Starbucks SBUX

Reference table

1. What the business is, and the transaction that reshaped it

Starbucks operates and licenses coffeehouses worldwide. Chief executive Brian R. Niccol; approximately 381,000 employees per the vendor profile — by far the largest workforce in this batch; Seattle, Washington; NASDAQ-listed since 1992. Fiscal year ends in late September; fiscal 2025 ended 2025-09-28.

Revenue by product type (seg_prod) and by geography (seg_geo) both reconcile exactly, which is worth recording given what happened to other names in this batch:

Both blocks sum to reported revenue exactly in every year shown, and the United States home market is present at 72.9%. China is broken out separately from the International segment, which is unusually useful here because China is the piece that has just been sold: $3,160.8 million, 8.5% of fiscal 2025 revenue.

Food has grown fastest — 21.5% over three years against Beverage's 15.3% — which matters because the 10-Q attributes part of the June quarter's ticket growth to "strength in customer food attach".

The China divestiture, from the 10-Q filed 2026-07-29

> "On March 30, 2026, we completed the divestiture of Starbucks retail operations in China ('the disposal group') ... Under the agreement, Boyu Capital acquired a 60% interest in Starbucks retail operations in China based on a cash-free, debt-free mutually agreed-upon total enterprise value of approximately $4 billion. The transaction was partially financed with debt issued by the newly formed joint venture and we received total consideration with respect to the transaction of $3.1 billion, inclusive of our share of the debt proceeds. Starbucks retained a 40% interest of approximately $1.2 billion, accounted for under the equity method ... The Company derecognized net assets with a carrying value of $3.4 billion, and reclassified approximately $282.8 million of cumulative translation adjustment losses and net investment hedge losses of $99.7 million ... We will continue to own and license the Starbucks brand and intellectual property to the joint venture."

And the accounting mechanics disclosed alongside, which explain some of the odd quarterly figures:

> "During the first and second quarters of fiscal 2026, we classified the assets and liabilities of the disposal group as held for sale ... which required us to cease property, plant, and equipment depreciation and operating lease right-of-use asset amortization of the related long-lived assets, resulting in reduced depreciation and amortization and store operating expenses. We also changed our indefinite reinvestment assertions upon classification as held for sale, resulting in an increase in our income tax expense. No impairment was recorded upon the classification of the disposal group as held for sale."

That last sentence explains the December 2025 quarter's bizarre tax charge of $471.6 million on pre-tax income of $764.8 million — a 61.7% effective rate — and the September 2025 quarter's $700.6 million on $833.7 million, an 84.0% rate. Neither is an operating event; both are the tax consequence of reclassifying China as held for sale. A reader looking at Starbucks' reported net income across fiscal 2026 without the filing would see a company whose earnings collapsed and then recovered for no visible reason.

Why this matters to every growth rate in the file. The 10-Q states: "consolidated net revenues decreased 1% to $9.3 billion compared to $9.5 billion in the third quarter of fiscal 2025, primarily due to the conversion of Starbucks retail operations in China to our licensed joint venture model." And: "Company-operated stores revenue decreased $306 million, primarily driven by the conversion of Starbucks retail operations in China to our licensed joint venture model ($776 million) and unfavorable foreign currency impacts ($52 million). This was partially offset by a 7.9% increase in comparable store sales ($524 million)."

So the June quarter decomposes as: minus $776 million of China conversion, minus $52 million of currency, plus $524 million of comparable sales, plus licensed-store and other revenue. Reported revenue fell 1% while the underlying business grew nearly 8%. No vendor field reflects any of this.

2. The comparable-store number, which is the whole operating case

From the 10-Q filed 2026-07-29:

> "Starbucks results for the third quarter of fiscal 2026 showed continued progress and momentum on key 'Back to Starbucks' initiatives ... global comparable store sales growth, consolidated operating margin expansion, and improved customer engagement. These initiatives included the Green Apron Service standard to improve operational consistency ..., engaging consumer marketing, disciplined menu innovation, a redesigned Starbucks Rewards program, and coffeehouse uplifts ... The decline in consolidated revenues was offset by a 7.9% increase in global comparable store sales, driven by a 7.9% increase in the U.S. market ... the increase in comparable store sales was driven by a 4.2% increase in comparable transactions and a 3.6% increase in average ticket, primarily driven by higher delivery sales and strength in customer food attach and beverage modifications. Consolidated operating margin expanded 60 basis points from the prior year to 10.5%, primarily driven by sales leverage and lower inflation paired with tariff refunds, offset by higher restructuring costs and labor investments largely in support of 'Back to Starbucks.'"

Three things in that passage deserve isolating.

First, transactions are up 4.2%. In restaurant retail this is the number that cannot be faked. A comparable-sales figure driven by price tells you the company raised prices; a comparable-sales figure driven by transactions tells you more people chose to come. Starbucks got 4.2 points of its 7.9 from customer counts. In this batch, Lowe's comped +0.6% with transactions DOWN 0.9%. The contrast is the single sharpest operating comparison available across these twelve names.

Second, the margin expanded despite the investment, not because it stopped. The 10-Q says expansion came from sales leverage "offset by higher restructuring costs and labor investments." That is the favourable order of events: leverage is already outrunning the cost of the turnaround while the turnaround is still being paid for.

Third, "tariff refunds" is in that sentence and is a one-off. It is not quantified, and a reader should not extrapolate the full 60 basis points.

The earnings series shows the inflection clearly:

Three consecutive misses, then two consecutive large beats. The earn_cal basis is consistent across all five rows, so this is a like-for-like series and the turn is real rather than definitional. The June quarter's $0.85 is the highest adjusted EPS in the file and 70% above the year-ago $0.50.

3. Margins, and the size of the repair still to come

Fiscal 2025 is the damage: operating margin fell 540 basis points, net income halved, and earnings per share fell 51%. That is the cost of the turnaround — labour hours restored, stores refurbished, restructuring taken, menus reset — and it is why the trailing price-to-earnings ratio of 60.6x is a number about the denominator rather than about the price.

The arithmetic of the recovery, and it is the whole medium-term case. On a revenue base of roughly $38 billion, each point of operating margin is approximately $380 million of operating profit, or about $0.25 of after-tax EPS on 1,140 million shares. Fiscal 2025 ran at 9.6%; the June quarter at 10.5%; fiscal 2024 was 15.0% and fiscal 2023 16.3%. Recovering to the fiscal 2024 level would be worth roughly $1.35 of EPS — more than half the entire fiscal 2026 consensus of $2.58. That is why consensus models EPS rising 21.6% in fiscal 2027 and 18.8% in fiscal 2028 on revenue growth of only 1.6% and 5.3%.

A note on depreciation, which is a genuine data point rather than an artefact. cf_a reports fiscal 2025 depreciation and amortisation of $2,606.2M against inc_a's $1,684.7M — a $921.5M gap that is impairments and restructuring-related charges flowing through the cash-flow reconciliation. In fiscal 2024 the two figures were identical at $1,592.4M. The gap appeared in the restructuring year and it is the size of the non-cash charges taken.

4. Balance sheet, cash flow and the capital-allocation problem

Filing-verified, 28 June 2026 (10-Q filed 2026-07-29), against the 28 September 2025 year-end:

And from the cash-flow statement for the nine months: repayments of long-term debt of $2,815.9M against new issuance of $1,748.5M. The $3.1 billion of China proceeds went substantially to debt reduction.

The full-year balance sheet at 28 September 2025 from bal_a, which is what every screen still shows:

Two structural points. First, $10,536.7 million of the $26,611.5 million of total debt is lease obligations — for a company operating roughly 40,000 leased stores that is a genuine obligation and belongs in the leverage calculation, and netDebtToEBITDATTM of 3.23x is computed on that basis and is the right number. Second, the shareholders' deficit is an artefact of a decade of buybacks — retained earnings are minus $8,272.5 million — and it makes priceToBookRatioTTM (−15.6x), debtToEquityRatioTTM (−2.92x) and returnOnEquityTTM all meaningless. They are rejected as lenses. returnOnInvestedCapitalTTM of 9.05% is the honest return measure and it is low, reflecting the depressed earnings.

The enterprise value. km_ttm.enterpriseValueTTM of $138,650,100,000 implies net debt of $18,984,300,000, against the September 2025 annual figure of $23,391.7M. The difference is consistent with the post-divestiture balance sheet: $2.8 billion of long-term debt repaid plus the removal of Chinese store leases with the disposal group. We accept the vendor EV as approximately reflecting the June 2026 position and note that we cannot reconcile it precisely because the 10-Q's lease disclosure was not extracted.

Cash flow, and the dividend that is not covered

Read the last row. In fiscal 2025 Starbucks paid $2,771.4 million of dividends out of $2,442.0 million of free cash flow — a shortfall of $329.4 million — and repurchased exactly nothing. The gap was funded from the balance sheet. freeCashFlowToEquityTTM in the vendor file is MINUS $53 million, which is the same statement in a different form: after interest, debt service and dividends, the trailing twelve months produced nothing for equity holders.

The dividend is $2.47 per share, a 2.35% yield, and dividendPayoutRatioTTM is 213% of trailing GAAP earnings. On the fiscal 2026 consensus adjusted EPS of $2.582 the payout would be 96%; on the fiscal 2027 consensus of $3.140 it would be 79%. The dividend becomes covered on an adjusted basis only if the margin recovery in Section 3 actually happens.

We are not forecasting a dividend cut and nothing in the filings suggests one. But a 213% GAAP payout ratio, a suspended buyback, negative equity of $7.7 billion, 3.23x net leverage and a store-refurbishment capital programme running at 6.2% of revenue are five constraints pulling on the same cash, and a reader owning this for the 2.35% yield should know that the yield is currently being funded by the recovery rather than by the earnings.

5. Valuation — priced in or room?

At $104.97 (market cap $119.67B, ~1,140M shares, enterprise value ~$138.65B):

Estimate coverage is good through FY2028 and thins after: 22-25 analysts on the near revenue lines, 16 and 20 on FY2026 and FY2027 EPS, 11 on FY2028 and 8 on FY2029. The FY2029 and FY2030 rows are not used for any conclusion.

est.ebitdaAvg and est.ebitAvg are fixed-ratio fabrications and are REJECTED. ebitdaAvg is exactly 19.88% of revenueAvg in every year from FY2026 to FY2030; ebitAvg is exactly 15.29%. Starbucks' actual fiscal 2025 EBITDA margin was 14.5% and operating margin 9.6%. The vendor's FY2026 ebitdaAvg of $7,537M overstates a realistic figure by roughly a third. All forward valuation runs on epsAvg.

est.epsAvg validates. FY2026's $2.582 against the three reported adjusted quarters ($0.56 + $0.50 + $0.85 = $1.91) plus the Q4 estimate of $0.69 gives $2.60, within 0.7%. The EPS row is reliable and carries the entire forward valuation.

Peer context. The vendor peer set is a consumer-discretionary grab-bag — Airbnb, Carvana, Marriott, MercadoLibre, Nike, O'Reilly, Ferrari, Royal Caribbean, Sea Limited — of which only Chipotle (CMG) is a genuine restaurant comparable, and it is a particularly relevant one because Brian Niccol was its chief executive before joining Starbucks. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn. Note that Lowe's is also in this batch and lists Starbucks in ITS peer set, which is a comment on how the sets are constructed rather than on the businesses.

5a. What today's price assumes (the inversion)

At $104.97 — 33.4x FY2027 consensus and 28.1x FY2028 — the price embeds:

• Adjusted EPS reaches $3.140 in fiscal 2027 and $3.730 in fiscal 2028 — +21.6% then +18.8% — (consensus; 20 and 11 analysts) — on revenue growth of only 1.6% and 5.3%. The entire earnings path is margin recovery.
• Operating margin recovers from 9.6% toward roughly 13-14%. (Our number, derived from the consensus EPS path.) Each point is approximately $0.25 of EPS; getting from a fiscal 2025 $1.63 to a fiscal 2028 $3.73 requires roughly 400-450 basis points of margin, on top of modest revenue growth. The June quarter delivered the first 60 basis points.
• Comparable transactions stay positive. (Filing-confirmed at +4.2%.) This is the most fragile assumption in the price, not because there is any current evidence against it, but because the entire margin-leverage argument depends on volume growth, and the comparison base gets harder every quarter from here as the turnaround laps itself.
• The dividend is maintained without further balance-sheet damage. (Our number.) $2.47 per share is $2.8 billion a year against $2.4 billion of fiscal 2025 free cash flow, negative equity of $7.7 billion and 3.23x net leverage.
• The market keeps paying roughly 28-33x forward earnings for a low-single-digit revenue grower. (Our number.) At 24x FY2028E the stock is $90; at 34x it is $127.

5b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+18.8%, from FY2027E $3.140 to FY2028E $3.730) + multiple drift (COMPRESSION, from 33.4x to roughly 30x on the respective forward year, −10%) + shareholder yield (+2.35%, all dividend, and not covered by earnings) ≈ +9% to +11%.

Our base assumes compression, and the reason is the nature of the earnings rather than a view on the business. A 33.4x multiple is appropriate to a company compounding earnings in the high teens indefinitely. Starbucks' consensus earnings growth is recovery growth: it is 21.6% and 18.8% while the margin repairs and roughly 17% thereafter on a revenue base growing 5-7%. Once the margin is repaired, the sustainable growth rate is revenue growth plus modest leverage — call it high single digits. The multiple should converge toward that at some point, and applying a little of that convergence now is the conservative treatment.

Almost all of the expected return is earnings recovery, and the yield leg is weak because the dividend is not covered. If the multiple simply held at 33.4x on FY2028E the price would be $125 (+18.7%); at 26x it would be $97 (−7.6%) — which is precisely the street's low target. The spread is 26 percentage points on a multiple assumption.

5c. Variant perception (where we differ, what would surprise)

• We differ from the reported revenue line, and the filing settles it in the company's favour. Every screen shows Starbucks' revenue falling 1% in the June quarter. The business grew comparable sales 7.9% with transactions up 4.2%; the decline is $776 million of Chinese store revenue moving to a licensed model. A reader who does not open the 10-Q sees a shrinking company. It is not shrinking; it has been restructured.
• We differ from the standard balance-sheet reading in the company's favour too. bal_a shows long-term debt of $14,575.9M at 28 September 2025. The 10-Q shows $11,780.2M at 28 June 2026 after $2,815.9M of repayments funded by the China proceeds. Every leverage screen on this name is nine months and $2.8 billion out of date.
• We differ from the market on the dividend's security, and this is our only genuinely negative variant view. A 2.35% yield on a consumer staple-like brand reads as safe. It is 213% of trailing GAAP earnings, it exceeded free cash flow in fiscal 2025 by $329.4 million, the buyback has been cut to exactly zero to fund it, and equity is negative $7.7 billion. It becomes covered only if the margin recovery lands. We are not forecasting a cut; we are saying the yield is a claim on a recovery rather than on current earnings, and it is not priced that way.
• The knowledge base made a correct, dated, falsifiable call on this name and it deserves to be recorded. A 2025-07-29 claim states: "the real overhang is China — finding a Chinese JV partner is the key catalyst." On 30 March 2026 Starbucks completed exactly that transaction with Boyu Capital. That is the clearest instance in this batch of a knowledge-base claim specifying a catalyst that subsequently occurred, and it materially raises our confidence in that channel's judgement on this name — though the same claim was also "skeptical on green apron service", which the 7.9% comp appears to refute.
• Positive surprise that would force a re-rate: comparable transactions above 4% again on 2026-10-28; operating margin above 12%; the buyback resuming from zero; or separate disclosure of the licensed China royalty stream showing it materially more profitable than the consolidated operation it replaced.
• Negative surprise that would break the thesis: comparable transactions turning negative, which would mean the turnaround was a price and marketing effect; operating margin failing to progress beyond 10.5%; any change to the dividend; a write-down of the retained 40% China interest; or fiscal 2027 guidance below roughly $2.90 of adjusted EPS.

Synthos fair values

All three anchors are multiples of the fiscal 2028 (year ending September 2028) consensus adjusted-EPS distribution — mean $3.730, low $3.025, high $4.314 — because fiscal 2028 is the first year in which the margin repair is substantially complete on the consensus path, and fiscal 2026 and 2027 earnings are transitional. Cross-checked against the better-covered FY2027E of $3.140 (20 analysts).

• Bear ~$73 — 24.1x the fiscal 2028 consensus LOW of $3.025. Cross-check: 25.4x the FY2027 consensus low of $2.868; roughly 7% below the 52-week low of $78.46. The scenario: comparable transactions revert to flat or negative as the turnaround laps itself; the labour investment proves structural rather than transitional and operating margin stalls near 10-11%; the dividend is cut or held while the balance sheet deteriorates further; and a low-growth restaurant with negative equity and 3.2x leverage earns a mid-twenties multiple on unrepaired earnings. −30.5%.
• Base ~$112 — 30.0x the fiscal 2028 consensus MEAN of $3.730. Cross-check: 35.7x FY2027E; a 2.2% free-cash-flow yield on the fiscal 2025 figure. Sensitivity, stated openly: 28x gives $104 — essentially spot — and 33x gives $123. The whole answer lives in a 28-33x band and the low end is beneath the current price. The scenario: comps moderate to mid single digits, operating margin grinds toward 13%, the dividend is maintained and becomes covered on an adjusted basis, the buyback stays suspended, and the multiple compresses slightly as recovery growth normalises. +6.7%, plus the 2.35% dividend.
• Bull ~$145 — 33.6x the fiscal 2028 consensus HIGH of $4.314. Cross-check: 41.9x FY2027E; 33.7% above the 52-week high of $108.37 and 11.5% above the street's high target of $130. The scenario: comparable transactions stay above 3% for two more years, operating margin returns to the fiscal 2024 level of 15.0% (worth roughly $1.35 of EPS on its own), the licensed China model proves materially more profitable than the operation it replaced, the buyback resumes, and the market pays a full brand multiple for a restored franchise. +38.1%.

Base is 6.7% above spot; asymmetry roughly 1.25:1 to the upside (30.5% down, 38.1% up). What holds this at Watch rather than Buy is that the base returns 6.7%, sits below the street's $115.15, and rests on a margin recovery whose first instalment was 60 basis points against roughly 450 required — while the dividend that supports the yield case is not currently covered by earnings.

6. Knowledge base — 20 claims, nine of them a misattributed 2018 management interview

Raw entity hits: 20. Excluded as misattributed management voice: 9 (45%). Excluded by quarantine policy: 2. Discarded as off-theme: 7. Usable independent claims: 5.

The misattribution, handled first, because it is the finding

Nine of the twenty claims naming Starbucks — 45% — are filed under a channel tagged eli_lilly_ceo, are all dated 2018-06-20, and are unmistakably a Starbucks management interview. Their content settles it:

> "Streamlined the company, freed $8B capital, raised dividend/buyback; transitioning to disciplined, data-driven 'growth at scale' focused on core priorities." (bullish 70)

> "Maintains high-single-digit revenue growth this quarter and long-range 3-5% comp guidance despite depressed comps; admits execution has been inconsistent." (bullish 60)

> "US Q3 comps guided to just 1% (pruning 150 stores vs normal 50, still opening 400); post-May 29 US rebounded to ~3% comp trending into June." (neutral 45)

No third-party commentator has intra-quarter US comparable-sales trends "post-May 29", and no one but the company describes its own execution as inconsistent while reaffirming long-range guidance. This is the attribution-crisis failure mode in its textbook form: a claim set correctly extracted but attributed to the wrong channel entirely, and by channel rather than by speaker.

All nine are excluded on three independent grounds: the channel attribution is wrong; the content is management voice, which house rules exclude from conviction pools; and the vintage is 2018, eight years stale, describing a company with a different chief executive, a different China strategy and a different capital structure. They are reported here rather than silently dropped, because a mechanical read of this lane would show a densely covered, strongly bullish name where 45% of the evidence is the company talking about itself in 2018 under someone else's name.

A further two claims sit in the quarantine_misattributed bucket and are excluded by policy without further consideration.

The off-theme cluster — seven discards

Seven claims, dated 2022-2023 and spread across real_vision (3), arthur_hayes (2) and the two quarantined records, are about "Starbucks Odyssey" — a web3 loyalty and non-fungible-token programme that Starbucks has since discontinued:

> "Starbucks Odyssey shows web3 digital collectibles can layer community, storytelling and gamification onto an existing loyalty program to drive engagement." (real_vision, 2023-08-16, conviction 75)

> "Every company will launch NFTs to engage and incentivize their community — like Starbucks and legacy brands such as the Boston Globe." (real_vision, 2022-09-12, conviction 72)

These are claims about crypto adoption that use Starbucks as an example. They contain no view on the equity and are discarded in full. Two further text-only discards: a claim about Brazilian franchises bought out of bankruptcy — a licensee, not the company — and a bitcoin claim using "priced on the Starbucks board" as a metaphor for retail payment.

The five usable independent claims

> 2025-07-29 · bullish · conviction 55 · horizon: thesis · entity: Starbucks · channel: compound_and_friends · no named speaker

> "Owns Starbucks; turnaround skeptical on 'green apron' service but the real overhang is China — finding a Chinese JV partner is the key catalyst."

This is the most valuable claim in this batch's entire knowledge-base work, and it should be graded as such. Made on 29 July 2025, it identified the single decisive corporate event — a Chinese joint-venture partner — eight months before Starbucks completed exactly that transaction with Boyu Capital on 30 March 2026. It is dated, specific, falsifiable and it was right. It was also wrong on the other half: the same speaker was "skeptical on green apron service", and the Green Apron Service standard is named by the 10-Q as a driver of the 7.9% comparable-sales increase. A claim that got the structural call right and the operational call wrong is more informative than one that was simply bullish, and it is weighted accordingly.

> 2025-07-01 · bullish · conviction 80 · horizon: principle · entity: Starbucks · channel: invest_like_the_best · no named speaker

> "Starbucks is a phenomenal innovation/franchise machine; affordable premium worth the price is a winning red-thread strategy."

> 2025-08-09 · BEARISH · conviction 68 · horizon: principle · entities: China, BYD, Luckin Coffee, Starbucks · channel: all_in · no named speaker

> "China wields state-sponsored corporatism — subsidizing BYD, Luckin — to undercut US brands; justifies aggressive tariffs, but only after boxing China out via free-trade networks."

That third claim is the structural explanation for why the China divestiture was necessary, and it names the specific competitor — Luckin Coffee — that took share. It is bearish and it is corroborated by the outcome: Starbucks sold 60% of a business it had spent twenty years building.

> 2022-10-24 · bullish · conviction 70 · horizon: principle · entities: Costco, Starbucks · channel: we_study_billionaires

> "Understand businesses via daily-life interaction (Lynch-style) — Costco membership appeal and Starbucks demand reveal durable competitive moats."

> 2023-07-15 · bullish · conviction 45 · entities: CSCO, SBUX, KO · speaker: ray_dalio · speaker_role: independent

> "Recommends holding Cisco, Starbucks and Coca-Cola as reliable long-term holdings, citing dividends and low beta."

Note the irony in that last one, which is why it is included. It is the only named-speaker claim in the lane, and its stated rationale — dividends and low beta — is precisely what the capital-allocation analysis in Section 4 puts under question: the dividend is 213% of trailing GAAP earnings and exceeded free cash flow in fiscal 2025. The beta at 0.969 is fine; the dividend rationale is weaker than the claim assumes.

Concentration test and conclusion

A concentration sensitivity test is not required on the usable set, because it is five claims from five distinct channels — the widest breadth-per-claim ratio in this batch. What the test WOULD have shown before exclusion is the point: eli_lilly_ceo supplied 45% of the raw lane, and removing it changes a densely covered bullish name into a thin one.

Conclusion: conviction Low, breadth 5, net positive-low after the misattribution exclusion. The lane's real value is not its aggregate stance but one correctly specified catalyst call that came true, one structural bearish claim about China that was also right, and a reminder that 45% of a lane can be a company's own management voice filed under the wrong name.

7. Data integrity — what we rejected and why

Ten findings.

1. THE MAJOR FINDING — the China divestiture appears in NO vendor field. The 10-Q filed 2026-07-29 discloses the completion on 30 March 2026 of the sale of 60% of Starbucks China retail to Boyu Capital at a cash-free, debt-free enterprise value of approximately $4 billion, with $3.1 billion of consideration received, a 40% retained equity-method interest of ~$1.2 billion, $3.4 billion of net assets derecognised, and $282.8 million of cumulative translation adjustment losses plus $99.7 million of net investment hedge losses reclassified. bal_a ends 28 September 2025 and cf_a likewise; seg_geo still shows China at $3,160.8M. This is the reason reported revenue fell 1% in a quarter when comparable sales rose 7.9%, and it is invisible to every screen.

2. The Q3 FY2026 gross profit in inc_q is impossible — REJECTED. The June 2026 quarter reports gross profit of $6,494.1M on revenue of $9,322.7M — a 69.7% gross margin — against 20.1% in the March quarter, 15.6% in December and 24.2% for fiscal 2025. It is also internally impossible: gross profit of $6,494.1M with operating income of $980.4M would require $5,513.7M of operating expenses in a single quarter, against $5,400.7M for the whole of fiscal 2025. The figure is a cost-of-revenue mapping error. It also corrupts ratios_ttm.grossProfitMarginTTM, which reads 31.75% against a fiscal 2025 actual of 24.2%. Both are rejected and no gross-margin figure from the vendor is used.

3. bal_a is nine months and $2.8 billion out of date on debt — filing used. bal_a shows long-term debt of $14,575.9M at 28 September 2025. The 10-Q shows $11,780.2M at 28 June 2026, after $2,815.9M of repayments against $1,748.5M of issuance. Every leverage screen on this name reflects the pre-divestiture balance sheet. The shareholders' deficit likewise improved from −$8,096.6M to −$7,674.3M.

4. est.ebitdaAvg and est.ebitAvg are fixed-ratio fabrications — REJECTED. ebitdaAvg is exactly 19.88% of revenueAvg and ebitAvg exactly 15.29% in every year from FY2026 to FY2030, against an actual fiscal 2025 EBITDA margin of 14.5% and operating margin of 9.6%. The FY2026 ebitdaAvg of $7,537M overstates a realistic figure by roughly a third.

5. Book-equity metrics are meaningless — REJECTED as lenses. priceToBookRatioTTM (−15.65x), debtToEquityRatioTTM (−2.92x), bookValuePerShareTTM (−$6.70) and returnOnEquityTTM (null in ratios_ttm, −0.243 in km_ttm — the two blocks disagree again) are all artefacts of a −$8,272.5M accumulated deficit produced by a decade of buybacks. returnOnInvestedCapitalTTM of 9.05% is used instead.

6. The two quarterly tax rates that look like errors are real and are explained by the filing. The September 2025 quarter shows income tax of $700.6M on pre-tax income of $833.7M — an 84.0% effective rate — and the December 2025 quarter $471.6M on $764.8M, 61.7%. The 10-Q explains both: "We also changed our indefinite reinvestment assertions upon classification as held for sale, resulting in an increase in our income tax expense." These are real charges, correctly reported, and would be unintelligible without the filing.

7. cf_a and inc_a depreciation disagree by $921.5M in fiscal 2025 — both are correct and the gap is informative. inc_a reports D&A of $1,684.7M; cf_a reports $2,606.2M. In fiscal 2024 the two were identical at $1,592.4M. The gap appeared in the restructuring year and represents impairment and restructuring charges flowing through the cash-flow reconciliation. We use the cash-flow figure for cash purposes and note the difference as a measure of the non-cash charges taken.

8. dividendPayoutRatioTTM of 213% is correct and is the most important ratio in the file — reported, not rejected. $2.47 of dividend against $1.733 of trailing GAAP diluted EPS. In fiscal 2025 dividends of $2,771.4M exceeded free cash flow of $2,442.0M by $329.4M, and freeCashFlowToEquityTTM in the same file is MINUS $53 million. Correctly computed, and the reason it is highlighted rather than corrected.

9. tech.ret_3m reads exactly 0.00% — flagged as unverifiable. A precise zero three-month return is possible but improbable. The six-month (+14.5%) and twelve-month (+20.8%) figures are ordinary. We use the six- and twelve-month figures and treat the three-month reading as uninformative rather than asserting it is wrong.

10. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — tech used. Quote: $109.23 / $77.99. Tech: $108.37 / $78.46. Intraday versus close. We use tech throughout. Note tech.pct_from_hi is −3.1% while tech.max_dd_from_peak is −16.7% — correctly different, because the six-year peak predates the last twelve months. Both are right.

Not defects, correctly reported and independently confirmed: seg_prod and seg_geo both sum to reported revenue exactly in every year shown, with the 72.9% United States home market present and China separately identified — this is one of the cleanest segment reconciliations in the batch; inc_q June-quarter revenue of $9,322.7M matches the earn_cal actual exactly; dividendPerShareTTM of $2.47 is consistent with the payout data; capexToRevenueTTM of 5.06% is consistent with $2,305.5M on $37,184.4M for the annual period; interestCoverageRatioTTM of 6.21x is consistent with the debt load; profile.currency and inc_a[0].reportedCurrency are both USD — no currency mixing; and the earn_cal adjusted basis is consistent across all five reported rows, so the three-miss-then-two-beat series is like-for-like.

8-K scan — completed. The three retained 8-Ks are 2026-05-20, 2026-06-12 and 2026-07-29, all short (2,952 to 4,430 characters) and all cover documents without extracted exhibits. No corporate action postdating the 28 June 2026 balance sheet appears. The China divestiture of 30 March 2026 was found in the 10-Q, not the 8-Ks, because the announcing and completing 8-Ks predate the three retained here.

Non-equity tripwire — checked and passed. SBUX is common stock, NASDAQ-listed. The $104.97 price is not par-like; beta is 0.969; the dividend is variable and has been raised over time rather than fixed; volume was 9.20M shares (~$966M of turnover); and the 52-week band of $78.46 to $108.37 is a 38% range. This is common equity. The negative book value prompted the check and is a buyback artefact.

8. Technicals

• Price $104.97, up 1.55% on the day. −3.1% from the 52-week high of $108.37; +33.8% above the low of $78.46; −16.7% from the six-year peak. Position within the annual range: 88.5th percentile.
• Moving averages, both below: +2.6% above a 50-day of $102.30; +10.4% above a 200-day of $95.11. Both rising.
• RSI 49.6 and MACD +0.37 — neutral. Beta 0.969.
• Relative performance: 3-month exactly 0.00% (flagged in Section 7) vs SPY +7.6%; 6-month +14.5% vs SPY +11.1% — outperformance; 12-month +20.8% vs SPY +24.3% and QQQ +30.8% — slight underperformance.
• Sentiment, and it is the most evenly divided in this batch: 28 buy, 28 hold, 3 sell across 59 analysts. Consensus $115.15 (+9.7%), median $112 (+6.7% — identical to our base), high $130 (+23.8%), low $97 (−7.6%).

What the technical picture does to the entry. The stock has done its recovering: it is at the 88th percentile of its annual range, 34% above the low, and 3% from the high. The neutral RSI and MACD say the move has paused rather than reversed. What the chart does not offer is a discount, and what the analyst split — exactly 28 against 28 — says is that the informed community regards this as fairly valued. Our base of $112 is identical to the analyst median. On a name where our number, the median target and the current price sit within 7% of each other, there is no informational edge to act on.

9. Insiders

The insider block for Starbucks contains no transactions that support inference in the records available. Reported as an absence of data rather than as a signal.

The company's own capital behaviour is the substitute and it is unambiguous: buybacks of exactly $0 in fiscal 2025, against $1,266.7M in fiscal 2024, $984.4M in fiscal 2023 and $4,013.0M in fiscal 2022. A company that has repurchased $6.3 billion of stock in three prior years and none in the most recent one is telling you where its cash priorities are: the dividend, the store programme, and the balance sheet. The $3.1 billion of China proceeds went to debt reduction rather than to buying back shares at what was, for much of the period, a materially lower price.

10. Verdict, kill-criteria and flip conditions

Watch.

What is genuinely good, and this is the strongest operating evidence in the batch: global comparable store sales up 7.9% with United States comparable TRANSACTIONS up 4.2% — customer counts rising, which is the only way a restaurant turnaround can be proved and which no other name here can claim; consolidated operating margin expanding 60 basis points to 10.5% while still absorbing the labour and restructuring investment; two consecutive large earnings beats (+17.6% and +28.8%) after three consecutive misses; the China structural risk removed via a $4 billion-valuation transaction that delivered $3.1 billion of cash, retained a 40% economic interest, and converted a besieged company-operated business into an asset-light licence; long-term debt cut from $14,575.9M to $11,780.2M; a shareholders' deficit that is improving; and a brand that independent voices describe as an innovation and franchise machine.

What holds us back: 60.6x trailing GAAP earnings and 33.4x the FY2027 consensus for a company consensus expects to grow revenue 1.6% and 5.3%; a base fair value of $112, just 6.7% above spot, below the street's $115.15 and identical to the analyst median of $112; a street split exactly 28 buy against 28 hold; a stock at the 88th percentile of its annual range, 3.1% from its high; a dividend of $2.47 that is 213% of trailing GAAP earnings and exceeded fiscal 2025 free cash flow by $329.4 million, with freeCashFlowToEquityTTM at MINUS $53 million; a buyback suspended at exactly zero; shareholders' equity of minus $7,674.3 million and net leverage of 3.23x; and an operating margin that must travel roughly 450 basis points to justify the consensus earnings path, of which the first 60 have been delivered.

The distinction that matters. This is not a case where we doubt the business — the June quarter is genuinely impressive and the transaction that removed the China overhang was well executed and correctly anticipated by an independent voice in our own knowledge base eight months before it happened. It is a case where the market has already paid for the recovery. The stock is 3% from its high, our number equals the analyst median, and the yield that makes it look defensive is currently funded by the recovery rather than by earnings. With 450 basis points of margin repair still required and 60 delivered, waiting for the next two prints costs 6.7% of base-case return and buys a great deal of confirmation.

Pre-registered KILL criteria — what would take this to Avoid:

• Comparable TRANSACTIONS turning negative in any quarter. This is the number that distinguishes a real turnaround from a price-and-marketing effect, and it is currently +4.2%.
• Operating margin failing to progress beyond roughly 11% through fiscal 2027, which would mean the labour investment is a permanent structural cost rather than a transitional one and would remove most of the consensus EPS path.
• Any reduction to the dividend, which at 213% of GAAP earnings and above free cash flow is the most likely place a balance-sheet problem would surface.
• A write-down of the retained 40% China interest (~$1.2 billion), which would say the transaction was a rescue rather than a strategic choice.
• Fiscal 2027 guidance below approximately $2.90 of adjusted EPS, against the $3.14 consensus.
• A break below the 200-day average of $95.11 on company-specific news.

Pre-registered FLIP conditions — what would take this to Buy — Tactical:

• Price below approximately $92 (24.7x FY2028E, 29.3x FY2027E) with comparable transactions still positive. At $92 the base case is +21.7% against a bear case only 21% lower, and the arithmetic becomes genuinely asymmetric. That is 12.4% below today's close and just below the 200-day average.
• Operating margin above 12% in the 2026-10-28 print, which would put the fiscal 2028 consensus of $3.73 comfortably in reach and validate the entire margin-repair thesis two years early.
• A second consecutive quarter of comparable transactions above 4%, confirming the customer-count recovery is durable rather than a marketing pulse.
• The buyback resuming at any scale, which would signal the board considers the balance sheet repaired and the dividend covered.
• Separate disclosure of the licensed China royalty stream showing materially higher margins than the consolidated operation it replaced, which would re-rate the quality of the remaining revenue.

Where SBUX fits in the Synthos Framework Portfolio. The consumer-brand / turnaround sleeve, at 0% today with a 1.5-2% target on a fill near $92 or on operating margin clearing 12%. Batch overlap note: Lowe's lists Starbucks in its own vendor peer set and is in this batch, which is a comment on peer-set construction rather than on the businesses — but the two make an instructive pair on exactly one axis, the comparable-transaction line: Starbucks +4.2%, Lowe's −0.9%. Both are consumer-facing recoveries priced as if they are working; only one has rising customer counts. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $104.97.

Single biggest risk: the dividend. $2.47 per share is 213% of trailing GAAP earnings. In fiscal 2025 it cost $2,771.4 million against free cash flow of $2,442.0 million, and the vendor's own freeCashFlowToEquityTTM of minus $53 million says that after interest, debt service and distributions the trailing twelve months produced nothing for shareholders. It is being paid by a company with shareholders' equity of minus $7,674.3 million, net leverage of 3.23x, a suspended buyback, and a capital-expenditure programme running at 6.2% of revenue that the store-refurbishment strategy requires. The dividend becomes covered on adjusted earnings only if operating margin travels from 10.5% back toward 13-14%. That recovery is plausible, is underway, and is the entire investment case — but until it lands, a 2.35% yield that looks like a defensive characteristic is in fact a claim on a turnaround, and the market is not pricing it that way.

Provenance & disclosures

• Traceability: 20 raw knowledge-base entity claims name Starbucks, of which only FIVE are usable. NINE (45%) are filed under a channel tagged eli_lilly_ceo, are all dated 2018-06-20, and are unmistakably a Starbucks MANAGEMENT interview — the content includes intra-quarter US comparable-sales trends "post-May 29", an admission that "execution has been inconsistent" alongside reaffirmed long-range guidance, and a description of having "streamlined the company, freed $8B capital". This is the attribution-crisis failure mode: correct extraction, wrong channel, channel-level rather than speaker-level attribution. All nine are excluded on three independent grounds — misattributed channel, management voice, and eight-year vintage. A further two sit in the quarantine_misattributed bucket and are excluded by policy. Seven more are discarded as off-theme: a 2022-2023 web3 and non-fungible-token cluster about the discontinued "Starbucks Odyssey" loyalty programme (real_vision ×3, arthur_hayes ×2, the two quarantined records), plus a claim about Brazilian franchises bought out of bankruptcy (a licensee, not the company) and a bitcoin claim using "priced on the Starbucks board" as a retail-payment metaphor. The five usable claims come from five distinct channels — the widest breadth-per-claim ratio in this batch — and the standout is compound_and_friends, 2025-07-29, bullish, conviction 55: "the real overhang is China — finding a Chinese JV partner is the key catalyst." Starbucks completed exactly that transaction with Boyu Capital on 30 March 2026, eight months later. That is the clearest instance in this batch of a knowledge-base claim specifying a catalyst that subsequently occurred, and it is recorded as such — while noting that the same claim was "skeptical on green apron service", which the 7.9% comparable-sales increase appears to refute. The remaining four are invest_like_the_best (2025-07-01, bullish 80, on the franchise), all_in (2025-08-09, BEARISH 68, on Chinese state subsidy of Luckin undercutting US brands — the structural reason the divestiture was necessary), we_study_billionaires (2022-10-24, bullish 70, a Lynch-style moat principle) and ray_dalio (2023-07-15, bullish 45, the only NAMED speaker in the lane, speaker_role: independent), whose stated rationale of "dividends and low beta" is directly questioned by the payout analysis in Section 4. No concentration test was required on the usable set because it is five claims from five channels; the test would have been decisive on the raw lane, where one channel supplied 45%. All quotes are verbatim from the stored claim text.
• Data as-of: most recent filing-verified financials 2026-06-28, from the 10-Q filed 2026-07-29 (balance sheet, comparable-store-sales detail, the China divestiture note, the held-for-sale accounting explanation, and the revenue bridge) · annual statements fiscal 2025 ended 2025-09-28 from inc_a/bal_a/cf_a, corroborated by the 10-K filed 2025-11-14 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873601 = 2026-08-04T20:00:01Z ($104.97, +1.55%; 50-DMA $102.30; 200-DMA $95.11; RSI 49.6; MACD +0.37; beta 0.969) · knowledge-base claims 2026-08-04. Fiscal note: Starbucks' fiscal year ends in late September. "Fiscal 2025" means the year ended 2025-09-28 and "third quarter of fiscal 2026" means the three months ended 2026-06-28. All figures come from the Synthos vendor data file for SBUX or from the SEC filings in the SBUX archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2025-11-14 (fiscal 2025); 10-Q filed 2026-04-28 (March 2026 quarter); 10-Q filed 2026-07-29 (June 2026 quarter, containing the China divestiture note and the comparable-store-sales detail used throughout); 8-K filed 2026-05-20; 8-K filed 2026-06-12; 8-K filed 2026-07-29 (Q3 results) — all three 8-Ks are short cover documents (2,952 to 4,430 characters) with no extracted exhibits and no financial data. The China divestiture of 30 March 2026 was found in the 10-Q, not the 8-Ks, because the announcing and completing filings predate the three retained here.
• Where the filings contradicted or corrected the vendor (detailed in Section 7): the 30 March 2026 China divestiture — 60% to Boyu Capital at a ~$4B enterprise value, $3.1B of consideration received, a 40% retained equity-method interest of ~$1.2B, $3.4B of net assets derecognised, $282.8M of CTA losses and $99.7M of net investment hedge losses reclassified — appears in NO vendor field, and is the reason reported revenue fell 1% while comparable sales rose 7.9%; bal_a is nine months out of date on debt, showing $14,575.9M of long-term debt against the 10-Q's $11,780.2M after $2,815.9M of repayments; inc_q gross profit for the June 2026 quarter ($6,494.1M, a 69.7% margin against 24.2% for fiscal 2025) rejected as internally impossible, along with the grossProfitMarginTTM it corrupts; est.ebitdaAvg and est.ebitAvg rejected as fixed-ratio fabrications at exactly 19.88% and 15.29% of revenue in every year against an actual fiscal 2025 EBITDA margin of 14.5%; book-equity metrics rejected as artefacts of a −$8,272.5M accumulated deficit, with returnOnEquityTTM again reported as null in ratios_ttm against −0.243 in km_ttm; the two anomalous quarterly tax rates (84.0% and 61.7%) confirmed as REAL and explained by the filing's held-for-sale reinvestment-assertion change — they would be unintelligible without it; the $921.5M fiscal 2025 gap between cf_a and inc_a depreciation identified as restructuring and impairment charges; tech.ret_3m of exactly 0.00% flagged as uninformative; quote.yearHigh/yearLow rejected in favour of tech.hi52/lo52. Where vendor and filing AGREED — worth recording: seg_prod and seg_geo both sum to reported revenue EXACTLY in every year, with the 72.9% United States home market present and China separately broken out at 8.5% — one of the two cleanest segment reconciliations in this batch; inc_q June-quarter revenue matches the earnings-calendar actual exactly; the earn_cal adjusted basis is consistent across all five rows.
• Basis note: analyst estimates and earn_cal actuals are adjusted (non-GAAP); reported diluted EPS is GAAP, and the gap is very wide — trailing adjusted EPS is $2.43 against trailing GAAP diluted of $1.733, a 40% difference driven by restructuring charges, impairments and the held-for-sale tax effects. The earn_cal basis is consistent across all five reported rows, so the surprise series (−22.7%, −6.5%, −4.4%, +17.6%, +28.8%) is like-for-like and the three-miss-then-two-beat inflection is genuine rather than definitional. Forward multiples use consensus adjusted EPS; the trailing multiple is shown on both bases (43.2x adjusted, 60.6x GAAP diluted). A reader should note that the 40% GAAP-to-adjusted gap is itself evidence of how much restructuring is still running through the accounts.
• Estimate coverage: 22-25 analysts on the fiscal 2026 to fiscal 2028 revenue lines; 16 and 20 on fiscal 2026 and fiscal 2027 EPS; 11 on fiscal 2028 EPS, on which the fair-value anchors rest, and 8 and 9 on fiscal 2029 and 2030, both of which are excluded from every conclusion. The fiscal 2028 anchor's eleven-analyst coverage is disclosed rather than buried; it was chosen over the better-covered fiscal 2027 row because fiscal 2027 earnings are still transitional and fiscal 2028 is the first year in which the margin repair is substantially complete on the consensus path.
• Peer note: the vendor peer set (Airbnb, Chipotle, Carvana, Marriott, MercadoLibre, Nike, O'Reilly, Ferrari, Royal Caribbean, Sea Limited) is a consumer-discretionary grab-bag of which only Chipotle is a genuine restaurant comparable — and a particularly relevant one, since Brian Niccol was its chief executive before joining Starbucks. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn. Lowe's, also in this batch, lists Starbucks in its own peer set, which illustrates how loosely these sets are constructed.
• Fair-value caveat: the $73 / $112 / $145 anchors are multiples of the fiscal 2028 consensus adjusted-EPS distribution — 24.1x the low of $3.025, 30.0x the mean of $3.730, and 33.6x the high of $4.314 — anchored on fiscal 2028 rather than the better-covered fiscal 2027 because fiscal 2026 and 2027 earnings are transitional and the margin repair is only substantially complete by fiscal 2028 on the consensus path. Cross-checked against FY2027E of $3.140. Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 28x gives $104, essentially spot; 33x gives $123. The base assumes mild multiple compression as recovery growth normalises toward the underlying revenue growth rate of 5-7%, and Section 5b states that explicitly.
• Timing: third-quarter fiscal 2026 results were released 2026-07-29, six days before this dive, and the 10-Q filed the same day, so the first nine months of the fiscal year including the completed China divestiture are in the data. The next print is 2026-10-28, 85 days away — the fourth quarter and full year — and there is no near-term company-specific catalyst, which is an explicit input to the Watch verdict. 2026-08-04 was a firm day for the stock (+1.55%) inside a broadly higher market; no company-specific news for that date appears in this file.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.