Philip Morris International PM

Reference table

1. What the business is, and how the disclosure actually works

Philip Morris International describes itself in the 10-Q as "a leading international consumer goods company, actively delivering a smoke-free future." It is a Virginia holding company, headquartered at 677 Washington Boulevard, Stamford, Connecticut, listed on the New York Stock Exchange since its 17 March 2008 separation from Altria. Group Chief Executive Jacek Olczak; approximately 84,900 employees. The common stock has no par value.

The critical structural fact, stated first because it changes how every table below is read: PMI sells nothing in the United States under the Philip Morris cigarette brands — those rights belong to Altria. Its U.S. business is entirely smoke-free: ZYN (acquired with Swedish Match in November 2022), the cigar business, the Aspeya wellness unit, and a limited IQOS roll-out. Everything else is international.

The reportable segments — and the vendor is reporting a taxonomy that no longer exists

This is the single most important disclosure-structure point in the file. The 10-K, at line 3306, states plainly:

> "As communicated in the fourth quarter of 2025, with our smoke-free business now operating at scale across our regions, including substantial growth from our U.S. business, we have implemented an evolved organizational model with two primary business units: International and U.S. ... This change was implemented effective January 1, 2026, and as a result we realigned our reportable segments accordingly. The four geographic segments have been replaced with three new reportable segments: International Smoke-Free, International Combustibles, and U.S."

The vendor's seg_geo block supplies the OLD four geographic segments and stops at fiscal 2025. It has no 2026 data and it never will, because the segments it describes were retired seven months before this dive. The new segments — International Smoke-Free, International Combustibles, U.S. (including the Aspeya wellness unit) — appear nowhere in the vendor file. Everything we say about 2026 segment performance below comes from the 10-Q prose, not the vendor.

Old segments, fiscal 2025 (seg_geo, verified line by line against the 10-K):

Three of the four tie exactly. The fourth does not, and we can name the error. The FY2025 column sums to $40,648M against the 10-K's stated "Net revenues of $40.6 billion for the year ended December 31, 2025" — an exact match. The FY2024 column sums to $37,545M against reported net revenues of $37,878M, a shortfall of $333M. Back-solving from the 10-K's stated Europe growth of 9.1% gives a FY2024 Europe figure of $15.684B, not the vendor's $15.357B — a difference of $327M, which is the shortfall. The vendor is carrying the originally-reported 2024 Europe number rather than the restated one (PMI moved the Wellness business into Europe in January 2025 following the Vectura disposal, and renamed PMI Duty Free to PMI Global Travel Retail). We use the FY2025 column, which ties exactly, and we discard the FY2024 geographic column.

The product split — and the gross-versus-net excise-tax test, which PASSES

On a tobacco name the classic tie failure is a vendor reporting revenue GROSS of excise taxes, which roughly doubles the number. We tested it explicitly.

The 10-K settles it verbatim, at line 4018: "Net revenues include $16.9 billion in 2025 and $14.7 billion in 2024 related to smoke-free." The vendor's $16.854B and $14.660B round to exactly those figures, and the two lines sum to reported net revenues in every year from 2018 forward. This is NET revenue, excluding excise taxes. The vendor is correct.

One legacy row is the exception and it is instructive. The FY2017 row reports Combustible Tobacco of $74.305B, Reduced-risk Products of $3.793B, and a separate Excise Taxes line of −$49.350B, summing to $28.748B. That is the gross presentation PMI used before 2018. Sum the first two lines without the third and you get 2.7x the true revenue. It is confined to the single FY2017 row and no figure in this dive touches it — but a reader taking the vendor's seg_prod block at face value across the full history would produce exactly the 2x-style error the data contract warns about.

Smoke-free went from 23.8% of net revenues in 2020 to 41.5% in 2025. That is the entire investment case in one line, and it is filing-confirmed.

The new segments, from the 10-Q prose (the only source that has them)

"Organic" here is the filing's own phrase — "excluding currency and acquisitions/divestitures."

Geography, brands and concentration

Marlboro accounted for approximately 43% of 2025 cigarette shipment volume, and the five international brands — Marlboro, Parliament, Chesterfield, L&M and Philip Morris — contributed 81%. The company holds at least 15% market share in approximately 100 markets. Cigarettes are sold in approximately 170 markets; smoke-free products in 109 markets as of 30 June 2026 (up from 106 at year-end 2025).

Two country exposures are individually disclosed and both matter. The 10-K states: "In 2025, Russia accounted for around 9% of our total cigarette and heated tobacco unit shipment volume, and around 6% of our total net revenues. Ukraine accounted for around 2% ... and around 1% of our total net revenues." The 10-Q adds the balance-sheet dimension: $2.8 billion of the $6.0 billion cash balance is held in Russia, and Russian countermeasures include "restrictions on lending to foreign borrowers, repatriation of dividends and transactions with securities and real estate involving companies from 'hostile' countries."

Customer concentration: none disclosed. Unlike a semiconductor or aerospace name, PMI's 10-K contains no "customer accounted for 10% of revenue" disclosure — the business sells through direct retail distribution, independent distributors and wholesalers across 170 markets. We searched for it and it is absent, which is itself the finding: this is a diversified-customer business with concentrated COUNTRY and REGULATOR risk instead.

Competitors are not enumerated in a discrete list in the way an equipment maker's are. The filing's competitive language is about "intense competition, including unfair competition from non-tax paid volume by certain manufacturers" — illicit trade — and, in the U.S. pouch category, about competitors marketing products "that were not previously marketed in the United States and did not have a marketing granted order." The vendor peer set names British American Tobacco ($127.5B), Altria ($113.7B), Turning Point Brands, Universal and RLX as genuine comparables, alongside Coca-Cola, PepsiCo, Procter & Gamble, Unilever and AB InBev, which are staples but not competitors. No peer earnings or multiples are supplied in the file, so no peer-multiple comparison is drawn anywhere in this dive.

2. The single most important thing to understand about these numbers: the consolidated growth rate is an average of three businesses moving in different directions, and half of it is the dollar

Read only the top line and PMI looks like a 10%-growth consumer company. It is not one, and the 10-Q says so directly.

Three things follow, and together they are the whole point of this section.

First, roughly 46% of the reported revenue growth in the half was currency. The 10-Q names the drivers: "The favorable currency impact in net revenues was due primarily to the Euro and Russian ruble, partly offset by the Japanese yen." Contrast fiscal 2025, when net revenues rose 7.3% reported and 6.5% organic — a currency contribution of only 0.8 points. The currency tailwind is a 2026 phenomenon, it is large, and it is not a business improvement.

Second, reported earnings per share went DOWN while all of that was happening. First-half diluted EPS of $3.36 fell 8.4%, and the 10-Q is explicit that excluding the $0.22 favourable currency impact it fell 14.4%. The causes are all itemised in the filing:

Read the last two lines together. Operations contributed +$0.21 of EPS in the half and currency contributed +$0.22 — the dollar did as much for reported earnings as the entire operating business did, and both were swamped by the impairment and the equity-securities reversal.

Third, the three segments are not one business. International Smoke-Free grew 13.7% organic with gross profit up 16.9% organic. International Combustibles grew 3.8% organic, entirely on price, against a 1.9% cigarette volume decline. And the U.S. — the ZYN business — fell 16.5% organic with gross profit down 27.8%. A single consolidated growth rate of 5.3% organic tells you nothing about any of them.

Why this section exists. Everything in Sections 3 through 6 is an attempt to value a company whose reported numbers are systematically inflated by currency on the revenue line and systematically deflated by one-off charges on the earnings line, while consensus is struck on a third basis that removes the charges but keeps the currency. We state the basis of every figure we use, every time, for exactly that reason.

3. Growth and margin trajectory

Annual, six fiscal years, from the vendor income series (all figures NET revenues, excluding excise taxes, confirmed against the 10-K):

Two warnings about that table, both material.

Warning one: the FY2024-to-FY2025 net income comparison is worthless as stated. Net income rose 60.8% and diluted EPS rose 60.6% — and essentially none of it was operating improvement. The 10-K explains: "We recorded an impairment charge of $2,316 million related to our RBH equity investment ... for the year ended December 31, 2024." FY2024 also carried a $199 million loss on the sale of Vectura Group, a $180 million restructuring charge and a $45 million Egypt sales-tax charge. Operating income rose 11.1% reported and 9.3% organic — that is the honest number. This is defect class 3 running in reverse: not a one-off gain capitalised as earnings, but a prior-year one-off charge making the current year look transformational. A screen ranking on EPS growth would put PM near the top of the consumer-staples universe on a number that is an accounting artefact.

Warning two: gross margin looks like it recovered and mostly it did, but the 2023 trough was the Swedish Match integration. Gross margin fell from 68.1% (FY2021) to 63.3% (FY2023) as the acquired oral business and its lower-margin device sales were absorbed, then recovered to 67.1% in FY2025 as smoke-free scale arrived. The FY2025 gross margin of 67.1% is the second-best of the six years and the operating margin of 36.7% is the third-best. Trailing twelve months, gross margin is 67.5% and operating margin 37.7% — the best readings in the file.

Quarterly, cross-checked against the earnings calendar:

Four observations.

The Q4 2024 GAAP loss of $0.37 per share is the RBH impairment quarter and is not an operating event. The vendor reports it correctly; a reader scanning the quarterly EPS series without the filing would conclude the company lost money in a quarter when operating income was $3.259 billion.

The GAAP-to-adjusted gap is large and widening. Q2 2026: $1.80 GAAP against $2.20 adjusted — a 22.2% gap. Q1 2026: $1.56 against $1.96, a 25.6% gap. Q4 2025: $1.37 against $1.70, a 24.1% gap. For KLAC in this same batch the gap was 2.7%. Here it is an order of magnitude larger, and every forward multiple in this dive is therefore labelled with its basis. We reconcile the Q2 bridge line by line in Section 6.

Revenue acceleration is real and it is currency-assisted. Year-on-year revenue growth ran +7.1%, +9.4%, +6.8%, +9.1%, +10.4% across the last five quarters. The Q2 2026 organic figure was +7.6% against +10.4% reported.

Q2 2026 operating margin of 40.5% is the best quarterly reading in the file. The 10-Q attributes the 22.0% reported operating-income growth (20.0% organic) to "a favorable pricing variance and lower restructuring charges in 2026, partly offset by unfavorable volume/mix."

Volumes — the growth case, from the filing prose only

Three things in that table are worth stopping on.

The international modern-oral number is much better than it looks. Total international oral smoke-free volume fell 7.0% in Q2 — but that is the legacy Nordic snus business shrinking, and modern oral pouches grew 14.7%, or 26.3% excluding the Nordics. The mix inside the mix is running the right way.

VEEV is the fastest-growing thing in the company and gets almost no attention. Shipments grew 55.1% in the quarter and 72.0% in the half, and the 10-Q states VEEV "now holds the #1 closed pod position in Europe."

Marlboro set a record while the category shrank. In Q2 the international cigarette category volume share was 25.3%, flat year on year, and "Marlboro continued to gain share (up by 0.3 percentage points) matching its record category share of 11.0%." A premium cigarette brand taking record share in a declining, heavily-taxed category is the pricing power that funds the smoke-free transition.

Guidance was RAISED in the July 10-Q, which is unusual for this industry and is a genuine positive: "For the full year 2026, we currently expect a broadly stable to slightly growing (previously broadly stable) total PMI cigarette and SFP shipment volume, with high-single digit SFP shipment volume growth, and a cigarette shipment volume decline of 2% to 3%, (previously around 3%)." We label this as management's own forward statement and half-weight it accordingly — but the direction of the revision is a fact, not a claim.

4. Balance sheet and the enterprise-value rebuild

The balance sheet in the vendor file is two quarters stale — and the vendor's own trailing metrics block is not. This is an unusual configuration and it needs to be stated before any number is used.

• bal_a[0] is dated 2025-12-31, from the 10-K filed 2026-02-06. There have been two subsequent quarters and both 10-Qs are in the archive, the most recent filed 2026-07-24 covering 30 June 2026. Defect class 1 is live on the bal_a block.
• But km_ttm and ratios_ttm are computed on the 30 June 2026 balance sheet, and we can prove it: debtToMarketCapTTM of 0.168588 against a market cap of $291.320B implies total debt of $49.113B, and cashPerShareTTM of $3.8386 implies cash of $5.998B. The 10-Q states: "Our total debt was $49.1 billion at June 30, 2026, and $48.8 billion at December 31, 2025" and "As of June 30, 2026 and December 31, 2025, we had cash and cash equivalents of $6.0 billion and $4.9 billion, respectively." Both match to the rounding of the filing's own disclosure. The trailing block is current; the annual balance sheet is not. We use the trailing block and the 10-Q.

The negative equity — what it is, and the four ratios we reject because of it

PMI has carried a stockholders' deficit for over a decade and the 10-Q's own statement is titled "condensed consolidated statements of stockholders' (deficit) equity." The arithmetic at 31 December 2025:

> Retained earnings $35.400B

> + additional paid-in capital $2.453B

> − treasury stock $35.551B

> − accumulated other comprehensive loss $12.296B

> = PMI stockholders' deficit −$9.994B

It is a legacy of two things and neither is distress: roughly $35.6 billion of stock bought back between 2008 and 2015, before PMI stopped repurchasing; and $12.3 billion of accumulated currency-translation losses from decades of reporting foreign operations in dollars.

Consequently the following four vendor metrics are arithmetically correct and economically meaningless, and we REJECT all four as lenses:

Note also that financialLeverageRatioTTM of −7.95x and debtToCapitalRatioTTM of 1.21x (above 1.0) are corrupted by the same denominator, and that the vendor's own composite rating scores PM 1 out of 5 on BOTH return-on-equity and price-to-book, dragging the overall grade to C+ / 2. That grade is an artefact of a negative denominator, not a judgement about the business.

The honest return measure is returnOnInvestedCapitalTTM of 26.3%, on invested capital of $34.872B. Cross-checks that do not touch equity: return on assets 15.9%, operating return on assets 23.4%, return on capital employed 38.5%. We use 26.3% ROIC throughout.

The enterprise-value rebuild — and it comes out clean

We rebuilt net debt and enterprise value from scratch, including the short-term-investments test that has broken this calculation on other names in this programme.

> Market cap $291.320B (1,558,613,439 shares × $186.91)

> + total debt $49.113B (10-Q: "$49.1 billion at June 30, 2026")

> − cash and cash equivalents $5.998B (10-Q: "$6.0 billion")

> − short-term investments $0 (the balance sheet carries none; the 10-Q's liquidity paragraph names only cash and cash equivalents)

> = rebuilt enterprise value $334.435B

The vendor reports $334.434B. The difference is $1 million, or 0.0003%. Defect class 2 — netDebt ignoring short-term investments — was tested here and PASSES. PMI genuinely holds no short-term investment portfolio; the $2.891B in longTermInvestments is equity-method and equity-security holdings (RBH at $51M, other Level 1 equity securities at $983M, plus equity-method investees), not liquidity. We say this explicitly because on AAPL, BRK-B and KLAC this same test moved enterprise value materially, and reporting a clean pass is as much a finding as reporting a failure.

One small EBITDA correction. ratios_ttm.ebitdaMarginTTM of 42.785% implies trailing EBITDA of $18.203B. Our build is operating income of $16.059B (four quarters: $4.263B + $3.373B + $3.893B + $4.530B) plus depreciation and amortisation of $2.049B, giving $18.108B, a 42.6% margin. The $95M residual (0.5%) is most likely equity-method income, which the vendor appears to include. We use $18.108B and disclose the difference rather than force a reconciliation.

How we derived D&A, because the vendor zeroed it. inc_q reports depreciation and amortisation of $0 for the June 2026 quarter, and an ebitda figure of $4.535B against operating income of $4.530B — implying $5M of D&A for a company that owns factories in dozens of countries. That is a zeroed field, not a fact. We derived the trailing figure from two independent vendor ratios: capexToRevenueTTM of 3.624% on trailing revenue of $42.545B gives capex of $1.542B, and capexToDepreciationTTM of 0.75256 then gives D&A of $2.049B. That implies a June-quarter D&A of $513M, consistent with $505M, $521M and $510M in the three prior quarters. We use $2.049B.

The debt, reconciled against the filings, and the three 8-Ks

Total debt was $49.1 billion at 30 June 2026, up from $48.8 billion at 31 December 2025. The composition, all from the 10-Q:

• Commercial paper: $2.7 billion outstanding at 30 June 2026, against NIL at 31 December 2025. Aggregate programme capacity is $8.0 billion across US and European programmes; the average balance in the first half was $4.7 billion against $3.0 billion across 2025.
• Other bank loans and uncommitted lines: $689 million at 30 June 2026, against $168 million at year end.
• Committed revolving credit facilities: $6.2 billion, entirely UNDRAWN, with no credit-rating triggers, no material-adverse-change clauses and no collateral provisions.
• Euro term loan (Swedish Match acquisition financing): €1.5 billion (approximately $1.7 billion) outstanding, expiring 23 June 2027.
• A shelf registration statement was filed 6 February 2026 allowing debt issuance over a three-year period.

The three 8-Ks in the archive were checked specifically for a financing that post-dates every statement in the file. Here is what they actually contain.

1. The 8-K of 2026-06-29 is a DELEVERAGING event, not an issuance. "On June 29, 2026, PMI prepaid €1.0 billion (approximately $1.1 billion), including a portion of the outstanding principal and accrued interest, under the 5-year tranche of its senior unsecured term loan facility ... Borrowings in the amount of €1.5 billion (approximately $1.7 billion) under the 5-year tranche of the Term Loan Facility remain outstanding, expiring on June 23, 2027." It occurred one day before the 30 June balance-sheet date, so it IS captured in the 10-Q's $49.1 billion. It is NOT captured in the vendor's bal_a, which stops at 31 December 2025.

2. There WAS a first-half 2026 bond issuance, and the numeric table is stripped. The 10-Q states "PMI's debt issuances in the first six months of 2026 were as follows:" followed by a table our extraction did not capture, with a footnote reading "(a) Interest is payable semi-annually, commencing in October 2026" — consistent with an April 2026 pricing. We identified the issues indirectly from the filings' own XBRL tagging: the Q2 10-Q introduces two note series, USDollarNotesDueApril2029 and USDollarNotesDueApril2036, that appear in neither the 10-K (February 2026) nor the Q1 10-Q (April 2026). The 10-Q states the proceeds "have been or will be used for general corporate purposes, including working capital requirements, repayment of commercial paper or to refinance certain of our outstanding notes due in 2026." The amounts are not recoverable from the extracted text and we do not estimate them. What matters for the leverage conclusion is that the NET effect is already in the $49.1 billion total-debt figure the 10-Q states in prose, which is what we use.

3. The 8-K of 2026-07-09 is a governance item, not a financing. It confirms that Massimo Andolina, previously President Europe Region, became Group Chief Financial Officer on 1 August 2026, replacing Emmanuel Babeau, who remains as Strategic Advisor to the Group CEO through 31 March 2027 on a base salary of CHF 1,260,012, with a lump-sum severance equal to that salary, a CHF 393,754 payment in lieu of pro-rated 2027 incentive, full vesting of outstanding equity awards, and a twenty-four month non-competition provision. A Form 3 filed 2026-08-03 shows Andolina holding 71,588 shares. This is the single most significant corporate-governance change in the file and it is worth flagging: the finance function changed hands eleven days after the Q2 print and three days before this dive.

4. The 8-K of 2026-07-22 is the Q2 earnings release and contains NO NUMBERS. It is present as the cover document only; Exhibits 99.1 (the release) and 99.2 (the non-GAAP reconciliations and glossary) were not extracted. This is the known extraction limit and it matters more here than usual, because Exhibit 99.2 is precisely where the adjusted-EPS reconciliation lives. We rebuilt that bridge from the 10-Q's MD&A instead — see Section 6.

What the leverage actually says

Net debt of $43.1 billion against $18.108 billion of trailing EBITDA is 2.38x. Interest coverage is 10.4x. Moody's revised the outlook from Stable to Positive on 14 April 2026. Free cash flow, guided by management to roughly $12.0 billion for 2026 ($13.5 billion operating cash flow less $1.4–1.6 billion capital expenditure), covers the $9.2 billion annual dividend cost 1.3x. That is investment grade and it is manageable. It is not a fortress.

And there is an adjustment we make that the vendor cannot. $2.8 billion of the $6.0 billion cash balance is held in Russia, where the 10-Q states countermeasures restrict "repatriation of dividends." Treating that cash as unavailable lifts net debt from $43.1 billion to $45.9 billion and leverage from 2.38x to 2.54x. We report both, use 2.38x as the headline because it is the reported figure, and flag 2.54x as the economically honest one.

5. Cash flow and capital returns

The most important capital-allocation fact about this company, and it is easy to miss: there is no buyback at all. The 10-K's Issuer Purchases table carries a note that settles it: "Shares repurchased represent shares tendered to us by employees who vested in restricted and performance share unit awards and used shares to pay all, or a portion of, the related taxes." The vendor's cash-flow statements agree — commonStockRepurchased is $0 in FY2023, FY2024 and FY2025. Every dollar of capital return is the dividend, and the share count is therefore flat to slightly rising: 1,552M diluted (FY2023) to 1,556M (FY2024) to 1,558M (FY2025) to 1,558,613,439 outstanding at 17 July 2026.

The dividend — and the vendor's lastDividend label is wrong in the usual way. The vendor reports lastDividend: 5.88 and dividendPerShareTTM: 5.88. That $5.88 is the ANNUALISED rate, not the last declared dividend. The 10-Q settles it: "During the third quarter of 2025, our Board of Directors approved an 8.9% increase in the quarterly dividend to $1.47 per common share. As a result, the present annualized dividend rate is $5.88 per common share."

The last declared quarterly dividend is $1.47. Four quarters at $1.47 is $5.88, so in this instance the annualised and trailing figures happen to coincide — the increase landed at the start of the trailing window. The yield is 3.15%, and the forward run-rate is the same 3.15%, which is NOT the case for most names where this field is mislabelled. But the label is still wrong, and a reader who took $5.88 as a quarterly payment would compute a 12.6% yield.

Cash-flow quality is good and the first half was unusually strong. incomeQualityTTM of 1.264 — operating cash flow comfortably above net income, the opposite of the working-capital drag seen on capital-equipment names. First-half operating cash flow of $5.1 billion against $3.1 billion a year earlier; the 10-Q attributes the $1.9 billion currency-neutral improvement to $1.5 billion of lower working-capital requirements, noting that the 2025 half included the approximately $0.8 billion disputed German heated-tobacco excise surcharge PMI elected to pay in January 2025, plus the final instalment of the 2017 US Tax Cuts and Jobs Act transition tax.

Capital expenditure is low and directed. $1.569 billion in FY2025, 3.9% of net revenues, and the 10-Q states 2026 capex "were primarily related to our ongoing investments in smoke-free product manufacturing capacity" with the full-year guide "predominantly supporting the smoke-free business."

Total shareholder yield is therefore 3.15% — all dividend, no buyback. That is materially more than KLAC's 1.37% in this same batch and it is the single largest component of the expected return. It is also the constraint: with a 76% payout of guided free cash flow and no buyback to suspend, the dividend is the first thing a deleveraging programme or a currency reversal would collide with.

6. Valuation — priced in or room?

Before any multiple: the basis, established from the filings, because the GAAP-to-adjusted gap here is 22%.

Consensus is struck on ADJUSTED DILUTED EPS, INCLUDING currency. We can prove both halves of that statement from the Q2 numbers. The earnings calendar reports an actual of $2.20 for the quarter ended 30 June 2026; the income statement reports GAAP diluted EPS of $1.80. The 10-Q's own items-affecting-comparability list bridges them:

> GAAP diluted EPS $1.80

> + amortisation of intangibles $0.13

> + RBH equity-investment impairment $0.33

> − fair-value gain on India and Sri Lanka equity securities $0.06

> − Egypt sales-tax settlement benefit $0.01

> = $2.19, against the reported adjusted $2.20

The bridge closes to a penny. The adjustments are amortisation, impairments, restructuring, equity-security fair-value movements and certain discrete tax items. Currency is NOT among them — the 10-Q reports currency as a separate +$0.03 driver of the quarter's reported EPS movement, not as an adjustment. So consensus is a reported-currency, adjusted-earnings number. Every forward multiple below is on that basis, and every trailing multiple is shown on both. (Defect class 6, tested and resolved with the filing rather than assumed.)

At $186.91 (market cap $291.320B, 1,558,613,439 shares, rebuilt enterprise value $334.435B):

Estimate coverage is adequate on the near years and collapses beyond FY2028. Eleven analysts on FY2027 and FY2028 revenue; nine on FY2027 EPS, five on FY2028 EPS, and exactly ONE on FY2029 and FY2030 EPS. The FY2029 and FY2030 EPS rows are excluded from every conclusion in this dive.

The est EBITDA rows were tested against actuals and, unusually, they PASS. Per the data contract, 33.5% of files carry an arithmetically impossible EBITDA row. PM's does not. ebitAvg is below ebitdaAvg in all eight years; no year is negative; and the FY2025 row can be checked against the actual: ebitdaAvg of $17.023B against our computed FY2025 EBITDA of $16.929B (operating income $14.933B plus D&A $1.996B) is 0.6% high, and ebitAvg of $15.383B against actual operating income of $14.933B is 3.0% high. Directionally consistent, small, and not corrupt. We report the pass — but per the contract all forward valuation still runs on epsAvg and revenueAvg.

Peer context, stated honestly and briefly. The vendor peer set contains three genuine comparables — British American Tobacco ($127.5B), Altria ($113.7B) and Turning Point Brands ($1.6B) — alongside consumer staples that are not competitors (Coca-Cola $372.4B, Procter & Gamble $351.9B, PepsiCo $190.0B, Unilever $138.0B, AB InBev $166.6B). No peer earnings or multiples are supplied in the file, so NO peer-multiple comparison is drawn. What can be said from the batch table is that PM is the only consumer-defensive name in this twelve-name group and one of only two decliners on the day.

6a. What today's price assumes (the inversion)

At $186.91 — 20.43x FY2027 consensus adjusted EPS and 22.34x FY2026 — the price embeds the following falsifiable claims, each with a number and a date:

• Net revenues compound at roughly 6% a year through 2028, reaching $45.955B in 2027 and $48.938B in 2028. (Consensus-derived; 11 analysts each.) That is a step DOWN from the 9.8% reported growth just delivered, and it is close to the 5.3% organic rate. In other words, consensus is implicitly assuming the currency tailwind goes away and the organic rate persists. The nearest test is the 2026-10-21 print, where consensus revenue of $11.185B would be only +3.1% year on year against the +10.4% just reported — the steepest implied deceleration anywhere in the estimate set.
• Adjusted EPS reaches $9.147 in 2027 and $10.021 in 2028, roughly 9-10% compound. (Consensus; 9 and 5 analysts.) Since revenue grows ~6%, this requires roughly 300-400 basis points a year of margin and below-the-line contribution — from smoke-free mix, from the absence of the 2026 one-offs, and from deleveraging. The tax line works AGAINST it: the effective rate was 19.7% in 2025 and management guides to approximately 21.5% for 2026, which is roughly a 2% headwind to earnings on its own.
• The U.S. segment stops shrinking. (Our number; the filing gives the decline, not the recovery.) The U.S. fell 16.5% organic in the first half with gross profit down 27.8%. Q2 stabilised at −0.9% organic. Consensus of ~6% total revenue growth is not achievable with the U.S. down double digits; it needs the U.S. flat to growing. This is the most fragile assumption in the price.
• The dollar does not strengthen materially. (Our number, derived from the filing's own currency disclosures.) First-half reported revenue growth exceeded organic growth by 4.5 points, and currency contributed +$0.22 to first-half diluted EPS. Consensus is struck on reported-currency adjusted EPS. A dollar rally cuts the consensus number directly, without any deterioration in the business.
• The market keeps paying roughly 20x or better on the forward year. (Our number.) At 17x FY2027E the stock is $155; at 24x it is $220. The multiple, not the earnings, is where the outcome distribution lives — and unusually, we can prove that: the FY2027 consensus EPS range is $9.047 to $9.286, a spread of just 2.6% across nine analysts. The street has almost no disagreement about the earnings and total disagreement about the multiple.

6b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+9.3%, from FY2026E $8.366 to FY2027E $9.147) + multiple drift (a modest HOLD-to-slight-compression, from 22.34x on the forward year today to 22.0x on the then-forward year, −1.5%) + shareholder yield (+3.15%, all dividend) ≈ +10.9%.

Our base case assumes the multiple essentially HOLDS, and we should be explicit about why that is the right call rather than the lazy one. Today the market pays 22.34x on the next calendar year's estimate. In twelve months, if it pays 22.0x on the then-next year's estimate, that is a 1.5% de-rating — near enough to flat. We assume a hold rather than compression because, unlike a cyclical, PMI's consensus growth rate does not decelerate on the forward path: EPS growth runs +9.3%, +9.6%, +10.1% across FY2027, FY2028 and FY2029. There is no growth cliff for the market to start discounting. We assume a hold rather than expansion because the leverage is 2.38x, the equity is negative, the U.S. leg is impaired and the regulatory pipeline in Europe is live — none of which argues for paying more.

Almost all of the expected return is earnings growth plus the dividend, and that is the structurally attractive part of this name. If the multiple simply held at 22.34x on FY2028E the price would be $224 (+19.8%); if it compressed to 17x the price would be $170 (−9.0%). The dividend is the asymmetry buffer: at 3.15%, a full year of multiple compression to 20x on FY2027E ($183, −2.1%) is still roughly break-even in total return. That is a materially different risk shape from a zero-yield growth name, and it is the main reason this is a Hold rather than a Watch.

The bull case at $230 does require modest expansion — 23.0x applied to the FY2028 consensus mean, which is 25.2x on FY2027E, a genuine re-rating. The bear case at $155 requires no earnings miss at all, only a de-rating to 17x on the unchanged FY2027 consensus mean. We are being deliberately symmetric about which leg needs help in which case, because it is the multiple doing the work in both.

6c. Variant perception (where we differ, what would surprise)

• We differ from consensus on the QUALITY of the reported growth, not its level. The street's ~6% revenue growth and ~9-10% EPS growth look right to us. What we think is under-appreciated is that first-half reported growth of 9.8% was only 5.3% organic, and that reported EPS actually FELL 8.4%. Anyone anchoring on the reported top line is carrying an unhedged short dollar position they did not choose. Watchable number: the currency line in the 2026-10-21 release. If reported and organic revenue growth converge, the tailwind is over and consensus for 2027 gets tested.
• We differ, materially, on how to read the balance sheet — but in the opposite direction to most names. Screens show PM with negative book equity, −112% return on equity and −5.72x debt-to-equity, and the vendor's own composite rating scores it 1 out of 5 on two of six sub-metrics as a result, producing a C+ grade. All of that is a negative-denominator artefact and should be ignored. But the correction cuts both ways: the real leverage of 2.38x — 2.54x excluding the $2.8 billion of Russian cash — is genuine and is higher than the sector's reputation for balance-sheet conservatism suggests. We are neither as alarmed as the screens nor as relaxed as the "staple" label implies.
• We think the Canadian litigation resolution is under-credited, and this is our clearest positive variant. The 10-Q reports 7 health-care cost-recovery cases pending at 30 June 2026 against 17 a year earlier, with nine proposed class actions and ten further cost-recovery cases released under the Canadian CCAA plan that became effective 29 August 2025. PMI's carrying value in RBH is now $51 million, down from $569 million, after cumulative impairments of $3.060 billion. The Canadian tail that hung over this name for seven years is closed and the remaining book exposure is fifty-one million dollars. The market treated the May 2026 $511 million impairment as bad news; on our reading it is the last instalment of a bill already paid.
• Positive surprise that would force a re-rate: FDA authorisation of IQOS ILUMA, pending since October 2023 and formally accepted for review in March 2024 — it opens the induction platform in a market the filing describes as having approximately 25 million adult smokers. Or U.S. segment net revenues returning to growth in the 2026-10-21 print with ZYN offtake accelerating on the ZYN ULTRA launch and the June 2026 modified-risk authorisation, which is the only such clearance in the category.
• Negative surprise that would break the thesis: a material dollar rally removing the 4.5-point revenue tailwind and the $0.22 of EPS; a second consecutive down half in the U.S. as unauthorised competitors exploit the FDA's 8 May 2026 non-enforcement guidance; EU Tobacco Excise Directive momentum toward its contemplated 1 January 2028 implementation, which would for the first time bring heated tobacco and nicotine pouches into EU excise scope; or a Russian escalation affecting 6% of net revenues, 9% of volume and $2.8 billion of cash. Any of these hits a stock with 2.38x leverage, no equity cushion and a 76% dividend payout.

Synthos fair values

All three anchors are multiples of the FY2027 consensus adjusted EPS of $9.14662 (9 analysts) or the FY2028 consensus adjusted EPS of $10.02097 (5 analysts), stated explicitly for each. Note that the FY2027 EPS dispersion is exceptionally narrow — low $9.047, high $9.286, a 2.6% spread — so unlike most names we cannot build a range from the estimate distribution. The range below is built from the MULTIPLE, and we say so rather than manufacture false precision.

• Bear ~$155 — 17.0x the FY2027 consensus MEAN of $9.147 ($155.49). Cross-check: 15.5x FY2028E; 8.6x rebuilt trailing EBITDA at an enterprise value of $285B; a 3.79% dividend yield at the unchanged $5.88 rate. The scenario requires NO earnings miss — only a de-rating to a mid-teens tobacco multiple, driven by a dollar rally that removes the reported-growth premium, a second down half in the U.S., and EU excise-directive momentum. −17.1%, cushioned to roughly −14% on a total-return basis by the dividend.
• Base ~$201 — 22.0x the FY2027 consensus MEAN of $9.147 ($201.23). Cross-check: 24.0x FY2026E; 20.1x FY2028E; rebuilt enterprise value of $357B on FY2027E revenue of $45.955B is 7.8x sales, in line with today. Sensitivity, stated openly: 21x gives $192, roughly 3% above spot; 23x gives $210, roughly the street's mean. The whole answer lives in a 21-23x band and we claim no more precision than that. The scenario: organic revenue compounds at 5-6%, smoke-free mix continues to lift gross margin, the U.S. stabilises without recovering fast, the currency tailwind fades to neutral, and the multiple holds. +7.5%, plus 3.15% of dividend — roughly +10.7% total return.
• Bull ~$230 — 23.0x the FY2028 consensus MEAN of $10.021 ($230.48). Cross-check: 25.2x FY2027E; a 2.56% dividend yield. This case DOES require a re-rating, and we name the two things that would justify one: IQOS ILUMA authorised in the United States, and smoke-free crossing 50% of net revenues with U.S. ZYN reaccelerating on the ULTRA launch and the modified-risk claim. Deleveraging below 2.0x would help. +23.3%.

Base is 7.5% above spot and 6.5% BELOW the street's $215 median; asymmetry is roughly 1.36:1 to the upside (17.1% down, 23.3% up), improving to about 1.6:1 once the 3.15% dividend is included on both sides. That is a reasonable but unexceptional shape. What makes it a Hold rather than a Buy is that the base sits below consensus, the knowledge base is empty and stale, the single most-hyped part of the business went backwards in the half, and roughly half of the reported growth is a currency move we have no view on.

7. Knowledge base — ninety raw hits, eighty-four of them a macroeconomic survey

Raw hits: 90. Entity matches: 4. Text-only matches: 86. DISCARDED for collision: 84 (93.3%). Used: 6. Name-level conviction claims: 3. Claims dated in 2026: ZERO.

This is the most heavily contaminated knowledge-base lane in the programme to date, and the contamination has a single cause.

The collision, quantified and named

"PMI" is also the standard abbreviation for the Purchasing Managers Index — a family of monthly business-survey series (ISM manufacturing, S&P Global composite, regional Federal Reserve surveys) that macro commentators discuss constantly. The Synthos knowledge base is heavily weighted toward macro voices, so a text search for "PMI" against 51,928 distilled claims returns an enormous false-positive set.

We classified all 90 hits programmatically and then read every one. Of the 86 text-only matches, 83 mention "PMI" or "PMIs" exclusively in the Purchasing Managers Index sense and contain no reference to Philip Morris, IQOS, ZYN, Marlboro or any tobacco entity. Representative examples, quoted verbatim to make the collision unmistakable:

> "Manufacturing PMI, after 20+ months below 50, is bottoming and will rise toward ~60 in the final four months on AI-infrastructure-driven capex and broadening revenues." — jordi_visser, 2025-08-31

> "May Eurozone composite PMI fell to 47.5, services crashed to 46.4 (63-month low), France composite 43.5 (five-year low)." — eurodollar_university, speaker Jeff, 2026-05-24

> "A roughly three-year PMI (growth acceleration) cycle is rolling over; declining-PMI environments reliably stress and break unstable parts of the system." — lyn_alden, 2022-05-29

And critically, one of the four ENTITY matches is the same collision — the extractor tagged "PMI" as an entity on a claim that is unambiguously about the survey:

> "Manufacturing recovery is the main story after a multi-year bear market in IP; PMIs going higher and I believe they'll get close to 60 next year, forcing people to adjust, all driven by AI."

> Test: "PMIs close to 60 next year" — jordi_visser, 2025-09-14, bullish, conviction 70, entities: ["manufacturing", "PMI"]

Discarded: 83 text-only + 1 entity match = 84 claims, 93.3% of the raw lane. The reason is stated once and applies to all of them: "PMI" here means Purchasing Managers Index, a macroeconomic data series, not Philip Morris International. Their date range is 2021-07-19 to 2026-05-24 and their channel distribution is jordi_visser (43), andreas_steno (12), lyn_alden (9), darius_dale (6), eurodollar_university (6), raoul_pal (3), forward_guidance (2), and one each from four others. Had these been counted, this dive would have reported a 90-claim lane with strongly positive net conviction on a company none of them mentions.

We also checked the bare ticker "PM" for the other obvious collisions — p.m. as a time of day, portfolio manager, project manager, prime minister. None appears in the entity lane, and none appears in the text lane either: the contamination is 100% Purchasing Managers Index. That is a useful negative finding, because it means the fix is a single disambiguation rule rather than a general-purpose one.

The six survivors, verbatim

Three are name-level claims about Philip Morris International. All three are entity matches.

> 2024-10-14 · bullish · conviction 60 · horizon: thesis · entities: Philip Morris, PM · channel: business_breakdowns · NO NAMED SPEAKER

> "Philip Morris at ~14x may even be cheap given its leading next-gen mix (~30-40%) with iCOS now its largest brand."

> 2024-10-14 · bullish · conviction 70 · horizon: thesis · entities: Zyn, Philip Morris, PM · channel: business_breakdowns · NO NAMED SPEAKER

> "Zyn's US stranglehold shows modern-oral brands, once established, are very hard to displace—strong loyalty plus PM's distribution fueled its growth."

> 2025-04-29 · neutral · conviction 55 · horizon: thesis · entities: Philip Morris · channel: invest_like_the_best · NO NAMED SPEAKER

> "Tobacco was historically a top-performing stock validating the brand-loyalty insight, but that insight is now largely priced in."

Three mention the company but are NOT about it, and are labelled as such rather than counted as conviction.

> 2024-10-14 · bullish · conviction 70 · entities: British American Tobacco, BTI · channel: business_breakdowns · no named speaker

> "As market realizes BAT isn't a melting ice cube (new products replacing combustibles over ~10yrs), it could re-rate toward Philip Morris's 14x and more than double."

> This is a claim about BAT. Philip Morris appears only as the valuation anchor it should converge toward.

> 2024-10-14 · bullish · conviction 65 · entities: British American Tobacco, Velo, BTI · channel: business_breakdowns · no named speaker

> "BAT's Velo dominates Scandinavia/Europe (+35% last year) and its V2.0 product, once US-approved, positions it to compete against Zyn."

> This is a competitive claim ABOUT a rival, and its content is mildly NEGATIVE for PM — the only bearish content in the entire surviving lane.

> 2021-10-28 · bullish · conviction 75 · entities: none · channel: macrovoices · SPEAKER: David Hay

> "Energy producers are very good investments for years — like Philip Morris, they'll raise prices, buy back stock and lift dividends; some run 15% free-cash-flow yields ... and universal ESG shunning makes it a great contrarian setup."

> This is a claim about energy producers. Philip Morris is the ANALOGY. It is also factually stale as applied to PM: the company has not bought back stock since 2022.

What this lane is, stated without inflation

Concentration is severe and it is single-document. Four of the six survivors — 67% — come from ONE Business Breakdowns episode dated 2024-10-14 (doc_id: business_breakdowns-JckerODmIu8). Two of the three name-level claims come from that same single episode. Under the per-speaker concentration policy this is the failure mode the policy exists to catch: a lane that looks like it has multiple observations when it has one conversation.

Sourcing is the weakest form the 4-lane policy admits. Five of the six carry speaker: null and are attributed only to a CHANNEL. Only the macrovoices claim has a named speaker (David Hay), and that is the energy analogy. Across the full 86-claim text lane, 63 have no named speaker. A channel-level attribution tells you where a claim was said, not who said it or what their record is. No claim in the lane carries a speaker_role of management, so the half-weight management discount does not apply anywhere — which also means we have no company voice to quote separately, and the LLY failure mode (a lane that was 68% the company's own CEO) is not present here.

The lane is stale, and the staleness is disqualifying rather than merely unhelpful. The most recent surviving claim is 2025-04-29 — fifteen months before this dive. There is NOT ONE claim about Philip Morris dated in 2026. In that window the company changed its entire segment structure, received a first-in-category modified-risk authorisation for ZYN, watched its U.S. business fall 16.5% organic, and rode a 4.5-point currency tailwind. None of that is reflected anywhere in the lane.

And the most bullish claim's premise has expired. The 2024-10-14 claim argues Philip Morris "at ~14x may even be cheap." Whatever basis that 14x was struck on, the stock today trades at 26.85x trailing GAAP earnings, 23.08x trailing adjusted earnings and 20.43x the FY2027 consensus. The claim's supporting detail — "leading next-gen mix (~30-40%)" — is also now out of date on the low side: smoke-free was 41.5% of net revenues in 2025. So the claim was directionally right about the business and its valuation premise has been consumed by the price. That is a claim we should grade as CORRECT and then decline to re-use. (A minor distillation artefact worth logging: the claim renders IQOS as "iCOS".)

Conclusion, stated as a finding rather than dressed up. The Synthos knowledge base has no current view on Philip Morris International. It has one fifteen-month-old bullish valuation claim whose premise no longer holds, one durable and still-correct structural claim about modern-oral brand loyalty and ZYN's distribution advantage, one neutral claim that the tobacco brand-loyalty insight is "now largely priced in" — which, notably, is the claim that has aged best — and three claims that are about something else. Conviction rating: Very Low. Breadth: 3 sources. Net conviction: positive-low and stale. The thinness and the age of this lane are direct inputs to the Hold verdict, and the 84 discarded collisions are logged as a data-quality defect against the extraction pipeline, not as a fact about this company.

8. Data integrity — what we rejected and why

Fourteen findings. Listed rather than silently corrected. On balance PM's vendor file is BETTER than most in this batch — the enterprise-value rebuild came out clean to $1 million, the share count reconciles to 0.0002%, the segment tables tie to the filings in the years that matter, and the estimate EBITDA rows pass the arithmetic test. The defects are concentrated in staleness, in one zeroed field, and in four ratios destroyed by a negative denominator.

1. bal_a[0] is TWO QUARTERS STALE — the annual balance sheet is 2025-12-31 while two 10-Qs sit in the archive. The most recent is the 10-Q filed 2026-07-24 covering 30 June 2026. However — and this is unusual — km_ttm and ratios_ttm ARE computed on the current balance sheet. debtToMarketCapTTM of 0.168588 implies total debt of $49.113B, and cashPerShareTTM of $3.8386 implies cash of $5.998B; the 10-Q states "Our total debt was $49.1 billion at June 30, 2026, and $48.8 billion at December 31, 2025" and "cash and cash equivalents of $6.0 billion and $4.9 billion." Both trailing figures match the filing; the annual block does not. We use the trailing block and the 10-Q for every balance-sheet conclusion, and we state clearly that the bal_a inventory, receivables, goodwill and equity lines in this dive are as of 31 December 2025.

2. The three 8-Ks were checked for a financing post-dating every statement — and the finding is the OPPOSITE of the expected one. The 8-K of 2026-06-29 records a PREPAYMENT of €1.0 billion (approximately $1.1 billion) of the Swedish Match term loan, executed one day before the 30 June balance-sheet date and therefore already inside the 10-Q's $49.1 billion. There WAS a first-half 2026 bond issuance — the 10-Q's "Debt Issuances" table is stripped by our extraction, but the Q2 10-Q's XBRL introduces two note series (USDollarNotesDueApril2029, USDollarNotesDueApril2036) absent from both the 10-K and the Q1 10-Q, and the footnote "Interest is payable semi-annually, commencing in October 2026" points to an April pricing. The amounts are not recoverable from the extracted text and we do NOT estimate them; the net effect is already captured in the $49.1 billion prose total we use. The 8-K of 2026-07-09 is the CFO transition, not a financing. The 8-K of 2026-07-22 is the Q2 release, present as the cover page ONLY — Exhibits 99.1 and 99.2 were not extracted and the document contains no financial figures.

3. netDebt and the enterprise-value rebuild — TESTED AND PASSED, reported because a pass is a finding. We rebuilt net debt and enterprise value from scratch including the short-term-investments check that has broken this calculation on AAPL, BRK-B and KLAC. PMI's balance sheet carries $0 of short-term investments in every year of the file, and the 10-Q's liquidity paragraph names only cash and cash equivalents. The $2.891B in longTermInvestments is equity-method and equity-security holdings (RBH $51M, Level 1 equity securities $983M), not liquidity. Our rebuilt enterprise value is $334.435B against the vendor's $334.434B — a $1 million difference, or 0.0003%. No correction required.

4. Book-equity-based metrics are economically MEANINGLESS and are REJECTED as lenses — this is the headline metric finding on this name. PMI carries a stockholders' DEFICIT of $9.994B at 31 December 2025 (−$11.750B in 2024, −$11.225B in 2023), confirmed by the 10-Q's own statement title, "condensed consolidated statements of stockholders' (deficit) equity." Consequently priceToBookRatioTTM (−34.03x), priceToFairValueTTM (−34.03x), returnOnEquityTTM (−112.2%) and debtToEquityRatioTTM (−5.72x) are all arithmetically valid and all economically useless, as are financialLeverageRatioTTM (−7.95x) and debtToCapitalRatioTTM (1.21x, above 1.0). We use returnOnInvestedCapitalTTM of 26.3% as the honest return measure, cross-checked by return on assets of 15.9% and return on capital employed of 38.5%. Note that the vendor's own composite rating scores PM 1 out of 5 on BOTH return-on-equity and price-to-book, which is what drags the overall grade to C+ / 2 — a negative-denominator artefact, not a judgement about a business earning 26.3% on invested capital.

5. Quarterly depreciation and amortisation is ZEROED for the June 2026 quarter, concealing roughly half a billion dollars — CORRECTED. inc_q[0].depreciationAndAmortization reads $0, and the same row's ebitda of $4.535B against operating income of $4.530B implies $5M of D&A for a global manufacturer. The three prior quarters read $510M, $521M and $505M. We rebuilt the trailing figure from two independent vendor ratios — capexToRevenueTTM of 3.624% on $42.545B of trailing revenue gives capex of $1.542B, and capexToDepreciationTTM of 0.75256 then gives D&A of $2.049B — implying $513M for the June quarter, consistent with the series. Every EBITDA figure in this dive uses $2.049B.

6. researchAndDevelopmentExpenses is $0 in FY2020's successor years and internally inconsistent across the series — NO R&D TREND CAN BE READ. The annual series reads $495M (FY2020), $617M (FY2021), $0 (FY2022), $0 (FY2023), $759M (FY2024), $0 (FY2025), and $0 in every quarter. PMI does not present research and development as a separate income-statement line; its face line is "Marketing, administration and research costs," which the vendor maps entirely to SG&A. The 10-K confirms R&D is real and material — "The research and development expense for our smoke-free portfolio accounted for approximately 100% of our total research and development expense for the years ended December 31, 2025 and 2024" — and refers the reader to a numeric note our extraction stripped. We state the gap rather than estimate around it: no R&D figure and no R&D trend appears anywhere in this dive.

7. seg_geo reports a SUPERSEDED segment taxonomy and its FY2024 column does not tie — the FY2024 column is DISCARDED. The vendor supplies PMI's four geographic segments and stops at fiscal 2025. Those segments were retired on 1 January 2026 and replaced by International Smoke-Free, International Combustibles and U.S., per the 10-K at line 3306. Separately, the FY2025 column sums to $40,648M, matching the 10-K's stated net revenues exactly, while the FY2024 column sums to $37,545M against reported $37,878M — short by $333M. Back-solving the 10-K's stated Europe growth of +9.1% gives a FY2024 Europe figure of $15.684B against the vendor's $15.357B, a $327M gap that accounts for the shortfall: the vendor carries the originally-reported 2024 Europe number rather than the restatement following the Wellness reallocation. We use FY2025, discard FY2024, and take all 2026 segment data from the 10-Q prose.

8. seg_prod — the gross-versus-net excise-tax test, PASSED for every year from 2018, FAILED for the single FY2017 legacy row. On a tobacco name this is the classic 2x-style tie failure and we tested it explicitly. For FY2018-FY2025 the two product lines sum exactly to reported NET revenues — FY2025's $23.794B + $16.854B = $40.648B, matching the 10-K's "Net revenues of $40.6 billion", and the smoke-free line matches the 10-K's "$16.9 billion in 2025 and $14.7 billion in 2024 related to smoke-free" verbatim. The FY2017 row is the exception: it reports Combustible Tobacco of $74.305B plus Reduced-risk Products of $3.793B plus a separate Excise Taxes line of −$49.350B, the pre-2018 GROSS presentation. Summing the first two lines without the third produces 2.7x the true figure. No number in this dive touches the FY2017 row.

9. est.ebitdaAvg and est.ebitAvg were tested against actuals and PASS — an uncommon result worth reporting. No forward year carries a negative EBITDA; ebitAvg is below ebitdaAvg in all eight years; and no year shows the ebitdaAvg == −revenueAvg signature. Against the FY2025 actual, ebitdaAvg of $17.023B is 0.6% above our computed $16.929B and ebitAvg of $15.383B is 3.0% above actual operating income of $14.933B. Small, directionally consistent, not corrupt. Per the data contract, forward valuation still runs on epsAvg and revenueAvg — but the rows are sound and we say so.

10. GAAP versus non-GAAP — the gap is 22% and the basis is established from the FILING, not assumed. inc_a/inc_q are GAAP; earn_cal and est are adjusted consensus. Q2 2026 GAAP diluted EPS was $1.80 and the reported adjusted actual was $2.20. We rebuilt the bridge from the 10-Q's own items-affecting-comparability disclosure and it closes to a penny: $1.80 + $0.13 amortisation + $0.33 RBH impairment − $0.06 equity-security gain − $0.01 Egypt benefit = $2.19. Currency is NOT an adjustment — it is disclosed separately as a +$0.03 driver — so consensus is a REPORTED-CURRENCY adjusted number. Trailing multiples in this dive are shown on both bases (26.85x GAAP, 23.08x adjusted); forward multiples are adjusted-only and labelled.

11. FY2025 earnings growth of +60.8% is an ACCOUNTING ARTEFACT of the prior year and is not used as a growth input. Net income attributable to PMI rose from $7.034B to $11.348B and diluted EPS from $4.52 to $7.26. The 10-K explains: "We recorded an impairment charge of $2,316 million related to our RBH equity investment ... for the year ended December 31, 2024," alongside a $199M Vectura disposal loss, $180M of restructuring and a $45M Egypt charge. Operating income grew 11.1% reported and 9.3% organic — that is the number we use. A screen ranking on EPS growth would place PM near the top of the consumer-staples universe on an artefact. This is defect class 3 running in reverse.

12. lastDividend: 5.88 is the ANNUALISED rate mislabelled as the last dividend — corrected from the filing. The 10-Q states: "During the third quarter of 2025, our Board of Directors approved an 8.9% increase in the quarterly dividend to $1.47 per common share. As a result, the present annualized dividend rate is $5.88 per common share." The last declared quarterly dividend is $1.47. In this instance the trailing and forward figures coincide at $5.88 because the increase landed at the start of the trailing window, so the 3.15% yield is correct on both bases — but a reader treating $5.88 as a quarterly payment would compute a 12.6% yield.

13. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — we use tech and say why. The quote block reports a 52-week high of $207.76 and low of $142.11; the computed technical block reports $200.17 and $144.33. The discrepancies are 3.8% and 1.5% — larger than KLAC's in this same batch. We use the tech figures throughout, because they are computed from the same six-year close series that produces the moving averages, drawdown and relative-return figures, and mixing sources within the technical section would make the percentages internally inconsistent.

14. The vendor profile description is materially STALE. It states PMI's "smoke-free innovations are currently available across 71 global markets." The 10-K says 106 markets at 31 December 2025 and the 10-Q says 109 markets at 30 June 2026. It also omits ZYN and Swedish Match entirely from a brand list that includes HEETS and TEREA. No figure in this dive comes from the profile description; the CEO name (Jacek Olczak), employee count (84,900), currency (USD), beta (0.405) and exchange fields were used and are correct.

Not defects, correctly reported and independently confirmed: the share count — the 10-Q cover states 1,558,613,439 shares outstanding at 2026-07-17, which multiplied by $186.91 gives $291,320,437,883 against the vendor's reported market cap of $291,319,795,100 — agreement to 0.0002%, and consistent with the FY2025 diluted weighted-average of 1,558,000,000 (defect class 5 tested and clean, with no splits, no share classes and no ADS ratio); reportedCurrency is USD in every income statement, balance sheet and cash-flow row, matching profile.currency USD and the NYSE quote currency (defect class 4 tested and clean — the currency issue on this name is economic, not an accounting-basis mixing error); total debt of $48.835B at 31 December 2025 matches the 10-Q's stated "$48.8 billion"; cash of $4.872B matches the stated "$4.9 billion"; net revenues of $40.648B match "Net revenues of $40.6 billion for the year ended December 31, 2025"; the smoke-free revenue split matches the 10-K to the rounded $0.1 billion in both 2025 and 2024; capitalLeaseObligations is $0 in all three balance-sheet years consistently, so there is no lease double-counting or year-on-year discontinuity (defect class 9 tested and clean); stockBasedCompensationToRevenueTTM of 0 is consistent with PMI reclassifying restricted and performance share-unit expense into "corporate expenses and other," as the 10-Q states; and interestCoverageRatioTTM of 10.4x is consistent with trailing operating income of $16.059B against the disclosed interest expense run-rate.

Non-equity tripwire — checked and passed. PM is common stock, NYSE-listed, and the 10-Q cover page states "shares outstanding of the registrant's common stock, no par value per share." The price of $186.91 is not par-like ($25, $50 or $1,000); the dividend is variable and has been raised repeatedly, most recently by 8.9% in Q3 2025, rather than fixed; the 52-week band of $144.33 to $200.17 is a 38.7% range, far wider than a fixed-income-like instrument; volume was 5,088,180 shares (roughly $951M of turnover); ISIN US7181721090, CUSIP 718172109; isAdr: false, isFund: false, isEtf: false. Beta of 0.405 is low, which is the only reading that could superficially suggest a bond-like instrument — but it is a genuine low-beta consumer defensive, not a preferred or a baby bond. This is common equity.

9. Technicals

• Price $186.91. −6.6% from the 52-week high of $200.17, and +29.5% above the 52-week low of $144.33. Position within the annual range: 76th percentile — the upper quarter.
• Moving averages, both supportive: +1.8% above a 50-day average of $183.55, and +9.8% above a 200-day average of $170.26. Both are rising. This is an intact uptrend, unlike most of this batch.
• RSI 56.2 — mid-range, neither overbought nor oversold. MACD +2.44 — positive.
• Maximum drawdown from peak over the trailing year: −6.6%. That is the shallowest drawdown of any name in this batch, and it is the clearest single expression of what a 0.405 beta buys you.
• Relative performance, and this is where the picture turns negative: 3-month +10.5% vs SPY +7.6% and QQQ +7.7% — outperforming; 6-month +3.9% vs SPY +11.1% and QQQ +15.7% — underperforming by 7 to 12 points; 12-month +14.7% vs SPY +24.3% and QQQ +30.8% — underperforming the S&P by 9.6 points and the Nasdaq by 16.1 points. PM has been a poor absolute performer against a strong market and a good performer against risk, which is precisely the trade a defensive is.
• Sentiment: 17 buy, 7 hold, 1 sell across 25 analysts; consensus mean $210.50 (+12.6%), median $215 (+15.0%), high $225 (+20.4%), low $182 — 2.6% BELOW spot. A well-dispersed target set with only a 23.6% spread between high and low, which is unusually tight and matches the unusually tight EPS dispersion noted in Section 6.

Today's move and what it does to the entry

PM closed 2026-08-04 at $186.91, DOWN 0.27% or $0.50 from a previous close of $187.41. It opened at $186.00, traded a $184.36 to $187.46 range, and closed near the upper half of that range on 5,088,180 shares — almost exactly its 5,070,000 average volume.

This was one of only TWO declines in a twelve-name batch on a strong day, alongside a broad semiconductor and technology melt-up (ARM +17.36%, MRVL +12.81%, INTC +10.92%, TXN +5.42%, PANW +5.53%, IBM +3.91%, ANET +3.06%). A 0.405-beta consumer defensive going nowhere while high-beta names gap 5-17% is exactly what the factor exposure predicts, and it is not company news. No company-specific event appears in this file for 2026-08-04; the last was the Q2 release on 2026-07-22, thirteen days earlier.

The honest read on the entry. Unlike most of this batch, buying PM today does not mean paying for a one-day gap — the day was flat and the stock is 6.6% below its own high. But it also does not mean buying a discount. At $186.91 the stock sits at the 76th percentile of its annual range, above both moving averages, at 20.43x FY2027 consensus, with our base fair value of $201 only 7.5% above and 6.5% below the street's $215 median. The absence of a gap is not the same as the presence of an opportunity.

What would change the setup, stated as prices rather than adjectives:

• A price near $168 — roughly 18.4x FY2027E, a 3.50% dividend yield, and near the 200-day average of $170.26 — would put our base $201 at a +20% premium and make the risk/reward genuinely asymmetric. That is 10.1% below today's close and well inside the range this stock has traversed in the last twelve months.
• Or the 2026-10-21 print showing U.S. segment net revenues back to growth with ZYN offtake accelerating, which would repair the one broken leg of 6a. At that point paying 20x is defensible, because the uncertainty being paid for would be resolved.

10. Insiders — a new chief financial officer, a director grant, and zero open-market activity

The eight transactions in this file split into exactly two events, and neither is a discretionary trade.

Event one: an identical 1,119-share award to seven directors on 2026-05-06 at $169.93. Seven officers, the same share count, the same day, the same price is the unmistakable signature of the annual non-employee director equity grant made at the annual meeting. It is coded A-Award — shares granted BY the company, not purchased in the market. It carries no information about anyone's view of the price.

Event two, and it is the one that matters: a Form 3 filed 2026-08-03 for Massimo Andolina as "officer: Group Chief Financial Officer," reporting 71,588 shares held directly, with zero shares transacted. A Form 3 is an initial statement of beneficial ownership filed when a person BECOMES an insider. It is a corporate event, not a trade. It confirms the CFO transition disclosed in the 8-K of 2026-07-09: Andolina, previously President of the Europe Region, replaced Emmanuel Babeau effective 1 August 2026.

What the file therefore contains is: zero open-market purchases and zero open-market sales. That is a NEUTRAL signal and it is reported as neutral. Note also what is NOT here: no F-InKind tax-withholding transactions at all, which distinguishes this file from most in the batch — PMI's equity awards vest on a different calendar and the withholding activity falls outside this eight-transaction window.

The one substantive reading is governance, not sentiment. The finance function of a company carrying $49.1 billion of debt, negative book equity and a currency-exposed earnings base changed hands three days before this dive. Babeau had been CFO since 2018. The 8-K discloses the terms in full — continued base salary of CHF 1,260,012 as Strategic Advisor through 31 March 2027, a lump-sum severance equal to that salary, CHF 393,754 in lieu of a pro-rated 2027 incentive, full vesting of outstanding equity, and a twenty-four month non-competition provision. The generosity and the length of the non-compete both read as an orderly, planned succession rather than a departure under stress, and the appointment was pre-announced in an 8-K of 20 May 2026. We flag it as a watch item on financial-policy continuity — leverage targets, dividend policy, hedging programme — and nothing more.

11. Verdict, kill-criteria and flip conditions

Hold.

This is the most misread balance sheet and the most misread growth rate in this batch, and correcting both leaves us close to the street.

What is genuinely excellent, and none of it is in dispute: a 67.5% trailing gross margin and 37.7% operating margin, both the best readings in a six-year file; 26.3% return on invested capital; a smoke-free business that has gone from 23.8% of net revenues in 2020 to 41.5% in 2025, grew 13.7% organic in the first half with gross profit up 16.9% organic, and holds the only FDA modified-risk authorisation in the U.S. nicotine-pouch category; Marlboro at a record 11.0% international category share funding the transition; $12.0 billion of guided 2026 free cash flow on $1.4–1.6 billion of capital expenditure; a 3.15% dividend covered 1.3x; beta of 0.405 and a 6.6% maximum drawdown over a year in which the S&P rose 24.3%; Moody's outlook raised to Positive; and a Canadian litigation tail that is now closed with $51 million of remaining book exposure against seven years of overhang.

What we are declining to pay up for: 20.43x FY2027 consensus adjusted EPS, 26.85x trailing GAAP, 18.47x rebuilt trailing EBITDA and 7.86x sales, against a base fair value of $201 that is 7.5% above spot and 6.5% BELOW the street's $215 median. Net debt of $43.1 billion at 2.38x EBITDA — 2.54x once the $2.8 billion of Russian cash is treated as unavailable — with NO equity cushion of any kind and NO buyback to suspend. A first half in which reported revenue grew 9.8% and organic revenue grew 5.3%, and reported EPS FELL 8.4%. A U.S. segment — the single most-discussed part of the story — down 16.5% organic with gross profit down 27.8%. And a knowledge-base lane in which 93.3% of the hits were a macroeconomic survey series and the freshest genuine claim is fifteen months old.

The distinction that matters. We are not saying PM is a bad business; it is a very good one, and the smoke-free transition is real, measurable and disclosed rather than promised. We are saying that our fair value sits below the street, that we have no independent information the market lacks, that roughly half of the reported growth is a currency move we have no view on, and that the growth leg everyone is watching went backwards in the half. In the Synthos frame, a name where our number lands below consensus, the conviction lane is empty and stale, and the dividend does most of the work is a Hold by construction. The difference between this and a Watch is the 3.15% yield and the 0.405 beta: this is a name a holder should keep, not a name a non-holder should chase.

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

• A second consecutive half of U.S. segment decline at the 2026-10-21 or 2027-02 prints, or ZYN offtake turning negative in a growing category. That converts an inventory story into a share-loss story and removes the entire growth premium in the multiple.
• A material dollar rally that pushes reported revenue growth below organic growth for two consecutive quarters. Consensus is struck on reported-currency adjusted EPS; this cuts the number being valued without any business deterioration.
• Net debt to EBITDA rising above 3.0x, or the dividend payout exceeding 90% of free cash flow. With no buyback to cut and negative equity, the dividend is the only lever, and cutting it on this shareholder base would be a repricing event of its own.
• The EU Tobacco Excise Directive achieving unanimous member-state approval on anything close to its contemplated 1 January 2028 implementation with heated tobacco and pouches in scope on unfavourable terms. That is a structural repricing of the smoke-free margin, not a one-year cost.
• A Russian escalation affecting the 6% of net revenues, 9% of volume, or the $2.8 billion of cash held there.
• An adverse ruling in the German heated-tobacco excise appeal, where an oral hearing is expected at the earliest in Q4 2026 and where PMI has already provisionally paid EUR 721 million (approximately $751 million).

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

• Price below approximately $168 (roughly 18.4x FY2027E, a 3.50% dividend yield, near the 200-day average of $170.26) with no deterioration in the fundamental picture. This is the single most likely path to a Buy and it is only 10.1% below today's close. At $168 the base case is +20% before the dividend.
• FDA authorisation of IQOS ILUMA, pending since October 2023 and accepted for review in March 2024. That opens the induction platform in a market the filing describes as having approximately 25 million adult smokers, and it is the largest single un-modelled option in the name.
• U.S. segment net revenues returning to growth with ZYN offtake accelerating on the ZYN ULTRA launch and the June 2026 modified-risk claim — the one broken leg of 6a repaired.
• Net debt to EBITDA falling below 2.0x on a reported basis, which would remove the principal reason we hold the multiple flat rather than expanding it.
• Genuine, current name-level knowledge-base conviction appearing. At present the lane's freshest claim is from April 2025 and its most bullish claim rests on a multiple the stock left behind. That alone is worth roughly one tier.

Where PM fits in the Synthos Framework Portfolio. The consumer-defensive / cash-return sleeve, at 0% today with a 1.5-2% target on a fill near $168 or on U.S. segment stabilisation confirmed at the October print. On the batch-overlap question this name is unusually clean: PM is the only consumer-defensive and the only sub-0.5-beta name in a twelve-name group otherwise dominated by semiconductors, software, aerospace and financials, and its 12-month underperformance against SPY is the price of that diversification. Held against the batch, PM is the only position that would have protected capital on a bad day and the only one whose expected return depends more on a dividend than on a multiple. That is a real portfolio function and it is why the verdict is Hold rather than Watch. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $186.91.

Single biggest risk: the currency tailwind reverses. First-half reported net revenue growth was 9.8% and organic growth was 5.3%; reported diluted EPS fell 8.4% and fell 14.4% excluding a $0.22 favourable currency contribution. Consensus — the $9.147 FY2027 number every multiple in this dive rests on — is struck on reported-currency adjusted EPS. A dollar rally does not slow this company down; it cuts the number the street is valuing, immediately and without warning, and it does so on a stock carrying 2.38x leverage and no equity cushion. The 10-Q names the exposures precisely: the Euro and the Russian ruble on the favourable side, the Japanese yen and the Swiss franc on the other, against a $52.2 billion gross notional derivative book that the filing itself says "could impact our debt levels and the pace of anticipated deleveraging."

Most fragile assumption in the price: that the U.S. segment stops shrinking. Every other assumption in 6a is a price or mix assumption and those are observable quarter by quarter with a long track record behind them. This one is a reversal assumption made against a first half in which U.S. net revenues fell 16.1% reported and 16.5% organic, gross profit fell 27.8%, ZYN shipments fell 11.2% to 5.2 billion pouches, and the company's own explanation for the Q2 stabilisation was that offtake was "flat to slightly growing versus the prior year in a growing category, largely as a result of the uneven competitive landscape." That last phrase is the company describing competitors selling unauthorised products which the FDA said on 8 May 2026 it does not intend to prioritise enforcing against. PMI's response is to spend more — the 10-Q says it intends "to accelerate U.S. investments in the second half" — which protects the volume and costs the margin. If the U.S. stays flat rather than recovering, total revenue growth lands nearer 4% than 6%, FY2028 EPS is nearer $9.30 than $10.02, and at 22x that is $205 — roughly today's price plus the dividend, and nothing else.

Provenance & disclosures

• Traceability: 90 raw knowledge-base hits touched PM (4 entity matches, 86 text-only). 84 were DISCARDED as collisions — 83 text-only plus 1 entity match — on the abbreviation "PMI" meaning Purchasing Managers Index, a macroeconomic business-survey series with no connection to Philip Morris International. That is 93.3% of the raw lane, spanning 2021-07-19 to 2026-05-24 across jordi_visser (43), andreas_steno (12), lyn_alden (9), darius_dale (6), eurodollar_university (6), raoul_pal (3), forward_guidance (2) and four others. Six claims survive (breadth 3 sources, net conviction positive-low and stale). Of the six, exactly THREE are name-level claims about Philip Morris: business_breakdowns 2024-10-14 (bullish, conviction 60, no named speaker — "Philip Morris at ~14x may even be cheap"), business_breakdowns 2024-10-14 (bullish, conviction 70, no named speaker — ZYN modern-oral brand loyalty), and invest_like_the_best 2025-04-29 (neutral, conviction 55, no named speaker — the brand-loyalty insight is "now largely priced in"). The other three mention PM only as a valuation anchor for British American Tobacco (two claims, business_breakdowns 2024-10-14) or as an analogy for energy producers (macrovoices 2021-10-28, speaker David Hay) and are labelled as such, NOT counted as name-level conviction. Four of the six survivors — 67% — come from a SINGLE Business Breakdowns document dated 2024-10-14, including two of the three name-level claims. Five of the six carry speaker: null and are attributed only to a channel — the weakest sourcing the 4-lane attribution policy admits; across the full 86-claim text lane, 63 carry no named speaker. No claim in the lane carries a management speaker_role, so the half-weight management discount does not apply anywhere. NOT ONE surviving claim is dated in 2026; the freshest is 2025-04-29, fifteen months before this dive. Searches covered the pre-extracted PM lane across 51,928 distilled claims, plus explicit collision checks on the bare ticker "PM" for the p.m.-time, portfolio-manager, project-manager and prime-minister senses — none of which appears; the contamination is 100% Purchasing Managers Index. All quotes are verbatim from the stored claim text; the rendering of IQOS as "iCOS" in one claim is a distillation artefact and is logged as such.
• Data as-of: fundamentals — income statement through the quarter ended 2026-06-30, filing-verified against the 10-Q filed 2026-07-24; balance sheet: the vendor's bal_a block stops at 2025-12-31 and is TWO QUARTERS STALE, while km_ttm/ratios_ttm are computed on the 2026-06-30 balance sheet and match the 10-Q's stated total debt of $49.1 billion and cash of $6.0 billion — we use the trailing block and the 10-Q; cash flow annual through 2025-12-31 with first-half 2026 taken from the 10-Q prose ($5.1 billion operating cash flow, $4.6 billion of dividends paid) · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873602 ≈ 2026-08-04T20:00:01Z ($186.91, −0.27%; 50-DMA $183.55; 200-DMA $170.26; RSI 56.2; MACD +2.44; beta 0.405) · knowledge-base claims 2026-08-04. Fiscal note: PMI's fiscal year is the calendar year. All figures come from the Synthos vendor data file for PM or from the SEC filings in the PM archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents and its gaps: 10-K filed 2026-02-06 (fiscal 2025); 10-Q filed 2026-04-24 (March 2026 quarter); 10-Q filed 2026-07-24 (June 2026 quarter — the primary source for this dive); 8-K filed 2026-06-29 (prepayment of €1.0 billion of the Swedish Match term loan); 8-K filed 2026-07-09 (CFO transition and Emmanuel Babeau separation agreement); 8-K filed 2026-07-22 (Q2 2026 earnings release). The 2026-07-22 8-K is present as the cover document ONLY — Exhibits 99.1 (the release) and 99.2 (the non-GAAP reconciliations and glossary) were not extracted and the file contains no financial figures. Known extraction limit, and it bites hard on this name: numeric tables are stripped from the current filing text extraction. The stripped tables include the condensed balance sheet, the long-term debt schedule, the "Debt Issuances in the first six months of 2026" table, the segment financial summary, the net-revenues-by-product-category table and the shipment-volume tables. Every volume, segment and balance-sheet figure in this dive therefore comes from filing PROSE, which PMI's MD&A supplies unusually generously, or from the vendor with the source stated.
• Where the filings contradicted or corrected the vendor (detailed in Section 8): the annual balance sheet is two quarters stale at 2025-12-31 while the 10-Q gives 2026-06-30 total debt of $49.1 billion and cash of $6.0 billion — we use the filing; four book-equity-based ratios REJECTED as economically meaningless (priceToBookRatioTTM −34.03x, priceToFairValueTTM −34.03x, returnOnEquityTTM −112.2%, debtToEquityRatioTTM −5.72x) against a filing-confirmed stockholders' DEFICIT of $9.994 billion, replaced by 26.3% return on invested capital; June-quarter depreciation and amortisation of $0 rebuilt to $513 million and trailing D&A to $2.049 billion, with trailing EBITDA recomputed to $18.108 billion against the vendor's implied $18.203 billion; seg_geo identified as a SUPERSEDED four-segment taxonomy retired 1 January 2026, with its FY2024 column discarded because it sums $333 million short of reported revenue and the 10-K's stated Europe growth of +9.1% implies a FY2024 Europe figure of $15.684 billion against the vendor's $15.357 billion; seg_prod's FY2017 row identified as the pre-2018 GROSS-of-excise presentation which would produce a 2.7x overstatement if summed naively; researchAndDevelopmentExpenses of $0 in FY2022, FY2023 and FY2025 identified as a zeroing artefact with no R&D figure used anywhere; FY2025's +60.8% net-income growth rejected as an artefact of the prior year's $2,316 million RBH impairment, replaced by operating income growth of +11.1% reported and +9.3% organic; lastDividend: 5.88 identified as the ANNUALISED rate, not the last declared dividend, which the 10-Q gives as $1.47 per quarter; quote.yearHigh/yearLow ($207.76/$142.11) rejected in favour of tech.hi52/lo52 ($200.17/$144.33); and the vendor profile's "71 global markets" corrected to the 10-Q's 109 markets at 30 June 2026. Where vendor and filing AGREED — worth recording, because on this name the agreements outnumber the corrections: the share count (10-Q cover 1,558,613,439 at 2026-07-17 × $186.91 = the vendor's market cap to 0.0002%); total debt of $48.835B at 31 December 2025 against the 10-Q's "$48.8 billion"; cash of $4.872B against "$4.9 billion"; net revenues of $40.648B against "Net revenues of $40.6 billion"; the smoke-free revenue split of $16.854B and $14.660B against the 10-K's "$16.9 billion in 2025 and $14.7 billion in 2024 related to smoke-free"; the FY2025 seg_geo column summing exactly to reported net revenues; reportedCurrency USD throughout, matching profile currency and quote currency; shortTermInvestments of $0, confirmed by the 10-Q's liquidity paragraph naming only cash and cash equivalents — so the vendor's enterprise value of $334.434B matched our independent rebuild of $334.435B to $1 million; and capitalLeaseObligations of $0 consistently across all three years, with no lease double-counting.
• Basis note: analyst estimates and the earnings-calendar actuals are adjusted, non-GAAP, and struck INCLUDING currency; reported diluted EPS is GAAP. The gap is 22.2% in the most recent quarter — $1.80 GAAP against $2.20 adjusted — and we rebuilt the bridge from the 10-Q's own items-affecting-comparability disclosure rather than assuming it: $1.80 + $0.13 amortisation of intangibles + $0.33 RBH impairment − $0.06 equity-security fair-value gain − $0.01 Egypt sales-tax benefit = $2.19, against the $2.20 reported. Currency is NOT an adjustment — the 10-Q discloses it separately as a +$0.03 driver of the quarter's EPS movement — which is why the currency exposure flows straight into the consensus number every forward multiple here rests on. Trailing multiples are shown on both bases (26.85x GAAP, 23.08x adjusted); forward multiples are adjusted-only and labelled at every use.
• Estimate coverage: 10 analysts on FY2026 revenue and 7 on FY2026 EPS; 11 and 9 on FY2027; 11 and 5 on FY2028. The FY2029 and FY2030 EPS rows rest on ONE analyst each and are excluded from every conclusion. The FY2027 EPS dispersion is exceptionally narrow — low $9.047, high $9.286, a 2.6% spread across nine analysts — so the fair-value range in this dive is built from the MULTIPLE rather than from the estimate distribution, and that construction is stated openly rather than disguised. est.ebitdaAvg and est.ebitAvg were tested against actuals and PASS (FY2025 ebitdaAvg 0.6% above our computed figure, ebitAvg 3.0% above actual operating income, no negatives, no inversions), but per the data contract all forward valuation still runs on epsAvg and revenueAvg.
• Peer note: the vendor peer set contains three genuine comparables — British American Tobacco ($127.5B), Altria ($113.7B) and Turning Point Brands ($1.6B), plus RLX ($2.4B) and Universal ($1.3B) — alongside consumer staples that are not competitors (Coca-Cola $372.4B, Procter & Gamble $351.9B, PepsiCo $190.0B, AB InBev $166.6B, Unilever $138.0B). No peer earnings or multiples are supplied in the file, so NO peer-multiple comparison is drawn anywhere in this dive. The 10-K does not publish a discrete competitor list; its competitive language concerns illicit and non-tax-paid volume, and, in the U.S. pouch category, competitors marketing products "that were not previously marketed in the United States and did not have a marketing granted order."
• Fair-value caveat: the $155 / $201 / $230 anchors are 17.0x and 22.0x the FY2027 consensus adjusted EPS mean of $9.14662, and 23.0x the FY2028 consensus adjusted EPS mean of $10.02097, respectively. Stated arithmetic, not a discounted cash flow. The base is deliberately sensitivity-disclosed: 21x gives $192, roughly 3% above spot; 23x gives $210, roughly the street's mean. The whole answer lives in a 21-23x band and no greater precision is claimed. The bear case requires NO earnings miss, only a de-rating; the bull case DOES require a re-rating and we name the two events that would justify one. The 3.15% dividend is additive to all three and is stated separately rather than folded in.
• Timing: second-quarter 2026 results were released 2026-07-22, thirteen days before this dive, and the 10-Q was filed 2026-07-24, so the entire first half is filing-verified. The next print is 2026-10-21, 78 days away — no near-term company-specific catalyst, which is an explicit input to the Hold verdict rather than an oversight. 2026-08-04 was a broad technology and semiconductor up-day (ARM +17.36%, MRVL +12.81%, INTC +10.92%, TXN +5.42%, PANW +5.53%, IBM +3.91%); PM's −0.27% made it one of only TWO decliners in the twelve-name batch, which is what a 0.405-beta consumer defensive does on a risk-on day and is treated as factor behaviour, not company news.
• 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.