveraging our nationwide store network, strong brand equity, and product abilities to better identify and serve the growing customer demand." — Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: The absence of any new-store productivity disclosure is now the largest information gap in this name. Luckin discloses store counts, same-store sales and store-level margin, but nothing that would let an outside investor separate "new stores are smaller by design and profitable on day one" from "new stores are cannibalizing the base." Given that store-level margin held at 21.3% through a 9.5% AUV decline, the evidence currently favours the benign reading, but it is inference, not disclosure.

4. Delivery Mix Reversion: How Much Is Left

Delivery expense fell in absolute renminbi for the first time in this cycle, down 3.1% to RMB1,617.9M against revenue growth of 28.5%. The CFO attributed it to two distinct drivers, mix and unit cost, which is a more durable combination than mix alone.

"This improvement was mainly driven by a lower proportion of delivery orders, as well as lower average delivery costs per order as we enhance fulfillment efficiency. In absolute terms, delivery expenses decreased by 3% year-over-year to RMB 1.6 billion." — Jing An, CFO

The line has travelled from roughly 7% of revenue before the subsidy war, to a peak of 18.9% in Q3 2025, to 10.2% now. Management explicitly linked the margin outlook to further progress on the same path.

"At the same time, we are seeing improving trends in our profitability and margins, supported by the continuous decline in delivery mix from peak levels since the third quarter of last year, as well as the improvements in our fulfillment efficiency." — Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: Roughly 300bps of further reversion is available if the mix returns to its pre-war shape, and the per-order cost component suggests some of the improvement is structural rather than purely mix. That is enough to fund another year of the materials and rent pressure visible in the cost table. It is also the whole cushion. We would rather see one quarter of margin expansion driven by something other than this line before treating the recovery as broad-based.

5. The Buyback: 65% of the Program in One Quarter

At Q1 we listed buyback execution pace as a commitment to watch, and specified what good would look like: front-loaded into weakness rather than trickled. Management repurchased 48.9M Class A ordinary shares, equivalent to 6.1M ADSs, for US$195.1M between April 29 and June 30, an average near US$32 per ADS. That is 65% of the twelve-month US$300M authorization spent in roughly two months, at a price close to the bottom of the stock's twelve-month range.

"During the second quarter, we repurchased 49 million Class A ordinary shares or 6 million ADS for a total consideration of $195 million under our $300 million share repurchase program." — Jing An, CFO

The balance sheet corroborates it: treasury shares of RMB1,330.4M appeared where there were none at December 31, and weighted-average basic shares fell by roughly 15.7 million sequentially.

Assessment: Commitment delivered, in the manner we asked for. This is the clearest evidence in three years that management treats the equity as an investment rather than a currency, and it converts the maiden authorization from a signalling gesture into a capital-allocation policy. The follow-on question is what happens when the remaining US$105M is spent, and whether the board re-ups. Management was not asked and did not volunteer.

6. Borrowing to Buy Back: The First Debt of the Cycle

The balance sheet carries RMB1,945.0M of short-term bank borrowings at June 30 against nil at December 31, and the CFO connected the two directly.

"During the quarter, we utilized short-term borrowings as part of our capital allocation strategy to support ongoing shareholder return initiatives. We continue to maintain a strong net cash position of RMB 9 billion and a healthy leverage profile, supported by our robust cash generation capabilities and a disciplined financial management." — Jing An, CFO

The arithmetic checks: RMB10,925.8M of cash, deposits and short-term investments less RMB1,945.0M of borrowings is RMB8,980.8M, which is the RMB9 billion net cash figure cited. It is worth being precise about what this does and does not mean. The quarter generated RMB2,625.3M of operating cash flow against a RMB1,330.4M repurchase, so the buyback was comfortably covered by the period's own cash generation. The borrowing is therefore not a funding necessity. The likeliest explanation, which management did not offer and we cannot confirm from the filing, is currency and location: the cash is generated onshore in renminbi and the repurchase settles offshore in dollars, and a short-term facility is the ordinary way to bridge that.

Assessment: Benign on the arithmetic and worth flagging on the principle. A company with RMB10.9B of gross cash borrowing to fund buybacks invites a question it did not answer, and for a China-domiciled issuer with this governance history the question of how freely cash moves offshore is not academic. We read it as a treasury mechanic rather than a stress signal, and we would like it addressed on the record next quarter.

7. Expansion Accelerates Into the Same-Store Decline

Luckin added 2,714 net new stores in the quarter, more than the 2,548 it added in Q1, in the same period that same-store sales fell 5.3% and while, by the Street's account, several competitors were slowing openings. Management's defence rested entirely on category runway.

"we remain very confident in the long-term growth potential of China's coffee market. Compared with more mature coffee markets globally, coffee consumption in China remains in the early stages of habit formation, with significant room for future penetration and frequency growth." — Jinyi Guo, Co-founder & CEO (via interpreter)

The operational claim underneath it is that expansion speed and store quality are not in tension because the process is automated.

"we leverage our digital capabilities and AI technologies to improve our decision-making efficiency. Our mature optimizational capabilities and strong execution actually ensure that we can open stores quickly while maintaining store quality, and continuously enhance our store performance." — Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: Strategically coherent and evidentially thin. Taking share while competitors hesitate is the right move if new-store economics hold, and the 21.3% store-level margin is a real datapoint in favour. But the claim that opening 2,714 stores a quarter does not degrade store quality was asserted rather than demonstrated, and the AUV trend is the counter-evidence management did not address. We are inclined to give the benefit of the doubt for two more quarters on the strength of the margin, not the rhetoric.

8. Input Costs and the Materials Line

Cost of materials rose 165bps to 38.6% of revenue, the second-largest adverse move in the cost table and the one with the least visibility. The CFO named two causes and quantified neither.

"Cost of materials as a percentage of total net revenues was 39%, compared to 37% in the same period of 2025, mainly due to fluctuations in average selling price and certain raw material price." — Jing An, CFO

Note the first cause. "Fluctuations in average selling price" means part of the materials-ratio increase is a pricing effect rather than an input-cost effect: if realized price per cup falls, the materials ratio rises even with flat bean costs. Management separately said its cup-size upgrade initiative "supported the overall average selling price," which points the other way. Both cannot dominate, and the release gives no per-cup pricing disclosure to adjudicate.

Assessment: Green-bean cost has now been a stated headwind for four consecutive quarters with no hedging disclosure, no quantification and no expected duration. The added wrinkle this quarter is the selling-price reference, which raises the possibility that competitive pricing is eating into the ratio alongside input costs. This remains the least transparent line in the P&L, and it is 38.6% of revenue.

9. The Platform Commission Tax

Sales and marketing expenses grew 56.1% to RMB925.1M, far outpacing 28.5% revenue growth, and the composition matters.

"In absolute terms, sales and marketing expenses increased by 56% year-over-year to RMB 925 million, mainly due to higher spending on advertising and promotion, as well as higher commission fees paid to the food delivery and live streaming platforms." — Jing An, CFO

Delivery expense is falling and platform commissions are rising, which are two halves of the same relationship moving in opposite directions. The delivery line captures fulfilment cost; the commission sits in sales and marketing. A cleaner read of the platform relationship nets the two, and on that basis the improvement is smaller than the delivery line alone suggests: delivery fell RMB51.9M while sales and marketing rose RMB332.6M.

Assessment: This is the most important qualification on the margin recovery in the whole print, and it is the reason we are not treating 300bps of further delivery reversion as pure upside. Some part of what leaves the delivery line reappears as commission. The recovery is still real, because operating income grew 22.0% with both lines included, but the headline 333bp delivery improvement overstates the economics of the platform relationship.

10. The Product and Customer Engine

Average monthly transacting customers reached a record 112.7M, up 22.9%, with more than 25 million new transacting customers added in the quarter and cumulative transacting customers approaching 500 million. The product cadence continued at 28 new freshly brewed beverages plus more than a dozen snack items.

"Driven by product innovation and brand building, we added more than 25 million new transaction customers in the second quarter, bringing average monthly transaction customers to a new record high of over 110 million, up 23% year-over-year." — Jinyi Guo, Co-founder & CEO (via interpreter)

Two product datapoints were specific enough to be checkable: the Little Butter Americano ranked among the top two sellers by volume in the quarter, and the Little Butter series passed 500 million cumulative cups, and the Lychee Americano passed 10 million cups within its first week. Five of the twenty-five products that have each passed 100 million cumulative cups are non-coffee.

Assessment: The demand side of this business is not the problem. Customer count grew 22.9% to a record while same-store sales fell 5.3%, which is exactly the signature of a business adding stores and customers faster than it grows spend per location. The product engine remains the most reliably excellent part of the franchise, and it is what makes the AUV dilution look like a distribution question rather than a brand question.

11. International: Small, Real, and Now in Four Markets

International reached 223 stores across Singapore, Malaysia and the United States, adding 46 in the quarter. The US base grew from 12 stores to 20.

"In the second quarter, we added 46 net new stores, bringing our total international store count to 223. This includes 89 self-operated stores in Singapore, 20 self-operated stores in the U.S., and 114 franchise stores in Malaysia." — Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: At 0.6% of the store base this is not yet a financial contributor, and management gave no international revenue or profitability disclosure. It matters as optionality and as a governance signal: a company opening US stores is a company that has to operate to US standards of scrutiny, which is faintly supportive of the dormant relisting ambition. Do not model it.

Guidance & Outlook

Luckin issues no numeric guidance, so there is no guide-versus-consensus comparison to make. What management did provide this quarter was more specific than usual: a named quarter for continued pressure, a named mechanism for continued margin improvement, and a deliberately hedged summary word.

"Overall, we remain cautiously optimistic about our operating performance in the second half of the year." — Jinyi Guo, Co-founder & CEO (via interpreter)

Implied trajectory: Q3 same-store sales should stay negative against the +14.3% peak comparison, then improve mechanically as the base falls to +1.3% in Q4 and -0.1% in Q1 2027. Margin should continue recovering on delivery reversion, partly offset by rising platform commissions and materials. Revenue growth decelerates from 28.5% but stays well above 20% purely on the 38.6% larger store base.

Street at: Consensus was US$0.60 non-GAAP EPADS and US$2.267B revenue for the quarter, and the print cleared both. With no forward guide issued, forward consensus will be set by the Q3 base-effect commentary rather than by any company number.

Guidance style: Consistent and, on the evidence of three calls, credible on direction. Management said at Q3 2025 and Q4 2025 that early 2026 would be choppy on same-store sales and margin, and it was. It said at Q1 that margin would recover as delivery normalized, and it did, faster than promised. The one place its framing runs ahead of the evidence is the claim that a -5.3% print was "fully consistent" with prior expectations, which is true of the sign and generous on the size.

Analyst Q&A Highlights

The Q&A ran two questions before the operator closed it for time, the shortest session in our coverage of this name. Both questions were substantive and both landed on the same tension: expansion pace against per-store performance. The entries below are organized by topic, drawn from those two exchanges.

Second-half same-store sales trajectory

The opening question set up the quarter as an outlier against a weak industry, then pressed on whether the July-August period, the hardest comparison in the year, would extend the same-store decline. Management answered with the most specific forward statement it made all call, naming Q3 explicitly and attributing the pressure to the timing of last year's subsidy peak rather than to current demand.

Q: "Congrats on a very strong 2Q, very encouraging in a weak macro environment, especially against some challenges with the whole industry year to date. And second quarter turned out to be a beat in both same-store sales and also margins, clearly diverging from and outperforming the whole industry. My question is mainly on second half outlook. Considering a tough phase during July to August, how should we think about same-store sales trends for the second half?"
— Jessie Xu, JPMorgan

A: "As for the third quarter, as platform subsidies were particularly intense during July and August of last year. We expect the resulting high base effect to persist."
— Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: Management answered the question asked, which is not always the case here, and it answered it against its own interest by confirming a third consecutive weak same-store quarter. That candour is worth something. The framing also quietly establishes the reset point: if the drag is the July-August subsidy peak, then the comparison eases materially from Q4, and any same-store weakness after that cannot be blamed on the base.

Margin and earnings outlook as delivery normalizes

The same question carried a second half asking how to think about margins and earnings as delivery mix continues to revert. The answer separated external and internal drivers, and the external one is the more important admission: subsidies came down faster than management itself had forecast, which is what produced both the margin upside and the same-store downside in the same quarter.

Q: "As the delivery mix continues to normalize, how should we think about the margin and the earnings outlook from here?"
— Jessie Xu, JPMorgan

A: "From an external perspective food delivery subsidies have been scaled back more quickly than we initially expected at the beginning of the year. ... At the same time, we are seeing improving trends in our profitability and margins, supported by the continuous decline in delivery mix from peak levels since the third quarter of last year, as well as the improvements in our fulfillment efficiency."
— Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: The most useful sentence on the call. Faster-than-expected subsidy withdrawal explains the whole quarter: it pulled the margin recovery forward and pushed the same-store comparison down, simultaneously. That single variable is why the print looks contradictory, and understanding it removes most of the apparent contradiction. It also means the two effects are linked, so a second-half same-store recovery and a further margin surge are somewhat mutually exclusive.

Justifying the expansion pace while comparable stores decline

The second questioner put the strategic tension directly, noting that competitors had slowed openings while Luckin accelerated, and asked how the company balances the two. The answer never engaged the balance. It reframed the question as one about total market runway and process capability, and did not address per-store performance at all.

Q: "So, the company has a very fast store opening pace in the first half, but in the meanwhile, we see same-store sales growth pressure under the delivery subsidy normalization, and multiple players actually slowed down store opening this year. On this backdrop, what's the support in Luckin's faster expansion, and what is the company's plan and strategy of store opening in the second half, and its opening outlook, and how to balance store opening and same-store growth?"
— Xinyu Ruan, Goldman Sachs

A: "So for Luckin, store expansion is not simply about adding more stores. It's actually more about leveraging our nationwide store network, strong brand equity, and product abilities to better identify and serve the growing customer demand. Supporting this is a highly scalable and replicable operating model, covering the full value chain across people, products, and places."
— Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: A dodge, and the most informative moment of the call for exactly that reason. The question contained three specific asks: the support for faster expansion, the second-half opening plan, and how the trade-off is balanced. None was answered. No opening target was given, no new-store payback or productivity figure was offered, and the phrase "not simply about adding more stores" does the work of avoiding the arithmetic. Management is confident about the strategy and unwilling to show the unit economics behind it.

Long-term store runway in China

The final part of the second question asked about the ceiling: how much room is left in China. This is the live debate on the Street, where saturation estimates are circulating that would put Luckin within a few thousand stores of the theoretical maximum. Management rejected the framing by pointing at consumption frequency rather than store count.

Q: "In the longer term, how do you think about the room for the store expansion in China?"
— Xinyu Ruan, Goldman Sachs

A: "To conclude, we believe there remains substantial headroom for store growth in China's coffee market. As coffee penetration and consumption frequency continue to increase, the overall market capacity will continue to expand. We are confident in our ability to maintain a competitive pace of store expansion, capture the long-term growth opportunities in China's coffee market, and continue to grow Luckin Coffee's market share."
— Jinyi Guo, Co-founder & CEO (via interpreter)

Assessment: The argument is that market capacity is a function of demand rather than a fixed number of viable sites, which is analytically correct and conveniently unfalsifiable. It is also the right answer if you believe penetration is genuinely early, and the 22.9% growth in transacting customers supports that. What it does not do is give an investor any way to test the claim, and no store-count ceiling, target or long-range plan was offered.

What They're NOT Saying

1. Average unit volume or new-store productivity: Revenue per self-operated store fell roughly 9.5% year over year and neither the release nor the call acknowledged it. There is no disclosure of new-store payback, first-year revenue, or how new-store volumes compare with the base, which is the exact data needed to separate the subsidy lap from cannibalization.
2. Any second-half store-opening target: Directly asked for the second-half opening plan and outlook, management gave neither a number nor a range, only a commitment to a "competitive pace."
3. Why the company borrowed while holding RMB10.9B of cash: The short-term borrowings were disclosed and linked to shareholder returns, but the reason a cash-rich company needed a facility was left unexplained, and nobody asked.
4. Green-bean hedging or cost duration: A fourth consecutive quarter of materials pressure with no hedging policy, no quantification of the bean-cost component, and no expected duration. The line is 38.6% of revenue.
5. Per-cup pricing: The CFO attributed part of the materials-ratio increase to "fluctuations in average selling price," which is the first time pricing has been named as a margin factor, and no realized-price disclosure exists to size it.
6. Plans after the current buyback is exhausted: With 65% of the US$300M spent in one quarter, the program will likely complete inside the fiscal year, and no one asked and management did not say whether the board would re-authorize.
7. The U.S. relisting: A third consecutive quarter with no mention, despite the company now operating 20 stores in the United States. The catalyst remains dormant.
8. International economics: 223 stores across three overseas markets and not one revenue, margin or profitability figure for any of them.
9. The truncated Q&A itself: Two questions and a hard close for time. For a company facing a live debate about whether its expansion is cannibalizing its base, taking two questions is a choice, and it is the second consecutive quarter with an unusually thin analyst session.

Market Reaction

• Pre-print setup: LKNCY closed at $34.68 on July 31, up 3.5% year to date against the S&P 500's 9.4%, down 6.5% over the trailing twelve months but up 11.6% over the trailing thirty days. The stock had already run almost 10% in the three sessions into the print, the last two on roughly 2.6x normal volume, against a 52-week closing range of $27.52 to $43.03.
• Print-day session (August 3, before the open): Shares gapped up 1.7% to open at $35.26, traded a $34.70 to $37.27 range, and closed at $37.07, up 6.9% on volume of 5.0M against a 1.2M thirty-day average (4.1x). The S&P 500 rose 1.5% on the session.
• Follow-through: The gain did not hold. Shares closed at $33.65 on August 7, giving back the entire print-day move and finishing 3.0% below the pre-print close. They have since rebuilt to $35.07 as of August 28, which is 1.1% above the pre-print close and 18.5% below the 52-week closing high.

The shape of that path is more informative than the print-day number. The market's first read was the margin recovery, and it bought it hard on four times normal volume. Within four sessions it had sold the whole move. The most plausible explanation is the one the print itself sets up: the initial reaction traded the 22.0% operating-income growth and the 21.3% store margin, and the second reaction traded the -5.3% same-store sales and management's confirmation that Q3 gets no easier. Both readings are correct, and the round trip is the market failing to decide which one is the quarter.

The pre-print run matters too. Nearly 10% of upside was already in the stock before the release, on rising volume, which compresses how much a good print can deliver and helps explain why a genuine beat on both lines produced a 6.9% day rather than the 15.9% that followed the Q1 print. Positioning was long into this one. It was capitulated into the last one.

Net of the whole episode, the stock is roughly where it started and remains well below both its 52-week high and the level at which the Street's published targets cluster. The market has not paid for the margin recovery. That is the setup, and it is the reason a quarter with a deteriorating headline metric does not change our rating.

Street Perspective

Debate: Is -5.3% a Base Effect or the Start of Cannibalization?

Bull view: The bull camp points to the arithmetic of the comparison. The subsidy hump peaked at +14.3% in Q3 2025 and the current decline is its mirror image; customers grew 22.9% to a record and cumulative transacting customers approach 500 million, which is not the profile of a business losing demand. On this view same-store sales turn mechanically as the base resets from Q4.

Bear view: The bear camp argues the base effect is a comfortable story that conceals a density problem: a business adding 2,714 stores a quarter into a market where competitors are slowing is necessarily thinning its own catchments, and the second consecutive negative print with no new-store productivity disclosure is exactly what that would look like. Saturation estimates circulating on the Street would leave Luckin with far less domestic runway than management implies.

Our take: Both are partly right, and the evidence available favours the bulls on the near term. The decisive datapoint is that store-level margin held at 21.3% while average unit volume fell about 9.5%: if new stores were structurally uneconomic, unit margin would have followed volume down, and it did not. But the bears are asking the right question, and the honest answer is that the disclosure does not exist to settle it. The base resets in two quarters and will settle it for us.

Debate: Is the Margin Recovery Durable or a One-Time Reversion?

Bull view: Delivery expense fell in absolute renminbi while revenue grew 28.5%, the CFO attributed part of it to permanently lower per-order fulfilment cost rather than mix alone, and roughly 300bps of further reversion remains available before the line returns to its pre-subsidy-war level. Operating income is compounding again at 22.0%.

Bear view: The skeptics note that every other cost line deteriorated, that operating margin still fell 71bps year over year, and that platform commissions inside sales and marketing rose RMB332.6M while delivery expense fell only RMB51.9M. On that netting, the platform relationship got more expensive, not less, and the recovery is an accounting migration between two lines.

Our take: The bears have identified the most important qualification in the print and overstated its conclusion. The netting argument is right that the 333bp delivery improvement overstates the economics, and it is wrong that nothing improved: operating income grew 22.0% with both lines included, and store-level margin recovered 770bps sequentially against a 450bp seasonal norm. The recovery is real and narrower than the headline. We want one quarter of margin expansion that does not depend on the delivery line before calling it broad-based.

Debate: Does the Buyback Justify a Re-Rating of the OTC Discount?

Bull view: The bull argument is that management spent 65% of a twelve-month authorization in two months, at prices near the bottom of the range, which is behaviour rather than signalling, and that a business generating RMB2.6B of quarterly operating cash flow with RMB9B of net cash has the capacity to keep doing it. Published price targets cluster well above the current price.

Bear view: The bear argument is that US$300M is small relative to the market value, that the program is nearly exhausted with no re-authorization committed, and that a pink-sheet listing with a fraud history does not re-rate on one quarter of repurchases. That the company borrowed to fund part of it is offered as evidence the capacity is more constrained than it appears.

Our take: We side with the bulls on the behaviour and with the bears on the timeline. Front-loading a maiden buyback into weakness is the most credible capital-allocation signal available to a management team, and Luckin sent it. But a discount built over six years does not close in one quarter, and the borrowing question deserves an answer. The buyback is a reason to own the stock; it is not by itself a re-rating catalyst. The re-rating catalyst, if it comes, is the same-store line turning positive.

Model Update Needed

Valuation impact: The model moves up on margin and down on same-store sales, and the margin move is worth more. Operating margin of roughly 11-12% for FY2026 against a prior 8-10% is a material upgrade to the earnings base, and it arrives with a share count that is now falling. The offset is that we push the same-store re-inflection out by two quarters and add an explicit average-unit-volume drag that the prior model did not carry. On balance the earnings power is higher and the quality of that earnings power is narrower, because it currently rests on one cost line. At $35.07 the stock sits 18.5% below its 52-week closing high and the market has given back the entire print-day reaction, so none of the upgrade is in the price. The thesis breaks if delivery reversion stalls above 10% of revenue while average unit volume keeps falling, because then the cost structure has no remaining offset.

Thesis Scorecard Post-Earnings

Grading the standing thesis carried from the Q1 2026 recap and the thesis of record.

Overall: Strengthened, with a narrower base. The margin pillar the Q1 upgrade rested on did not merely hold, it delivered a year early, and the capital-return pillar converted from authorization to execution. Against that, the deceleration bear point escalated materially and a new bear point on average unit volume has been added, because the same-store metric understates what is happening to the average store. We listed six commitments at Q1 and two were met: store margin recovered off the 13.6% trough, and buyback pace was front-loaded into weakness. Four were not. Revenue growth of 28.5% fell short of the 35%+ we said we wanted through the comp lap. Green-bean cost trajectory went a fourth quarter without disclosure. The relisting and dividend questions went unmentioned again. And back-half same-store re-inflection, which we called the single most important confirm, is now pushed to 2027. The thesis still strengthens on a two-of-six scorecard because the two that were met are the two the upgrade was built on, and three of the four misses are disclosure or optionality items rather than operating failures. The revenue shortfall against our own bar is a genuine miss and we are not reclassifying it.

Action: Maintain Outperform. Hold the position. The recovery mechanism worked, the buyback commitment was delivered rather than promised, and the market gave back the entire print-day reaction, so the improved earnings base is available at roughly the pre-print price. We are not adding until either the same-store line stops deteriorating or the company discloses something that lets an outside investor test new-store productivity.