Merck & Co. MRK
Healthcare · Drug Manufacturers - General · Synthos Deep Dive · 2026-08-04
The Overview
Merck sells prescription medicines. About half its sales come from a single cancer drug, Keytruda, which helps the immune system attack tumours. In the three months to June, Keytruda and its new injectable-under-the-skin version brought in $8.4 billion out of $16.6 billion of total sales.
This morning Merck reported that it lost money in the quarter — 54 cents a share on the standard accounting measure. The shares went up slightly anyway, and that was the right reaction, because the loss is not a loss in any economic sense.
Here is why. When Merck buys a small biotechnology company whose only real asset is one experimental drug, the accounting rules require it to treat the entire purchase price as an immediate expense rather than putting it on the balance sheet. Merck has been buying such companies aggressively. In January it bought Cidara for $9.2 billion, and the filing says this produced a charge of $9.0 billion — $3.62 a share. In May it bought Terns for $6.7 billion, producing a charge of $2.31 a share. Those two charges together are $5.93 a share, and they are the entire reason the company is reporting losses.
Strip them out and Merck expects to earn about $8.76 a share this year. At $128.04, that is a little under 15 times earnings — cheap by any normal standard for a large pharmaceutical company.
But it is cheap for a reason that is entirely public. Keytruda's American patent protection runs out around 2028. When that happens, competitors can make copies. Analysts already assume this: they expect Merck's sales to stop growing in 2028 and to be no higher in 2030 than in 2028, and they expect earnings per share to peak in 2028 and then fall.
Merck is spending very heavily to replace what it will lose — roughly $26 billion of acquisitions in nine months — and its borrowings have risen from $25 billion to $44 billion in fifteen months as a result. Some of the new products are working well: sales of a lung-artery drug called Winrevair rose 75%, a kidney-cancer drug called Welireg rose 67%, and this quarter American regulators approved the first once-a-day pill for lowering cholesterol in a class that previously required injections. The company has also started selling an under-the-skin version of Keytruda, which had $463 million of sales — this matters because that version has its own separate patent protection.
We think the shares are worth about $139 against $128.04 — a 9% gap. The average analyst says $140.18, which is the same. The dividend is 2.62%, and this is one of the least volatile shares in the market. So the answer is to hold it if you own it, collect the dividend, and watch what happens to the subcutaneous Keytruda number.
- Downside Risk 6/10. Very stable share, strong margins — and half the revenue goes off patent in about two years.
- Growth Quality 5/10. Sales rose 5%, but underlying profit per share was flat, because the drugs going away are as big as the ones arriving.
- Exponential Potential 5/10. A genuinely deep set of new drugs, mostly bought rather than invented.
Putting a number on it: our fair-value estimate is $139 against a current price of $148.26 — a premium price for a business we still like.
Our summary metrics
"Rated 6 — an extraordinarily low-volatility security with one very large, very dated problem. The supports: beta of 0.211, the lowest in this batch; interest coverage of 17.4x; a 75.9% gross margin; a 2.62% dividend; and a share that has fallen only 2.9% from its 52-week high while rising 65.0% from its low. Against that: KEYTRUDA and KEYTRUDA QLEX were $8.366 billion of the June quarter's $16.607 billion — 50.4% of total sales — with United States composition-of-matter protection lapsing around 2028, and consensus already models revenue flat at roughly $74 billion from FY2028 through FY2030 with earnings per share PEAKING at $10.590 in FY2028 and falling to $9.810 by FY2030. Net debt has risen from $25.028 billion at end-2024 to $43.790 billion at 2026-03-31 — a $18.8 billion increase in fifteen months — to fund roughly $26 billion of acquisitions, and the Terns transaction closed after that balance-sheet date. Tangible book value per share is NEGATIVE at −$0.56, with $21.579 billion of goodwill and $26.681 billion of intangibles against $52.606 billion of equity. And the geographic disclosure records China revenue falling from $5.494 billion (FY2024) to $1.939 billion (FY2025), a 64.7% decline in one year. A 6 rather than higher because the balance sheet is investment-grade and the cliff is four years pre-announced; rather than lower because half the company genuinely does go off patent."
"Rated 5 — real top-line growth, and underlying earnings that are not growing. Second-quarter sales of $16.607 billion grew 5% (4% excluding currency). The launch portfolio is genuinely working: WINREVAIR $588 million (+75%), WELIREG $271 million (+67%), CAPVAXIVE $184 million (+42%), PREVYMIS $295 million (+29%), OHTUVAYRE $204 million (new, from the October 2025 Verona acquisition), and KEYTRUDA QLEX — the subcutaneous formulation that is the entire conversion strategy — at $463 million. Animal Health grew 8% to $1.775 billion. But the earnings line does not follow. Adding back the disclosed $2.31 Terns charge gives clean second-quarter non-GAAP EPS of $2.18, against $2.13 reported a year earlier which itself carried a $0.07 charge — so on a like-for-like basis underlying earnings went from roughly $2.20 to $2.18, DOWN 1%, on 5% revenue growth. The declining lines explain it: JANUVIA/JANUMET −31%, VAXNEUVANCE −35%, LAGEVRIO −95%, PROQUAD/M-M-R II/VARIVAX −3%. A 5: the new products are growing quickly enough to offset the old ones and not yet quickly enough to grow the company."
"Rated 5 — a genuinely deep pipeline attached to a business consensus expects to shrink after 2028. The optionality disclosed in the quarter is substantial and specific: United States approval of LIPFENDRA (enlicitide), described in the release as 'the First and Only Once-Daily Oral PCSK9 Inhibitor To Reduce LDL-C', which addresses a market currently served only by injectables; positive TroFuse-005 data for sacituzumab tirumotecan in advanced endometrial cancer; positive Phase 3 results for a once-weekly oral HIV regimen developed with Gilead; MK-1406, the long-acting influenza-prevention antiviral acquired with Cidara, in Phase 3, which one knowledge-base claim puts at 60-76% single-shot protection in Phase 2 with peak sales guidance above $5 billion; and MK-4208, the oral allosteric BCR::ABL1 inhibitor acquired with Terns. The counterweight is the same knowledge base's most incisive observation, from a named speaker: 'Merck's Keytruda-replacement pipeline is ~70-80% acquired drugs.' Buying optionality is not the same as generating it, and roughly $26 billion has been spent in nine months. A 5: more genuine shots on goal than any name in this batch except Palantir, and a revenue base that a consensus of 18 analysts expects to stop growing in 2028."
What does “fair value” mean?
Fair value is Synthos’s estimate of what one share is worth today, based on our model of the company’s future cash generation and the risks to it. It is not a price target or a prediction of where the stock trades next quarter — it’s the price where we think risk and reward are balanced. Above it, you’re paying for outcomes better than our base case; below it, the market is offering a margin of safety.
The model’s inputs and weightings are proprietary — they’re the product. What we publish is the output, plus an independent, public-math cross-check further down the page (“Check our number”) so you can judge it for yourself.
What do the 5 tiers mean? (Core · Tactical · Watch · Hold · Avoid)
The Road Ahead
What we expect to matter in each window, and the evidence that would prove us wrong.
Short term 0-6 months
Neutral- Driver
- "A large recovery that has flattened at the top. Price $128.04 is 3.8% above a 50-day average of $123.40 and 14.3% above a 200-day average of $112.02, with RSI at 60.8 and MACD at +1.81 — constructive but not extended. The shares are 2.9% below the 52-week high of $131.82 and 65.0% above the low of $77.60, placing them at the 93rd percentile of their annual range — the highest position of any name in this batch. Twelve-month return is +61.48% against SPY's +24.26%, a 37-point outperformance on a 0.211-beta security. On the day of this dive Merck reported a GAAP loss per share of $0.54 and the shares rose 0.21% to $128.04 on 8.35M shares against a 10.08M average. That reaction is the correct one and it is also the point: the market has already worked out that the losses are acquisition charges, and the re-rating from 11x to 13.3x forward has largely happened."
- What we’re watching
- "Whether the raised full-year guidance holds. The release narrowed and RAISED expected 2026 sales to $66.3-67.3 billion and set non-GAAP EPS at $2.66-2.76 including $2.43 of Terns charges. Whether KEYTRUDA QLEX conversion accelerates from the $463 million recorded this quarter — it is the single most important operational number for the 2028 question and it is now disclosed separately. Whether the LIPFENDRA launch produces a visible revenue line; an oral PCSK9 inhibitor in a market served by injectables is the most commercially interesting approval in this batch. Whether China stabilises after falling from $5.494 billion to $1.939 billion of revenue between FY2024 and FY2025. And whether another acquisition is announced — the knowledge base has tracked a rumoured Revolution Medicine transaction at $28-32 billion, which would be four times the size of Terns."
- Confidence
- Medium
Medium term 6-24 months
Neutral- Driver
- "The medium term is a two-year window in which nothing much is supposed to go wrong and the market is already looking past it. Consensus has FY2027 revenue at $70.171 billion (+5.0%) and non-GAAP EPS at $9.611 (14 and 12 analysts), then FY2028 revenue at $74.305 billion and EPS at $10.590. Our derived clean FY2026 earnings power of approximately $8.76 — the raised guidance of $2.71 plus the $2.43 of disclosed Terns charges plus the $3.62 of disclosed Cidara charges — makes the FY2027 consensus a 9.7% increase, which is achievable on 5% revenue growth plus the absence of a $6 per share acquisition-charge year. The genuine question is what the acquisition programme costs from here. Merck spent roughly $26 billion in nine months (Verona ~$10B, Cidara $9.2B, Terns $6.7B) and the March 10-Q discloses the accounting: because the acquired assets are single clinical-stage candidates, the transactions are treated as ASSET acquisitions and the consideration goes straight to research and development expense rather than to the balance sheet. That is conservative accounting and it means every future deal of this shape will produce another loss quarter."
- What we’re watching
- "Whether the acquisition charges keep recurring at $3-4 per share a year. The mechanism is disclosed and mechanical — an asset acquisition of a single-candidate biotech expenses the whole purchase price — so the size of the charge is a direct function of how much Merck spends. A rumoured $28-32 billion transaction would produce a charge of roughly $11-13 per share. Whether net debt, at $43.790 billion at 2026-03-31 before the Terns closing, stabilises; it was $25.028 billion fifteen months earlier. Whether the FY2028 consensus revenue of $74.305 billion survives contact with biosimilar KEYTRUDA — 18 analysts currently carry that number and it is 11.5% above FY2026 guidance. Whether the Inflation Reduction Act's Medicare Part B price-setting, which the 10-Q notes begins in 2028, catches KEYTRUDA in the same year the patent lapses. And whether the dividend, currently absorbing 92% of a charge-depressed trailing GAAP EPS but only about 38% of clean earnings, is maintained through the transition."
- Confidence
- Medium
Long term 2+ years
No differentiated view- Driver
- "This is the only name in this batch where our long-horizon stance is a headwind, and the reason is arithmetic rather than opinion. KEYTRUDA and KEYTRUDA QLEX were 50.4% of second-quarter sales. Consensus — 18 analysts on FY2028 revenue, falling to 11 and 9 in the two years after — has revenue at $74.305 billion (FY2028), $74.324 billion (FY2029) and $74.045 billion (FY2030): flat to declining across three years. Earnings per share follows the same shape, peaking at $10.590 in FY2028 and falling to $10.165 and then $9.810. The market is not disputing that Merck can replace the revenue; it is saying that on current visibility the replacement roughly offsets the loss and no more. The knowledge base's best claim, from a named independent speaker, adds the nuance that matters: KEYTRUDA is a BIOLOGIC, so 'won't suffer rapid 80% erosion at its 2028 IP expiry' the way a small molecule would — but 'the FDA interchangeable-biosimilar pathway and political scrutiny are bear-case caveats.' Subcutaneous conversion via KEYTRUDA QLEX, at $463 million and rising, is the company's defence, and it is the number to track."
- What we’re watching
- "The KEYTRUDA QLEX conversion rate above all — a subcutaneous formulation with its own protection can retain a substantial share of the franchise if patients and payers move before biosimilar entry, and the company has now begun disclosing the figure separately. Whether MK-1406 (influenza prevention, Phase 3), sacituzumab tirumotecan (multiple solid tumours), enlicitide/LIPFENDRA (oral PCSK9) and the once-weekly oral HIV regimen with Gilead collectively reach the scale required — the knowledge base's own claim that the replacement pipeline is '70-80% acquired drugs' is the risk statement. Whether the acquisition programme is disciplined; the same lane records a rumoured $28-32 billion transaction described as 'a rich EV/sales multiple'. Whether Medicare Part B price-setting from 2028 compounds the patent problem. And leadership: Robert M. Davis is chairman and chief executive and the file carries no succession disclosure."
- Confidence
- Low
Exponential Potential
"Rated 5 — a genuinely deep pipeline attached to a business consensus expects to shrink after 2028. The optionality disclosed in the quarter is substantial and specific: United States approval of LIPFENDRA (enlicitide), described in the release as 'the First and Only Once-Daily Oral PCSK9 Inhibitor To Reduce LDL-C', which addresses a market currently served only by injectables; positive TroFuse-005 data for sacituzumab tirumotecan in advanced endometrial cancer; positive Phase 3 results for a once-weekly oral HIV regimen developed with Gilead; MK-1406, the long-acting influenza-prevention antiviral acquired with Cidara, in Phase 3, which one knowledge-base claim puts at 60-76% single-shot protection in Phase 2 with peak sales guidance above $5 billion; and MK-4208, the oral allosteric BCR::ABL1 inhibitor acquired with Terns. The counterweight is the same knowledge base's most incisive observation, from a named speaker: 'Merck's Keytruda-replacement pipeline is ~70-80% acquired drugs.' Buying optionality is not the same as generating it, and roughly $26 billion has been spent in nine months. A 5: more genuine shots on goal than any name in this batch except Palantir, and a revenue base that a consensus of 18 analysts expects to stop growing in 2028."
“Pharma AI drug-discovery names like Eli Lilly and Merck will still be growing revenues three years out at relatively low multiples as they transition to a new AI phase.”
“Merck is an 11 P/E stock (vs expensive Mag7) that may be at the point where expenses/earnings grow; a cheap multiple that can re-rate as pharma multiples expand.”
What could take this further than the base case — and where the market may be underpricing it. Full forward-growth and acceleration math in Deeper analysis, §4 below.
Clinical pipeline
Clinical-trial data for MRK hasn’t been pulled yet — check back soon.
Deeper analysis
Technicals, fundamentals, valuation, and the full expert-claim evidence panel — the detail behind the numbers above.
> ## ⚑ EARNINGS BANNER — REPORTED THIS MORNING
> Second-quarter 2026 results were released via 8-K/Exhibit 99.1 on 2026-08-04, the day of this dive. The headline was a GAAP loss per share of $0.54 and a non-GAAP loss per share of $0.13 — both including a $2.31 per share charge for the acquisition of Terns Pharmaceuticals. The stock closed UP 0.21% at $128.04. Full-year 2026 guidance was narrowed and RAISED to sales of $66.3-67.3 billion and non-GAAP EPS of $2.66-2.76. The vendor's quarterly income statement does NOT contain this quarter — it ends at 2026-03-31 — so every June-quarter figure in this dive comes from the earnings release itself.
Reference table
| Street consensus | $140.18 (+9.5%) · median $145 · high $155 · low $105 · 25 buy / 11 hold / 1 sell across 37 analysts |
| Valuation | 14.6x our derived clean FY2026 non-GAAP EPS of $8.76 · 13.3x FY2027E ($9.611) · 12.1x FY2028E ($10.590) · 47.2x reported FY2026 guidance of $2.71 — meaningless and stated so · 4.82x sales |
| Clean-earnings derivation | FY2026 guidance $2.71 + $2.43 disclosed Terns charges (8-K, 2026-08-04) + $3.62 disclosed Cidara charge (10-Q, 2026-05-04) = $8.76. Both add-backs are company-disclosed per-share figures, not estimates |
| Balance sheet | Net debt $43.790B at 2026-03-31 — cash $5.327B against total debt of $49.117B — up from $25.028B at end-2024. The vendor's enterprise value of $360.026B implies exactly this figure and is verified clean; it is bal_a (2025-12-31) that is stale |
| Conviction | Medium-low. 24 entity claims — but one is a genuine homograph collision (Merck KGaA), five arrive on a channel named after a competitor's chief executive, and four are the same claim repeated |
| Technicals | −2.9% from the 52-week high of $131.82, +65.0% above the low of $77.60 — the 93rd percentile of the annual range, highest in this batch; RSI 60.8; MACD +1.81; 12-month return +61.48% vs SPY +24.26% on a 0.211 beta |
What the experts actually said 10 traceable claims on MRK · showing the highest-conviction voices
“Merck at an 11 PE; expect it to head up to the 20s over the next couple years as an AI-benefiting margin story.”
“Merck's ~$10B Cidara buy backs CD388 flu-prevention drug with peak sales guidance over $5B; robust cost-effectiveness work supported conviction.”
“Industry revenue shifted from ~70% internally-sourced drugs (1995) to ~50-60% externally sourced; Merck's Keytruda-replacement pipeline is ~70-80% acquired drugs.”
Every claim reconciles to a real claim_id in the Synthos knowledge base — this is the evidence the verdict is built on, not vibes. Management (the company itself) is shown but half-weighted; one cautionary voice is included on purpose.
Price & moving averages 12 months · 50 & 200-day averages · 52-week range
Solid line = price · dashed line = 50-day average · dotted line = 200-day average · the two thin horizontal lines mark the 52-week high and low. Price above both averages is an uptrend.
Data summary: last close $148.35, 13% above the 50-day average ($131), 27% above the 200-day average ($117) — an uptrend. 5% below the 52-week high of $156, 91% above the 52-week low of $78.
Bollinger Bands 20-day average ± 2 standard deviations
The shaded band widens when the stock gets more volatile. Riding the upper edge = strong momentum (sometimes stretched); the lower edge = weak / potentially oversold.
Data summary: price $148.35 is currently inside the band (band $119–$160).
RSI (14) momentum gauge · 0–100
Above 70 (overbought zone, shaded) = overbought, below 30 (oversold zone, shaded) = oversold. Currently 63.
MACD 12 / 26 / 9 · trend & momentum
The MACD line crossing above the signal line (bars flip to the up color) = momentum turning up; crossing below (bars flip to the down color) = turning down. Bar height = the size of that gap.
Data summary: MACD is currently above its signal line by 0.90, positive momentum.
Relative performance vs S&P 500 & its sector (XLV (sector)), set to 100 a year ago
Solid = MRK · dashed = S&P 500 · dotted = XLV (sector). A rising line means it is beating that benchmark — the sector line shows whether it is a leader or laggard within its own group.
Forward revenue & earnings actual → estimate · "FY" = fiscal year, "E" = estimate
Darker bars = actual results, brighter = analyst estimates. Taller bars to the right = expected growth.
Key stats an RIA wants
1. The quarter, and the charge that made it a loss
From the 8-K/Exhibit 99.1 released 2026-08-04:
| $ millions except per share | Q2 2026 | Q2 2025 | Change |
|---|---|---|---|
| Sales | $16,607 | $15,806 | +5% (+4% ex-FX) |
| GAAP net (loss) income | $(1,335) | $4,427 | N/M |
| GAAP EPS | $(0.54) | $1.76 | N/M |
| Non-GAAP net (loss) income | $(330) | $5,366 | N/M |
| Non-GAAP EPS | $(0.13) | $2.13 | N/M |
| — of which Terns acquisition charge | $(2.31) | — | — |
| — Q2 2025 Hengrui licence charge | — | $(0.07) | — |
| CLEAN non-GAAP EPS (our derivation) | $2.18 | $2.20 | −1% |
Read the last line, because it is the whole quarter. Sales grew 5% and underlying earnings per share fell 1%. The reported loss is an accounting artefact of how Merck buys pipeline; the flat underlying earnings are not an artefact at all.
Top products, from the release:
| $ millions | Q2 2026 | Q2 2025 | Change | Share of sales |
|---|---|---|---|---|
| KEYTRUDA / KEYTRUDA QLEX | $8,366 | $7,956 | +5% | 50.4% |
| — of which KEYTRUDA QLEX | $463 | — | new | 2.8% |
| GARDASIL / GARDASIL 9 | $1,169 | $1,126 | +4% | 7.0% |
| ProQuad, M-M-R II, VARIVAX | $592 | $609 | −3% | 3.6% |
| WINREVAIR | $588 | $336 | +75% | 3.5% |
| BRIDION | $497 | $461 | +8% | 3.0% |
| JANUVIA / JANUMET | $429 | $623 | −31% | 2.6% |
| Lynparza (alliance) | $365 | $370 | −1% | 2.2% |
| PREVYMIS | $295 | $228 | +29% | 1.8% |
| Lenvima (alliance) | $283 | $265 | +7% | 1.7% |
| WELIREG | $271 | $162 | +67% | 1.6% |
| OHTUVAYRE | $204 | — | new (Verona) | 1.2% |
| CAPVAXIVE | $184 | $129 | +42% | 1.1% |
| VAXNEUVANCE | $148 | $229 | −35% | 0.9% |
| LAGEVRIO | $5 | $83 | −95% | 0.0% |
| Animal Health | $1,775 | $1,646 | +8% | 10.7% |
| — Livestock | $1,041 | $961 | +8% | — |
| — Companion Animal | $734 | $685 | +7% | — |
Three observations from that table.
First, KEYTRUDA QLEX is now disclosed separately and it is the single most important number in this dive. $463 million in a quarter, out of $8,366 million of total KEYTRUDA franchise sales — 5.5% converted. The subcutaneous formulation carries its own protection; every dollar moved into it before biosimilar entry is a dollar defended past 2028. The conversion rate is the metric that decides whether the FY2028-FY2030 consensus is too low or too high, and it is now trackable quarter by quarter.
Second, the growth and decline are roughly matched. The five fastest-growing lines (WINREVAIR, WELIREG, CAPVAXIVE, PREVYMIS, OHTUVAYRE) added approximately $536 million year on year. The four declining lines (JANUVIA/JANUMET, VAXNEUVANCE, LAGEVRIO, the paediatric vaccines) subtracted approximately $310 million. Net, the new portfolio is winning — but not by enough to move earnings per share, which is why the clean figure went from $2.20 to $2.18.
Third, GARDASIL grew 4% and the geographic disclosure explains why that is remarkable. The seg_geo block records China revenue falling from $5,494 million (FY2024) to $1,939 million (FY2025) — a 64.7% decline in one year — and the release attributes GARDASIL's growth to "higher demand in Asia Pacific and Europe... partially offset by lower demand in certain other international markets." A $3.6 billion revenue hole in one country has been absorbed and the company still grew.
Full-year 2026 outlook, narrowed and raised in the same release: sales of $66.3-67.3 billion; non-GAAP EPS of $2.66-2.76, with the release stating that the outlook "Includes Charges of $2.43 per Share for the Acquisition of Terns, Comprised of a One-Time Charge of $2.31 per Share as Well as Costs of Approximately $0.12 per Share To Finance the Acquisition and Advance MK-4208."
2. The clean-earnings derivation — the central calculation of this dive
Merck's reported earnings are currently unreadable, and the filings supply every number needed to fix them.
> Full-year 2026 non-GAAP EPS guidance (8-K, 2026-08-04): $2.66–$2.76, midpoint $2.71
> + Terns Pharmaceuticals charges: $2.43 per share — disclosed in the same release ($2.31 one-time plus ~$0.12 of financing and development)
> + Cidara Therapeutics charge: $3.62 per share — disclosed in the 10-Q filed 2026-05-04: "Merck recorded a charge of $9.0 billion to Research and development expenses, or $3.62 per share, (which primarily represented acquired in-process research and development with no alternative future use)"
> = clean FY2026 non-GAAP earnings power ≈ $8.76 per share
Both add-backs are company-disclosed per-share figures. Neither is our estimate. The only judgement we apply is that acquisition charges of this shape — whole purchase price expensed because the target's value is a single clinical candidate with no alternative future use — are capital expenditure in economic substance and should not be treated as operating earnings.
The 10-Q states the accounting mechanism explicitly for both deals. On Terns: "Merck anticipates the transaction will be accounted for as an asset acquisition since TERN-701 is expected to account for substantially all of the fair value of the gross assets to be acquired... Merck expects to record a charge of approximately $5.8 billion to Research and development expenses, or approximately $2.35 per share." On Cidara: "The transaction was accounted for as an asset acquisition since MK-1406 accounted for substantially all of the fair value of the gross assets acquired."
Three consequences, and the third is a forward warning.
First, the valuation. At $128.04, clean FY2026 earnings power of $8.76 is 14.6x. The FY2027 consensus of $9.611 — a year in which no such charge is currently modelled — is 13.3x. Against the reported guidance of $2.71 the multiple is 47.2x, which is the figure a screen will show and which is economically meaningless.
Second, the cross-check works. The FY2025 estimate row carries epsAvg of $8.935 — a year with no comparable charge — against our derived clean FY2026 of $8.76. Two independent routes to roughly the same underlying earnings power, one from actuals and one from guidance plus disclosed add-backs. It also means clean earnings are down about 2% year on year, consistent with the quarterly derivation in Section 1.
Third — and we pre-register this — every future acquisition of the same shape produces another loss quarter. The knowledge base has tracked a rumoured Revolution Medicine transaction at $28-32 billion. On the same accounting treatment that would produce a charge of roughly $11-13 per share. When it happens, the reported EPS will be deeply negative and it will not mean what it appears to mean. This dive says so in advance.
3. The 2028 problem, stated precisely
KEYTRUDA and KEYTRUDA QLEX were 50.4% of second-quarter sales. United States composition-of-matter protection lapses around 2028. Consensus is not in denial about this — it is the clearest thing in the estimate block:
| FY2026 | FY2027 | FY2028 | FY2029 | FY2030 | |
|---|---|---|---|---|---|
| Revenue (consensus) | $66.844B | $70.171B | $74.305B | $74.324B | $74.045B |
| Revenue growth | +2.9% | +5.0% | +5.9% | +0.0% | −0.4% |
| Analysts (revenue) | 16 | 14 | 18 | 11 | 9 |
| Non-GAAP EPS (consensus) | $2.784 (charge-laden) | $9.611 | $10.590 | $10.165 | $9.810 |
| EPS growth | — | — | +10.2% | −4.0% | −3.5% |
| Analysts (EPS) | 12 | 12 | 8 | 6 | 6 |
Revenue flat from FY2028 and earnings per share PEAKING in FY2028 and falling for two years thereafter. That is a consensus of 18 analysts on the FY2028 revenue line — the best-covered year in the whole block — saying that the replacement roughly offsets the loss and no more.
The knowledge base supplies the most useful qualification available, and it comes from a named independent speaker:
> 2025-02-05 · neutral · conviction 45 · channel business_breakdowns · speaker Ashwin Varma · speaker_role independent
> "Keytruda, a biologic, won't suffer rapid 80% erosion at its 2028 IP expiry, but the FDA interchangeable-biosimilar pathway and political scrutiny are bear-case caveats."
That distinction matters more than any other qualitative point in this dive. Small-molecule drugs lose 80-90% of revenue within a year of generic entry; biologics erode far more slowly because biosimilars require their own trials, manufacturing and physician acceptance. KEYTRUDA QLEX is the second defence — a subcutaneous formulation with independent protection, now running at $463 million a quarter and disclosed separately for the first time.
The same speaker supplies the risk statement:
> 2025-02-05 · neutral · conviction 55 · speaker Ashwin Varma
> "Industry revenue shifted from ~70% internally-sourced drugs (1995) to ~50-60% externally sourced; Merck's Keytruda-replacement pipeline is ~70-80% acquired drugs."
And the record bears it out. In nine months Merck has closed: Verona Pharma (~$10 billion, October 2025 — the source of OHTUVAYRE), Cidara Therapeutics ($9.2 billion, January 2026 — MK-1406, long-acting influenza prevention, Phase 3) and Terns Pharmaceuticals ($6.7 billion at $53 per share, closed May 2026 — MK-4208, an oral allosteric BCR::ABL1 inhibitor in Phase 1/2). That is approximately $26 billion, and net debt rose from $25.028 billion at end-2024 to $43.790 billion at 2026-03-31 to fund it.
What is being delivered against that spend, from the release:
- United States FDA approval of LIPFENDRA (enlicitide) — "the First and Only Once-Daily Oral PCSK9 Inhibitor To Reduce LDL-C in Adults With Hypercholesterolemia." An oral entrant into a category served exclusively by injectables is the most commercially significant approval in this batch.
- Positive TroFuse-005 data for sacituzumab tirumotecan in advanced or recurrent endometrial cancer.
- Positive Phase 3 results for a once-weekly investigational oral HIV regimen combining islatravir with lenacapavir, in collaboration with Gilead.
- MK-1406 in Phase 3 for influenza prevention; one knowledge-base claim records Phase 2 protection of "60-76%" from a single shot with "peak sales guidance over $5B."
None of these is yet a revenue line of consequence, and the four fastest-growing products in the quarter together are 7.4% of sales. The pipeline is real, deep, expensive and unproven.
4. Balance sheet, cash flow and capital returns
From the 10-Q filed 2026-05-04 (2026-03-31), against the vendor's 2025-12-31 balance sheet:
| Filing 2026-03-31 | Vendor 2025-12-31 | Vendor 2024-12-31 | |
|---|---|---|---|
| Cash and equivalents | $5,327M | $14,565M | $13,242M |
| Loans payable and current portion | $2,444M | $2,589M (shown as $2,883M short-term debt) | $2,649M |
| Long-term debt | $46,673M | $46,750M | $34,462M |
| Total debt | $49,117M | $50,534M | $38,270M |
| Net debt | $43,790M | $35,969M | $25,028M |
| Shares issued | 3,577,103,522 | 3,577,103,522 | — |
| Less treasury | 1,107,410,143 | 1,102,476,756 | — |
| Shares outstanding | 2,469,693,379 | 2,474,626,766 | — |
Two checks, and both pass — which is worth recording because they usually do not.
Check one: the vendor's enterprise value is CORRECT and current. enterpriseValueTTM of $360,025,752,800 less market capitalisation of $316,235,752,800 implies net debt of exactly $43,790,000,000 — which is precisely the 2026-03-31 figure ($49,117M of debt less $5,327M of cash). The enterprise value was struck on the March balance sheet even though bal_a was not. It is bal_a that is two quarters stale, not the EV. We use $43.790B.
Check two: the share count is right. 3,577,103,522 issued less 1,107,410,143 treasury equals 2,469,693,379; the vendor's implied count from market capitalisation divided by price is 2,469,819,000. Agreement to 0.005%. No share-class or partial-unit error.
The caveat both checks leave open: the Terns transaction closed in May 2026, AFTER the balance-sheet date, at $6.7 billion. No June-quarter balance sheet exists in this archive. Net debt at 2026-06-30 is therefore materially above $43.790 billion and this dive cannot quantify it. Stated as a gap.
Net debt has risen $18.762 billion — 75% — in fifteen months. Against clean EBITDA of roughly $27.8 billion (the vendor's implied trailing figure of $18.78 billion plus the $9.0 billion Cidara charge), that is approximately 1.6x — comfortable, and rising.
Cash flow:
| Fiscal year | Operating cash flow | Capex | Free cash flow | Acquisitions | Buyback |
|---|---|---|---|---|---|
| FY2022 | $19,095M | $4,388M | $14,707M | $121M | $0 |
| FY2023 | $13,006M | $3,863M | $9,143M | $12,032M | $1,346M |
| FY2024 | $21,468M | $3,372M | $18,096M | $4,093M | $1,306M |
| FY2025 | $16,472M | $4,112M | $12,360M | $10,042M | $5,084M |
Free cash flow of $12.360 billion in FY2025 is a 3.91% yield on the current market capitalisation — and it is depressed by the same mechanism as earnings, since asset-acquisition charges flow through operating cash flow. freeCashFlowYieldTTM of 4.46% is the trailing figure and is similarly contaminated.
Dividend and buyback. The trailing dividend is $3.36 per share, a 2.62% yield. The vendor's dividendPayoutRatioTTM of 92.1% is computed on charge-depressed trailing GAAP EPS of $3.61; on our clean earnings power of $8.76 the payout is approximately 38%, which is the honest figure. FY2025 repurchases of $5.084 billion are 1.61% of market capitalisation, up sharply from $1.306 billion in FY2024. Total shareholder yield is approximately 4.23% — the highest in this batch after Chevron.
5. Valuation — priced in or room?
At $128.04 (market cap $316.24B, 2,469,693,379 shares, enterprise value $360.03B at 2026-03-31):
| Clean FY2026 (derived) | FY2027E | FY2028E | FY2029E | |
|---|---|---|---|---|
| Revenue | $66.3-67.3B (guidance) | $70.171B (14 analysts) | $74.305B (18) | $74.324B (11) |
| Revenue growth | +2.9% | +5.0% | +5.9% | +0.0% |
| Non-GAAP EPS | $8.76 (our derivation) | $9.611 (12) | $10.590 (8 — thin) | $10.165 (6 — thin) |
| Reported non-GAAP EPS | $2.71 (guidance) | — | — | — |
| P/E on clean / consensus | 14.6x | 13.3x | 12.1x | 12.6x |
| P/E on reported guidance | 47.2x — rejected | — | — | — |
| EV/Sales | 5.39x | 5.13x | 4.85x | — |
| EV/EBITDA (clean, derived) | ~12.9x | — | — | — |
| Dividend yield | 2.62% | — | — | — |
Estimate coverage is good on FY2027 (14 analysts on revenue, 12 on EPS) and FY2028 revenue (18) but THIN on FY2028 EPS (8) and thinner still on FY2029 and FY2030 (6 each). No conclusion here rests on FY2029 or FY2030 beyond noting their SHAPE — flat revenue and declining EPS — which is itself the finding.
Per the data contract we tested est.ebitAvg and est.ebitdaAvg and found a basis inconsistency WITHIN the FY2026 row. That row carries ebitAvg of $26.919 billion and netIncomeAvg of $7.142 billion — a gap of $19.8 billion on a company whose interest expense is roughly $1.4 billion and whose tax rate is 23%. The EBIT line excludes the acquisition charges and the net income and EPS lines include them. The FY2027 row is internally consistent ($28.259 billion of EBIT against $23.782 billion of net income). We therefore use epsAvg and revenueAvg only, and we treat the FY2026 epsAvg of $2.784 as a charge-laden figure rather than an earnings estimate — which is exactly what the company's own guidance says it is.
Peer context. The vendor peer set is a sector-code grouping rather than a comparable set: Abbott, AstraZeneca, Cigna, CVS Health, Elevance, Humana, Novo Nordisk, Novartis, Thermo Fisher and UnitedHealth. Four of the ten are managed-care companies and two are tools and diagnostics businesses. The genuine pharmaceutical comparables are AstraZeneca ($241.21B), Novartis ($291.32B) and Novo Nordisk ($196.77B). No peer multiples are supplied in the file, so no peer-multiple comparison is drawn. Merck's $316.24B market capitalisation is the largest of that pharmaceutical group.
5a. What today's price assumes (the inversion)
At $128.04 — 13.3x FY2027 consensus and 14.6x our derived clean FY2026 — the price embeds:
- Non-GAAP EPS reaches $9.611 in FY2027 and $10.590 in FY2028. (Consensus-derived; 12 and 8 analysts.) Against our clean FY2026 of $8.76 that is 9.7% then 10.2% growth, on revenue growing 5.0% and 5.9%. Achievable, and it requires no acquisition charge in either year — which on the recent run-rate is itself an assumption.
- KEYTRUDA erodes SLOWLY after 2028 rather than collapsing. (Our reading of the consensus revenue path.) Revenue flat at roughly $74 billion in FY2029 and FY2030 with KEYTRUDA at 50.4% of the base means the market is modelling partial erosion offset by launches. This is the most fragile assumption in the price and it is a biologics-erosion assumption, not a pipeline assumption.
- The market keeps paying 12-14x forward earnings. (Our number.) At 10x FY2027E the stock is $96; at 18x it is $173. The multiple band is worth $77 — 60% of the current price — on unchanged estimates.
- The acquisition programme stops, or at least slows. (Our assumption.) FY2027 and FY2028 consensus contain no charge of the Cidara or Terns type. A single further transaction on the recent scale removes $2-4 per share from reported non-GAAP EPS in the year it closes, and a transaction on the rumoured $28-32 billion scale would remove $11-13.
- China does not deteriorate further. (Filing-confirmed decline; our assumption on the direction.) The geographic block records China revenue falling from $5.494 billion to $1.939 billion between FY2024 and FY2025 — a $3.555 billion hole already absorbed. A repeat would take another 5% off the top line.
5b. The return bridge (why the multiple moves)
Expected return over the next twelve months decomposes as: EPS growth (+10.2%, from FY2027E $9.611 to FY2028E $10.590) + multiple drift (COMPRESSION, from 13.3x to roughly 12.5x on the respective forward year, −6.0%) + shareholder yield (+4.23%) ≈ +8% to +9%.
Our base assumes mild multiple compression and it is not a de-rating thesis — it is the observation that a company whose consensus EPS peaks in FY2028 should see its forward multiple fall as that peak approaches. A 13.3x multiple two years before a peak is not the same as a 13.3x multiple in the peak year.
The counter-case is the one voice in the knowledge base made, four times, in November 2025: "Merck at 11 PE; expect it to head up to the 20s over the course of the next couple years." That call is partially resolved in his favour — the shares are up 61.5% over twelve months and the forward multiple has gone from roughly 11x to 13.3x. A further re-rating to the high teens is the bull case and it is what the $180 anchor represents. We do not underwrite it, because the arithmetic of a 2028 earnings peak argues the other way.
Sensitivity, stated openly. On the FY2027 consensus of $9.611: 10x gives $96, 12x gives $115, 14.5x gives $139, 16x gives $154, 18.7x gives $180. The street's $140.18 implies 14.6x. The entire disagreement between us and the street is one-tenth of a multiple point.
5c. Variant perception (where we differ, what would surprise)
- We do not differ from the street on value at all, and that is a finding. Our base of $139 against a consensus target of $140.18 is a 0.8% gap — the closest agreement of any name in this batch. Twenty-five of thirty-seven analysts are buyers. When our arithmetic lands on top of the crowd's on a name where the central issue is a publicly-known patent date, the honest conclusion is that we have no edge, and no-edge names default toward Hold.
- Where we do add something is the clean-earnings bridge, and we suspect most screens do not have it. A stock trading at 47.2x its own reported non-GAAP guidance and 14.6x its disclosed-charge-adjusted earnings will appear expensive in any automated system. Both add-backs — $2.43 for Terns and $3.62 for Cidara — are company-disclosed per-share figures sitting in two different documents, one filed today and one filed in May. We have combined them and the result is a materially different valuation.
- We think KEYTRUDA QLEX is the number the market is not yet tracking closely enough. $463 million in a quarter against an $8.366 billion franchise is 5.5% converted. If that reaches 30-40% before biosimilar entry, the FY2029-FY2030 consensus of flat revenue is too pessimistic; if it stalls below 15%, it is too optimistic. The company has begun disclosing it separately, which means it is now a quarterly falsifier. This is the single most watchable number in the dive.
- Positive surprise that would force a re-rate: KEYTRUDA QLEX conversion accelerating materially above $463 million a quarter; a LIPFENDRA launch trajectory resembling WINREVAIR's (+75%); Phase 3 success for MK-1406 in influenza prevention against the ">$5B peak sales" the knowledge base records; or China stabilising after a 64.7% revenue decline.
- Negative surprise that would break the thesis: an interchangeable biosimilar KEYTRUDA designation arriving earlier than expected; a further large acquisition at a price the knowledge base would again describe as "a rich EV/sales multiple"; the FY2028 consensus revenue of $74.305 billion being cut; or Medicare Part B price-setting — which the 10-Q notes begins in 2028 — catching KEYTRUDA in the same year protection lapses. That last item is a genuine double-hit that neither the consensus nor the knowledge base appears to be discussing.
Synthos fair values
All three anchors are multiples of the FY2027 consensus non-GAAP EPS of $9.611 (12 analysts — the first clean year), cross-checked against our derived clean FY2026 of $8.76 and against FY2028E.
- Bear ~$96 — 10.0x FY2027E. Cross-checks: 11.0x our clean FY2026; 9.1x FY2028E; 10.5x the FY2027 consensus LOW of $9.186; a 3.50% dividend yield; 14.3% below the 200-day average of $112.02. The scenario: biosimilar KEYTRUDA erodes faster than the biologics precedent suggests, QLEX conversion stalls below 15%, another large acquisition raises leverage, and the multiple contracts to where the market prices a shrinking pharmaceutical. −25.0%.
- Base ~$139 — 14.5x FY2027E. Cross-checks: 15.9x our clean FY2026 of $8.76; 13.1x FY2028E; enterprise value of $387.1B at 5.5x FY2027E revenue. Sensitivity: 12x gives $115, 16x gives $154, and the street's $140.18 implies 14.6x. The scenario: the launch portfolio keeps offsetting the declining portfolio, the acquisition programme moderates, QLEX converts at a useful rate, and the multiple compresses mildly as the FY2028 earnings peak approaches. +8.6%, plus 4.23% of shareholder yield.
- Bull ~$180 — 18.7x FY2027E. Cross-check: 17.0x FY2028E; 20.5x our clean FY2026. This bull case REQUIRES multiple expansion of roughly 40% and that is its weak leg — though it is exactly the call one knowledge-base voice made in November 2025 and has been partly right about. The scenario: KEYTRUDA QLEX conversion runs, LIPFENDRA and MK-1406 both land, the FY2029-FY2030 revenue plateau is revised upward, and pharmaceutical multiples re-rate toward the market. +40.6%.
Base is 8.6% above spot; asymmetry roughly 1.62:1 to the upside (25.0% down, 40.6% up). That ratio is decent. What makes this a Hold rather than a Buy is that our number is identical to the street's, the shares sit at the 93rd percentile of their annual range after a 61.5% twelve-month run, and the single fact that determines the outcome — how fast a biologic erodes after 2028 — is one on which nobody in this file, including us, has an edge.
6. Knowledge base — 24 claims, one genuine homograph, and a call that came good
Raw entity hits: 24. Text matches: 5. Discarded: 1 (a genuine homograph collision). Independent voice-days: approximately 8.
The search covered MRK, Merck and Merck & Co across the entity field of all 51,928 distilled claims. Stance: 18 bullish, 3 neutral, 3 bearish.
The homograph collision, named because it is the clearest example in this batch:
> Entity-tagged, 2025-08-27, bullish, conviction 75, channel biotech_hangout:
> "Merck-Springworks acquisition (~$5B) likely announced next week; Merck needs oncology/neuro assets and can take them global."
>
> Text-matched, 2025-08-21, neutral, conviction 50, same channel:
> "Merck KGaA to acquire Springworks in $3.5B deal — slight premium but well below February trading levels."
Merck KGaA is a German company entirely separate from Merck & Co., Inc. — they share a name for historical reasons and are different registrants, different listings and different businesses. The SpringWorks transaction was Merck KGaA's; the entity tagger attributed it to Merck & Co. The claim is discarded in full. This is a textbook homograph and it is worth flagging because it will recur.
The attribution structure of the remaining lane:
| Source | Claims | Note |
|---|---|---|
channel biotech_hangout, no named speaker | 9 | Acquisition-tracking lane; genuinely useful |
channel eli_lilly_ceo, no named speaker | 5 | A channel named after a DIRECT COMPETITOR's chief executive |
channel jordi_visser / speaker jordi_visser | 5 | Four are the same claim on 2025-11-15 and 2025-11-16 |
| speaker Ashwin Varma (channel business_breakdowns) | 2 | The two most analytically useful claims in the lane |
channel business_breakdowns | 1 | — |
channel huberman_lab | 1 | Clinical, not financial |
channel lex_fridman | 1 | Historical (Vioxx, 2004) |
Two attribution problems.
The eli_lilly_ceo channel carries five of the twenty-four claims — 21% of the lane — and it is named after the chief executive of a direct competitor. Two of the five are explicitly comparative: "On valuation Pfizer or Merck look preferable to Lilly — lower P/E (Lilly ~70x), higher dividend yield, and higher net margin/ROA" and "IRA is one-sided price-setting mislabeled as negotiation." A channel attributed to a rival company's chief executive is precisely the channel-versus-speaker failure the 4-lane policy exists to catch, and whether the underlying speaker is that executive or a guest, the claims cannot carry independent weight on the competitor's equity. We down-weight all five and use none of them for conviction.
Four claims dated 2025-11-15 and 2025-11-16 are the same Jordi Visser assertion, restated four times with convictions of 50, 65, 64 and 65: "Merck at 11 PE; expect it to head up to the 20s over the course of the next couple years." Counted once.
And it is worth grading, because it has substantially come good. At the time of the claim the shares were near their 52-week low of $77.60; they closed 2026-08-04 at $128.04, up 61.5% over twelve months, and the forward multiple has moved from roughly 11x to 13.3x. The direction was right and the magnitude of the re-rating is not yet there — the multiple has expanded 21%, not to "the 20s." A partial hit, recorded as such.
The two claims that carry a genuine independent named speaker are the two we actually use, and both are quoted in Section 3: that KEYTRUDA as a biologic will not suffer rapid 80% erosion but faces interchangeable-biosimilar and political caveats, and that Merck's replacement pipeline is 70-80% acquired. A single named independent analyst supplying both the bull qualification and the bear qualification on the central issue is the most valuable thing this lane contains.
The biotech_hangout lane, though channel-attributed, functions well as an acquisition tracker and its record is checkable. It flagged the Verona buyout as "priced-for-buyout... premium was modest not astounding" (2025-08-21); the Cidara transaction at "$9.2B... adds CD388, a long-acting antiviral showing 60-76% single-shot flu protection in phase 2" (2025-11-14) — the 10-Q confirms $9.2 billion and MK-1406 in Phase 3; and characterised the strategy accurately: "an assortment of deals (Cidara, Verona, Acceleron) plus oral PCSK9, not one blockbuster, to backfill and grow past the Keytruda cliff." The oral PCSK9 reference was made in November 2025 and LIPFENDRA was approved in the quarter reported today. That is a hit.
One competitive claim is worth recording as a risk:
> 2025-10-10 · bullish · conviction 70 · channel biotech_hangout
> "Arcus's casdatifan (HIF-2α) phase 1 in RCC shows 12.2mo PFS vs Merck belzutifan's 5.6mo and 31% response, looking better than Welireg in a $5B market."
WELIREG (belzutifan) grew 67% to $271 million this quarter and is one of the four fastest-growing products in the portfolio. A competitor with better Phase 1 data in the same $5 billion market is a specific, dated, checkable threat to one of the replacement assets.
Conclusion. The knowledge base has a positively-signed, medium-depth view of Merck built almost entirely on the acquisition programme, plus one named analyst supplying both sides of the KEYTRUDA question and one repeated valuation call that has partly come good. After discarding the homograph and down-weighting the competitor-channel claims, roughly eight independent voice-days remain. We use it in Section 3 and in the variant perception; it does not move the fair value.
7. Data integrity — what we rejected from the vendor file and why
Nine findings, and two verified-clean checks that are worth as much as the rejections.
1. inc_q does not contain the June 2026 quarter — the vendor income statement is one quarter stale. The nine quarterly rows end at 2026-03-31. The June quarter was reported on the morning of this dive and appears only in earn_cal (revenue $16,607M, non-GAAP EPS −$0.13). Every June-quarter figure in this dive comes from the 8-K/Exhibit 99.1 released 2026-08-04. Consequently netIncomePerShareTTM of $3.614, priceToEarningsRatioTTM of 35.97 and every other trailing metric in the payload are struck on a window ending 2026-03-31 and are not trailing twelve months to the most recent quarter.
2. The FY2026 estimate row mixes bases WITHIN the row — EBIT/EBITDA rejected. ebitAvg of $26.919B against netIncomeAvg of $7.142B is a $19.8 billion gap on a company with $1.4 billion of interest expense and a 23% tax rate. The EBIT and EBITDA lines exclude the acquisition charges; the net income and EPS lines include them. The FY2027 row is internally consistent. We use epsAvg and revenueAvg only, and we treat the FY2026 epsAvg of $2.784 as a charge-laden guidance figure rather than an earnings estimate.
3. seg_geo FY2023 contains SEGMENT names, not geography — REJECTED for that year. The FY2023 entry reads {"Pharmaceutical segment": 53,583,000,000, "Animal Health segment": 5,625,000,000, "Other Segments": 907,000,000} — an exact duplicate of the FY2023 seg_prod row. There is no FY2023 geographic split in the vendor file. FY2025 and FY2024 are correct and are used.
4. The seg_geo block records a 64.7% collapse in China revenue — NOT a defect, a finding. China revenue reads $5,494M (FY2024) and $1,939M (FY2025) — a $3,555 million decline in one year. The FY2025 lines (United States $36,510M, EMEA $14,580M, Latin America $3,410M, Asia Pacific $2,983M, International $2,878M, Japan $2,711M, China $1,939M) sum to $65,011M against reported revenue of $64,926M — an $85 million, 0.13% overshoot consistent with a minor allocation residual. We use the block and flag the China collapse as material. The same $85M residual appears in seg_prod.
5. dividendPayoutRatioTTM of 92.1% is computed on charge-depressed earnings — corrected. The trailing GAAP EPS denominator of $3.614 reflects the $3.62-per-share Cidara charge. On our derived clean earnings power of $8.76 the payout ratio is approximately 38%. The reported figure would suggest a dividend at risk; the corrected one does not.
6. researchAndDevelopementToRevenueTTM of 33.8% is an artefact of the same charge — flagged. Merck's ordinary research and development expense runs at roughly 19-20% of revenue (FY2025: $12.514B on $64.926B). The trailing figure is inflated by the $9.0 billion Cidara charge routed through research and development. The same contamination sits in the FY2023 annual row, where researchAndDevelopmentExpenses of $30.531 billion on $60.115 billion of revenue produced GAAP diluted EPS of $0.14. Merck's research-and-development line is not comparable across years and no trend should be read from it.
7. tangibleBookValuePerShareTTM is NEGATIVE at −$0.564 — correctly reported, and the associated rating sub-scores are harsh rather than wrong. Goodwill of $21.579B and intangibles of $26.681B exceed equity of $52.606B less debt. The vendor's composite rating of B / 3 carries debtToEquityScore of 1 and priceToBookScore of 1 against returnOnEquityScore of 5. The inputs here are all defined — debt-to-equity of 1.07x and price-to-book of 6.90x are real — so this is not the undefined-input artefact seen elsewhere; it is simply an aggressive scoring of an ordinary pharmaceutical capital structure. We report it and do not use it.
8. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — we use tech. The quote block reports $135.05 / $77.58; the computed block reports $131.82 / $77.60. The high discrepancy is 2.4% and it changes the drawdown read from −5.2% to −2.9%. We use tech. Note also that tech.max_dd_from_peak of −3.70% differs from pct_from_hi of −2.87% because the peak is a multi-year maximum; per the data contract this is NOT a defect.
9. No 10-Q exists for the June 2026 quarter and the Terns closing post-dates every balance sheet in the file. The most recent filing-verified balance sheet is 2026-03-31. The Terns acquisition closed in May 2026 at $6.7 billion. Net debt at 2026-06-30 is therefore materially above the $43.790 billion used here and this dive cannot quantify it. Stated as a gap rather than estimated around.
VERIFIED CLEAN — two checks that passed and are worth as much as the rejections.
Check A: enterpriseValueTTM is correct and current. The vendor's $360,025,752,800 less market capitalisation of $316,235,752,800 implies net debt of exactly $43,790,000,000, which reconciles precisely to the 10-Q's 2026-03-31 position (loans payable $2,444M plus long-term debt $46,673M less cash $5,327M). The enterprise value was struck on the March balance sheet even though bal_a was not. Against the stale bal_a figure of $35.969B the residual would be 17.9% and the data contract would withhold the multiple; against the filing it is zero and the multiple is usable. This is the reverse of the usual finding and it is recorded as such.
Check B: the share count is exact. The 10-Q reports 3,577,103,522 shares issued and 1,107,410,143 in treasury at 2026-03-31, giving 2,469,693,379 outstanding. The vendor's market capitalisation divided by price implies 2,469,819,000. Agreement to 0.005%. No share-class error, no partial-unit error.
Non-equity tripwire — checked and passed. MRK is common stock, $0.50 par value, NYSE-listed, one class. Beta of 0.211 is the lowest in this batch and low enough to demand the check — but the 52-week range of $77.60 to $131.82 is a 69.9% spread, the dividend is variable, the price is not par-like, and volume was 8.35M shares (roughly $1.07B of turnover). This is common equity, and the low beta reflects sector defensiveness rather than any fixed-income characteristic.
8. Technicals
- Price $128.04. −2.87% from the 52-week high of $131.82; +65.00% above the 52-week low of $77.60. Position within the annual range: 93rd percentile — the highest of any name in this batch.
- Moving averages, both supportive: +3.8% above a 50-day average of $123.40; +14.3% above a 200-day average of $112.02.
- RSI 60.8, MACD +1.81 — constructive, not extended.
- Maximum drawdown from the multi-year peak: −3.70%. Merck has fully recovered a much larger prior decline.
- Relative performance: 3-month +13.20% vs SPY +7.59% and QQQ +7.67%; 6-month +12.94% vs SPY +11.09% and QQQ +15.71%; 12-month +61.48% vs SPY +24.26% and QQQ +30.80%. A 37-point twelve-month outperformance on a 0.211-beta security is an extraordinary combination and it is why the shares are no longer cheap in the way they were.
- Sentiment: 25 buy, 11 hold, 1 sell across 37 analysts. Consensus target $140.18 (+9.5%), median $145 (+13.2%), high $155 (+21.1%), low $105 (−18.0%).
Today's move
MRK closed 2026-08-04 at $128.04, up 0.21% or $0.27 from a $127.77 close, on 8.35M shares against a 10.08M average. The second-quarter release was published the same morning with a headline GAAP loss per share of $0.54 and a raised full-year outlook. A twenty-basis-point move on a reported loss is the market saying it has already done the arithmetic in Section 2 — the loss is a Terns charge, the guidance went up, and nothing changed. We agree with that read, and it is the reason this dive spends its effort on the 2028 question rather than on the quarter.
9. Insiders
The vendor insider block for MRK contains no transactions of substance in the window covered by this file. We report that as a gap rather than constructing a narrative from it. For a company that has completed roughly $26 billion of acquisitions in nine months and is approaching a defining patent event, the absence of disclosed officer or director open-market activity is neither positive nor negative — it is simply an absence, and it carries no signal.
10. Verdict, kill-criteria and flip conditions
Hold.
What is genuinely good: sales of $16.607 billion (+5%, +4% ex-currency) in the June quarter with a launch portfolio that is visibly working — WINREVAIR +75%, WELIREG +67%, CAPVAXIVE +42%, PREVYMIS +29%, plus OHTUVAYRE arriving from Verona; United States approval of LIPFENDRA, the first once-daily oral PCSK9 inhibitor; positive Phase 3 data for a once-weekly oral HIV regimen and positive TroFuse-005 data in endometrial cancer; KEYTRUDA QLEX at $463 million and disclosed separately for the first time; a 75.9% gross margin, 17.4x interest coverage, 0.211 beta — the lowest in this batch — and a 4.23% total shareholder yield; and a valuation of 14.6x clean FY2026 earnings power and 13.3x FY2027 consensus once two company-disclosed acquisition charges totalling $6.05 per share are added back.
What prevents a Buy: KEYTRUDA at 50.4% of sales with United States protection lapsing around 2028; a consensus of 18 analysts that has revenue flat at roughly $74 billion from FY2028 through FY2030 and earnings per share peaking at $10.590 in FY2028 before declining to $9.810; underlying non-GAAP EPS that went from $2.20 to $2.18 — down 1% — on 5% revenue growth; net debt up $18.8 billion in fifteen months to $43.8 billion before the Terns closing, funding roughly $26 billion of acquisitions; China revenue down 64.7% in one year; a base fair value of $139 that is 0.8% BELOW the street's $140.18; and a share price at the 93rd percentile of its annual range after a 61.5% twelve-month run.
The distinction that matters. Merck is cheap and it is cheap for a reason that is public, dated, and correctly understood by the market. We add one thing the market may not have fully processed — the clean-earnings bridge that turns a 47x reported multiple into a 14.6x real one — and we take that finding to a fair value that lands within 1% of consensus. That is a no-edge outcome on the valuation and a genuine information gain on the accounting. A holder collects 4.23% a year and owns the lowest-beta security in this batch. A buyer is paying full value for a company whose consensus earnings peak in two years. Hold.
Pre-registered KILL criteria — what would take this to Avoid:
- KEYTRUDA QLEX conversion stalling below roughly $500 million a quarter through 2027, which would mean the subcutaneous defence is not working.
- An interchangeable biosimilar KEYTRUDA designation arriving before 2028, or a court decision shortening the protection.
- FY2028 consensus revenue being cut below $70 billion, which would mean the 18-analyst view that the replacement offsets the loss has broken.
- A further acquisition above roughly $20 billion, particularly at what the knowledge base would call "a rich EV/sales multiple" — this would both raise leverage and produce a $8-13 per share reported charge.
- Net debt above roughly $60 billion, or an interest-coverage ratio below 10x.
- Medicare Part B price-setting, which the 10-Q notes begins in 2028, being applied to KEYTRUDA in the same year protection lapses.
Pre-registered FLIP conditions — what would take this to Buy — Tactical:
- Price below approximately $110 — roughly 11.4x FY2027E and near the 200-day average of $112.02 — at which the base case is +26% and the dividend yield 3.05%. This is 14% below today's close.
- KEYTRUDA QLEX conversion above roughly 20% of franchise sales (currently 5.5%), which would materially raise the FY2029-FY2030 revenue floor.
- A LIPFENDRA launch trajectory resembling WINREVAIR's, i.e. above $500 million of quarterly sales within eighteen months. An oral entrant to an injectable-only category is the highest-optionality asset in the portfolio.
- Phase 3 success for MK-1406 against the ">$5 billion peak sales" the knowledge base records, which would validate the $9.2 billion Cidara price.
- FY2029 and FY2030 consensus revenue being revised UPWARD from the current flat $74 billion, which would be the market conceding that biologic erosion is slower than modelled.
Where MRK fits in the Synthos Framework Portfolio. The healthcare / defensive-value sleeve, held rather than added to, at a 2% position appropriate to a 0.211-beta asset with a 4.23% shareholder yield, and a 3.5% target on a fill near $110. On batch overlap: Merck and UnitedHealth are the two healthcare names here and they are genuinely uncorrelated exposures — Merck's risk is a patent date, UnitedHealth's is reimbursement policy and litigation — and both are near-uncorrelated with the eight technology and financial names. At a 0.211 beta Merck is the single lowest-volatility security in this batch and the best pure diversifier in it. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $128.04.
Single biggest risk: KEYTRUDA at 50.4% of sales, off patent around 2028. $8.366 billion of $16.607 billion in the June quarter, from one molecule, with United States composition-of-matter protection lapsing in roughly two years. The market's assumption — visible in a consensus of 18 analysts holding revenue flat at $74 billion from FY2028 to FY2030 — is that biologic erosion is slow and the launches offset it. The knowledge base's named independent speaker agrees KEYTRUDA "won't suffer rapid 80% erosion" while naming "the FDA interchangeable-biosimilar pathway and political scrutiny" as the caveats. If erosion runs at 25% a year rather than 10%, roughly $8 billion of high-margin revenue disappears over three years, and at 13x that is the $96 bear case. Compounding it, the 10-Q notes Medicare Part B price-setting begins in 2028 — the same year.
Most fragile assumption in the price: that the acquisition programme stops. FY2027 and FY2028 consensus contain no charge of the Cidara or Terns type. But the entire replacement strategy is acquisition-led — one named speaker puts the replacement pipeline at "70-80% acquired drugs" — and the accounting is unforgiving: because these targets are single-candidate biotechs, the whole purchase price is expensed on close. Roughly $26 billion has been spent in nine months. A rumoured transaction the knowledge base has tracked at $28-32 billion would, on the same treatment, produce a charge of roughly $11-13 per share, and the reported non-GAAP EPS for that year would be deeply negative. It would not mean the business had deteriorated — but it would mean another year in which no screen in the market shows Merck's real earnings, and it would mean another $30 billion of capital committed against a patent cliff that is now two years away.
Provenance & disclosures
- Traceability: 24 tagged knowledge-base claims name Merck by entity, plus 5 text matches; 1 claim DISCARDED as a genuine homograph collision. The collision is documented in Section 6: an entity-tagged claim asserting a "Merck-Springworks acquisition (~$5B)" conflates Merck & Co., Inc. with Merck KGaA, a separate German company that in fact agreed to acquire SpringWorks for $3.5 billion — a fact correctly recorded in a text-matched claim from the same channel six days earlier. Two further attribution problems are disclosed and applied: FIVE claims (21% of the lane) arrive on a channel labelled
eli_lilly_ceo— a channel named after the chief executive of a direct competitor, which is the channel-versus-speaker failure the 4-lane policy exists to catch — and all five are down-weighted and used for none of the conviction; and FOUR claims dated 2025-11-15 and 2025-11-16 are the same Jordi Visser assertion restated with convictions of 50, 65, 64 and 65, counted once. After correction roughly eight independent voice-days remain, of which only two carry a genuine independent named speaker (Ashwin Varma, both 2025-02-05) and those two supply both the bull qualification and the bear qualification on the central KEYTRUDA question. Stance 18 bullish / 3 neutral / 3 bearish. Breadth 8, net conviction positive-medium. Two claims are graded against outcomes: the repeated November 2025 call that "Merck at 11 PE... will head up to the 20s" is a PARTIAL HIT (shares +61.5% over twelve months, forward multiple from ~11x to 13.3x, not "the 20s"); and a November 2025 claim describing the strategy as "an assortment of deals... plus oral PCSK9" is a HIT, since LIPFENDRA (enlicitide), the first once-daily oral PCSK9 inhibitor, was approved in the quarter reported today. All quotes verbatim from stored claim text. - Data as-of: second-quarter 2026 income data from the 8-K/Exhibit 99.1 released 2026-08-04, the day of this dive · balance sheet, acquisition accounting and per-share charge disclosures from the Form 10-Q filed 2026-05-04 (2026-03-31) · fiscal 2025 from the 10-K filed 2026-02-24 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873599 = 2026-08-04T19:59:59Z ($128.04, +0.21%; 50-DMA $123.40; 200-DMA $112.02; RSI 60.8; MACD +1.81) · knowledge-base claims 2026-08-04. Merck's fiscal year is the calendar year. All figures come from the Synthos vendor data file for MRK or from the SEC filings in the MRK archive; no figure comes from memory, recall or external retrieval.
- Filing archive contents and its gaps: 10-K filed 2026-02-24 (fiscal 2025, tables preserved); 10-Q filed 2025-11-05 (September 2025 quarter); 10-Q filed 2026-05-04 (March 2026 quarter); 8-Ks filed 2026-05-22 and 2026-05-28; 8-K filed 2026-08-04 carrying the full Exhibit 99.1 second-quarter release with the financial summary, top-product table, pipeline highlights and full-year outlook. There is NO 10-Q for the June 2026 quarter, so the most recent filing-verified balance sheet is 2026-03-31 and the Terns acquisition, which closed in May 2026 at $6.7 billion, post-dates every balance sheet in the file. Tables were available on all substantive documents.
- Where the filings corrected or extended the vendor (detailed in Section 7): the vendor's
inc_qidentified as one quarter stale — it ends 2026-03-31 and does not contain the June quarter reported the morning of this dive, so every trailing metric in the payload is struck on the wrong window; the clean-earnings bridge constructed entirely from company-disclosed per-share figures — FY2026 non-GAAP guidance of $2.71 plus $2.43 of Terns charges (8-K, 2026-08-04) plus $3.62 of Cidara charges (10-Q, 2026-05-04) = $8.76, against a reported-guidance multiple of 47.2x and a clean multiple of 14.6x; the FY2026 estimate row found to mix bases within the row ($26.919B of EBIT against $7.142B of net income), rejecting the EBIT and EBITDA lines;seg_geoFY2023 rejected as a duplicate of the segment table containing no geography;dividendPayoutRatioTTMof 92.1% corrected to approximately 38% on clean earnings;researchAndDevelopementToRevenueTTMof 33.8% flagged as contaminated by the $9.0B Cidara charge routed through research and development, making the R&D line non-comparable across years; andquote.yearHigh/yearLowrejected in favour oftech.hi52/lo52. TWO VERIFIED-CLEAN CHECKS, recorded because they are as informative as the rejections: (A)enterpriseValueTTMof $360.026B implies net debt of EXACTLY $43.790B, reconciling precisely to the 10-Q's 2026-03-31 position — the enterprise value was struck on the March balance sheet even thoughbal_awas not, so what would have looked like a 17.9% residual against the stale figure is in fact zero against the filing; and (B) the share count reconciles exactly — 3,577,103,522 issued less 1,107,410,143 treasury equals 2,469,693,379, against a vendor-implied 2,469,819,000, agreement to 0.005%. - Basis note: Merck reports GAAP results and a non-GAAP measure that excludes acquisition- and divestiture-related costs, restructuring costs and equity-investment gains and losses — but INCLUDES charges for acquired in-process research and development, which is why both the GAAP and non-GAAP second-quarter figures are losses. Analyst estimates are struck on the same non-GAAP basis, confirmed by the FY2025
epsAvgof $8.935 against reported GAAP diluted EPS of $7.28. All forward multiples in this dive use consensus non-GAAP EPS. The "clean" figures used for valuation add back only the two company-disclosed acquisition charges and are labelled as our derivation at every occurrence. - Estimate coverage: 16 analysts on FY2026 revenue and 12 on FY2026 EPS; 14 and 12 on FY2027; 18 and 8 on FY2028; 11 and 6 on FY2029; 9 and 6 on FY2030. No conclusion rests on the FY2029 or FY2030 EPS rows beyond noting their SHAPE — flat revenue and declining earnings — which is itself the central finding of Section 3.
est.ebitAvgandest.ebitdaAvgare rejected for the FY2026 within-row basis inconsistency. - Peer note: the vendor peer set is a sector-code grouping rather than a comparable set — four of the ten entries (Cigna, CVS Health, Elevance, Humana) are managed-care companies and two (Thermo Fisher, Abbott) are tools and diagnostics businesses. The genuine pharmaceutical comparables are AstraZeneca ($241.21B), Novartis ($291.32B) and Novo Nordisk ($196.77B). No peer multiples are supplied in the file, so no peer-multiple comparison is drawn.
- Fair-value caveat: the $96 / $139 / $180 anchors are multiples of the FY2027 consensus non-GAAP EPS of $9.611 — 10.0x, 14.5x and 18.7x — cross-checked against our derived clean FY2026 of $8.76 and against FY2028E. Stated arithmetic, not a discounted cash flow. Sensitivity: 12x gives $115, 16x gives $154; the street's $140.18 implies 14.6x, so the entire disagreement between us and the street is one-tenth of a multiple point. The bear case requires no acquisition failure, only faster biologic erosion and a de-rating; the bull case REQUIRES roughly 40% multiple expansion and that is named as its weak leg. No case incorporates a further acquisition; a transaction on the rumoured $28-32 billion scale would, on the disclosed asset-acquisition accounting, produce a reported charge of roughly $11-13 per share in the year it closed, and this dive pre-registers that so it is not later mistaken for a deterioration in the business.
- Timing: second-quarter 2026 results were released on the morning of 2026-08-04, the day of this dive, with full-year guidance narrowed and raised to sales of $66.3-67.3B and non-GAAP EPS of $2.66-2.76. The shares closed +0.21%. The next print is 2026-10-29, 86 days away, with consensus of $2.27 non-GAAP EPS on $17.298B of revenue.
- 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.