Thermo Fisher Scientific TMO

Reference table

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

The 10-K opens with a claim that is close to true: Thermo Fisher is "the world leader in serving science." The company was established in 1956, is headquartered at 168 Third Avenue, Waltham, Massachusetts, listed on the NYSE since 17 March 1980, and is run by chief executive Marc N. Casper. It employed approximately 125,000 colleagues at 31 December 2025 — 59,000 in the Americas, 44,000 in Europe/Middle East/Africa and 22,000 in Asia-Pacific — including approximately 14,000 sales personnel. Its brands are Thermo Scientific, Applied Biosystems, Invitrogen, Fisher Scientific, Unity Lab Services, Patheon and PPD.

The four reportable segments (10-K Item 1 and Note 11)

• Life Sciences Solutions — reagents, instruments and consumables for biological and medical research and for the production of biologics. Three businesses: biosciences, genetic sciences, bioproduction. Bioproduction is the fastest-growing line the filings name.
• Analytical Instruments — three businesses: chromatography and mass spectrometry, chemical analysis, electron microscopy. Electron microscopy explicitly serves the semiconductor market as well as life and materials sciences.
• Specialty Diagnostics — five businesses: clinical diagnostics, immunodiagnostics, microbiology, transplant diagnostics and the healthcare market channel. The microbiology business is being sold (see below).
• Laboratory Products and Biopharma Services — four businesses: laboratory products, the research and safety market channel, pharma services and clinical research. This is the segment that houses Patheon (contract drug manufacturing), PPD (clinical research) and now Clario.

The 10-Q confirms at Note 11 that the company reports in four segments and that "During 2026, there have been no changes to the company's basis of segmentation."

seg_prod is NOT the segment table — and this is the biggest presentation defect in the file

The vendor's seg_prod block supplies three lines: Consumables, Instruments and Service. Those are not segments. They are the revenue-by-type disaggregation that appears at 10-K Note 6 and 10-Q Note 6 under the heading "Revenues by type are as follows", and the filings are explicit that the two are orthogonal: "Each reportable segment earns revenues from consumables, instruments and services in North America, Europe, Asia-Pacific and other regions."

Read naively, seg_prod would tell a reader that Thermo Fisher has three segments. It has four, and none of them is called Consumables. The four-segment names do appear in the vendor block — but only in the FY2017 and FY2015 rows and earlier, and under a superseded taxonomy: the 2017 row reads "Laboratory Products and Services", which is the name the segment carried before it was renamed Laboratory Products and Biopharma Services. The 2013 row uses "Analytical Technologies", a segment that no longer exists. The 2012, 2011 and 2010 rows contain nothing but "Intersegment Eliminations" and "Excluded from Reportable Segment Operating Income" — line items, not segments.

So the vendor cannot answer the segment question at all, and neither can the filings as extracted, because the Note 11 segment tables are numeric and the extraction strips numeric tables. The four-segment revenue and margin split is not available from any source in this file, and no such figure appears anywhere in this dive.

What seg_prod CAN do, and does cleanly, is the revenue-by-type reconciliation. We summed every line.

Three of the five years reconcile EXACTLY and the other two are within $1 million of rounding. There is no double-counting and no vanished line. We use this table, correctly labelled as the revenue-by-type disaggregation rather than as a segment table.

And this table is where the whole story of the last four years lives — see Section 2.

Geography (seg_geo), and a label change that breaks the comparison

The sums reconcile to reported revenue to within $1 million in all three years. But the FY2023-and-earlier rows are labelled "North America" and "Other Regions" while FY2024 and FY2025 are labelled "United States" and "All Other Countries". North America includes Canada and Mexico; the United States does not. The apparent FY2023-to-FY2024 decline in the top line from $22.764B to $22.504B is therefore not a like-for-like comparison and we do not draw a conclusion from it. The FY2024-to-FY2025 comparison is clean, and it shows the US up 2.4% and Europe up 8.9% — the largest regional move in the file, and one that carries currency translation inside it, because the 10-K states revenues are translated at average exchange rates for the period.

China is not broken out. The 10-K discusses China qualitatively (revenue "declined in China" in 2025; a named risk factor covers "Chinese regulations requiring the use of local suppliers") and the 10-Q says China grew in the first half of 2026, but no China revenue figure is available in this file.

Customer concentration: absent, and that is a positive

We searched the 10-K for customer-concentration language and found none. There is no "customers accounted for 10% of revenue" disclosure. For a $44.6 billion company selling to pharmaceutical firms, hospitals, universities and government agencies across roughly 70 countries, this is a genuine structural difference from the concentrated capital-equipment names elsewhere in this batch, and it is the main reason the risk score is a 5 rather than a 6.

2. The single most important thing to understand about these numbers: four years of nothing, bridged by acquisitions

Start with the top line, because everything else follows from it.

Read the 2022 and 2025 rows next to each other. Revenue $44.915B → $44.556B, a decline of 0.8% over three years, or minus 0.27% compounded. Diluted EPS $17.64 → $17.74, a gain of 0.6% over three years — and the share count fell 4.3% over the same period, meaning net income actually fell 3.1% and the buyback did all the work.

Peak operating margin was 2021 at 26.3%. It has been 17.4-19.0% every year since.

Now the composition, which is the actual explanation

Two large forces moved in opposite directions and roughly cancelled.

Force one — the COVID runoff, visible in Consumables. Consumables revenue: $22.608B (2021) → $20.624B (2022) → $17.597B (2023) → $17.587B (2024) → $18.664B (2025). That is a $5.0 billion decline from the 2021 peak to the 2024 trough, a 22.2% fall, followed by a partial recovery. The 10-K names the cause directly in the segment commentary for Laboratory Products and Biopharma Services: organic growth in 2025 was "partially offset by moderation in COVID-19 vaccines and therapies-related activity", and for the pharma and biotech market as a whole, growth was "partially offset by reduced demand for COVID-19 vaccine and therapy related products and services."

Force two — the capital-equipment recession, visible in Instruments. Instruments revenue: $7.924B (2022) → $7.646B (2023) → $7.446B (2024) → $7.301B (2025). Three consecutive annual declines, minus 7.9% cumulative. The 10-K's stated causes are pharma and biotech customer caution, an academic and government market where "Revenues... declined, driven by customer hesitancy in a more uncertain environment in the U.S. and macro conditions in China", and China itself where sales "declined" in 2025. The risk factors add the specific 2025 event: "In October 2025, the federal government entered a shutdown due to a lapse in appropriations... A similar impasse in federal government budget decisions could lead to substantial delays or reductions in federal spending."

Force three — the offsetting purchase, visible in Service. Service revenue: $8.850B (2021) → $16.367B (2022) → $17.614B (2023) → $17.845B (2024) → $18.592B (2025). The 2021-to-2022 jump is +84.9% in a single year. That is the first full year of PPD (clinical research) and the pharma-services businesses. Since that step, Service has grown 4.5%, 1.3% and 4.2% — mid-single-digit at best.

Put the three together and the four flat years are fully explained: a $5.0 billion COVID hole plus a $0.6 billion instrument decline, filled by a $9.7 billion acquired services business whose own organic growth is mid-single-digit.

And this is what it cost

Nineteen point seven billion dollars of acquisitions, five point three billion dollars of added goodwill, six point eight billion dollars of added debt — and revenue is smaller than it was.

The named acquisitions, from the filings: The Binding Site Group (Specialty Diagnostics, closed 3 January 2023); Olink Holding AB (Life Sciences Solutions, closed 10 July 2024, next-generation proteomics); Solventum's filtration and separation business (Life Sciences Solutions, closed 1 September 2025); a sterile fill-finish and packaging facility (Laboratory Products and Biopharma Services, 2025); Clario Holdings, Inc. (Laboratory Products and Biopharma Services, closed 24 March 2026, endpoint data solutions for clinical trials); and two proteomics software companies, one US and one German, within Analytical Instruments in 2026.

Going the other way: on 27 April 2026 the company agreed to sell its microbiology business to Astorg for approximately $1.075 billion, consisting of cash and a $50 million seller note. The business sits in Specialty Diagnostics and the sale "is expected to close in the third quarter of 2026."

This section is not an argument that the acquisitions were bad. Olink, the filtration business and Clario are all defensible strategic purchases in growing niches, and the 10-K's stated logic — synergies, commercial infrastructure leverage, elimination of redundancy — is the standard and often correct case. The argument is narrower and it is about what you are being asked to pay for. A company that grows earnings per share by buying revenue and retiring shares, financed with debt, at a return on invested capital of 7.44%, is a different asset from a company that compounds organically. The multiple you pay should reflect which one it is, and the honest answer is that on the evidence available in this file we cannot fully tell — because the organic growth rate is not disclosed in any source we have.

3. Growth and margin trajectory — and the number we cannot see

Quarterly, from the vendor income series, cross-checked against the earnings calendar:

Five consecutive beats on both lines, ranging from +1.9% to +5.4% on adjusted EPS and +1.3% to +2.5% on revenue. The second quarter of 2026 was the largest beat of the five and the fastest year-on-year revenue growth in the file. First-half 2026 revenue of $22.999B is up 8.4% on the first half of 2025's $21.219B.

The seasonality is real and should not be mistaken for a trend. The 10-K states plainly: "Revenues in the fourth quarter are historically stronger than in other quarters due to the capital spending patterns of industrial, pharmaceutical and government customers." That is why every first quarter falls 9-10% sequentially. Compare Q1 to Q1 and Q2 to Q2, never sequentially.

The number that is missing, and why it matters more here than anywhere else in this batch

Thermo Fisher's primary non-GAAP measure is organic revenue growth, which the 10-Q defines as "reported revenue growth, excluding the impacts of revenues from acquired/divested businesses and the effects of currency translation." The company reports it prominently and management uses it "to forecast and evaluate the operational performance of the company."

That figure appears in this dive's sources nowhere. It sits in the MD&A tables of the 10-K and both 10-Qs, and numeric tables are stripped from the extracted filing text. It is not in inc_a, inc_q, est, km_ttm, ratios_ttm or any other vendor key. We state the gap rather than estimating around it.

What we can say, from the prose that did survive, is how much of the reported growth the company itself attributes to acquisition:

> "On a reported basis, the clinical research business grew $441 million, which contributed 7 percentage points of reported growth in the segment, primarily driven by the impact of the Clario acquisition." — 10-Q, second quarter 2026, Laboratory Products and Biopharma Services

> "the bioproduction business grew $196 million, which contributed 8 percentage points of reported growth in the segment, driven by higher demand from pharma and biotech customers, as well as the impact from the 2025 acquisition of the filtration and separation business." — 10-Q, second quarter 2026, Life Sciences Solutions

$441 million of clinical-research growth in one quarter, against total company revenue growth of $1.139 billion year on year. If all of that were Clario — and the filing says "primarily" — it would be 39% of the entire quarter's growth from one acquisition that closed three months earlier. We cannot resolve it more precisely than that. It is the single largest open question in this dive and it directly caps the growth score at 4.

The margin picture — and a vendor series that contradicts itself

Gross margin on the trailing twelve months (Q3 2025 through Q2 2026) is 40.37%: $18.707B of gross profit on $46.336B of revenue. That reconciles exactly to ratios_ttm.grossProfitMarginTTM of 0.40372, so the vendor's trailing ratios are computed from the quarterly series.

But the annual series disagrees with its own quarters. The four FY2025 quarters sum to $26.975B of cost of revenue and $17.581B of gross profit; inc_a for FY2025 reports $27.778B and $16.778B. The annual gross profit is $803 million LOWER than the sum of the same year's four quarters. The same year's SG&A is $912 million lower in the annual row than in the quarterly rows. The FY2024-to-FY2025 gross-margin "compression" from 41.34% to 37.66% that the annual series appears to show is therefore an artefact of inconsistent classification, not an economic event, and we reject it. (Full detail at Finding 4.)

Operating margin, which is far less sensitive to the classification problem, is the honest margin lens here. It was 17.4% (2023), 17.9% (2024), 18.2% (2025) and 17.69% on the trailing twelve months — flat to very mildly improving, against 26.3% in 2021. The 10-K and both 10-Qs attribute the improvement to the same thing in every single period: "very strong productivity improvements", which the company defines as the output of its Practical Process Improvement (PPI) Business System — global sourcing, price realisation, headcount reduction and facility consolidation.

That is cost-out, not growth, and the filings quantify it. Restructuring and related charges were "approximately $0.3 billion in 2025 and $0.3 billion in 2024", expected to deliver "annual cost savings of approximately $0.5 billion and $0.2 billion, respectively." Severance actions touched approximately 5% of the workforce in 2025, 2% in 2024 and 5% in 2023. As of 31 July 2026 the company had identified further restructuring actions, primarily in Laboratory Products and Biopharma Services, expected to produce approximately $220 million of additional charges.

Research and development was $1.397B in FY2025 — 3.14% of revenue, and $1.404B on the trailing twelve months (3.03%). It has been essentially flat in dollars since 2021 ($1.406B) while revenue rose 13.6%. For comparison, the semiconductor names in this batch run 11-15% of revenue on R&D. Thermo Fisher's innovation model is substantially acquisitive rather than internal, which is consistent with everything else in this section.

Stock-based compensation is $310 million on $44.556B of revenue — 0.70% of revenue, 0.18% on the vendor's trailing measure. That is genuinely low and it is a real positive: the reported earnings are not being quietly diluted away.

4. Balance sheet and the enterprise-value rebuild

First, the freshness problem — and it is the largest single data defect in this file

bal_a[0] is dated 2025-12-31. This dive is dated 2026-08-04. The balance sheet in the vendor payload is SEVEN MONTHS STALE, and the things that happened in between are not small:

The 10-K had already flagged the first tranche as a subsequent event: "In the first quarter of 2026, the company issued $3.80 billion of senior notes", and "Early in the first quarter of 2026, the company repurchased $3.00 billion (4.9 million shares)."

We scanned all three 8-Ks for financings or acquisitions post-dating the statements and found none. Two of the three (2026-04-23 and 2026-07-23) are bare cover pages announcing quarterly results with the substance in an Exhibit 99.1 that was not extracted — they contain no financial figures whatsoever. The third (2026-05-26) reports the annual meeting and is discussed in Section 10.

The balance sheet as filed, with the two-year comparison

The 10-K corroborates the short-term debt line to the dollar: "As of December 31, 2025, the company's short-term obligations and current maturities of long-term obligations totaled $3.53 billion" against the vendor's $3.533B. Vendor and filing agree. The 10-Q gives the same figure at each subsequent date: $3.09 billion at 28 March 2026 and $3.37 billion at 27 June 2026.

The debt rebuild, and what we can and cannot verify

The June-2026 balance sheet exists in the 10-Q but its numeric table was stripped by the extraction. We therefore rebuild the debt from the cash-flow narrative, which is prose and did survive:

> Total debt at 2025-12-31 $39.385B

> + debt issued in H1 2026 $5.24B

> − debt repaid in H1 2026 $1.41B

> = approximately $43.2B of gross debt at 2026-06-27, before currency translation and discount accretion

The vendor's km_ttm implies net debt of $38.485B (enterprise value $248.358B less market capitalisation $209.873B, which resolves to exactly $38,485,000,000). That is $8.95B higher than bal_a[0]'s $29.533B, which confirms that km_ttm is struck on a June-2026 balance sheet even though bal_a is not. Against our rebuilt gross debt of ~$43.2B, a net figure of $38.485B implies cash and equivalents of approximately $4.7 billion at 27 June 2026, down from $10.105 billion at year end — which is exactly the direction the cash-flow narrative demands after $8.87B of acquisitions and $4.00B of buybacks against $5.24B of new debt.

We use $38.485B of net debt, and we state plainly that it is a vendor figure corroborated by a filing-derived bridge rather than a figure read off a filed balance sheet.

Defect class 2, tested: does the vendor omit short-term investments?

Yes, but the amount is small here. At 2025-12-31 the vendor's netDebt of $29.533B is total debt less cash and equivalents only, ignoring $253 million of short-term investments; the correct figure is $29.280B. At 2024-12-31 the omission was larger, $1.561 billion ($28.766B reported against $27.195B correct). We cannot test the June-2026 figure because that balance sheet is not available, but given a year-end short-term investment balance of only $253 million, the likely error on $38.485B is well under 1% — a materially different situation from META's $45.7B or BRK-B's $321.4B. Tested; real; immaterial at the current date; stated rather than suppressed.

The rebuilt enterprise value

> Market capitalisation $208.815B (369,747,286 shares from the 10-Q cover × $564.75)

> + net debt $38.485B

> = corrected enterprise value $247.300B

The vendor reports $248.358B. The overstatement is $1.058 billion, or 0.43%, and it comes entirely from the share count (Finding 6), not from the debt.

On EBITDA we do not accept the vendor's number. ratios_ttm.ebitdaMarginTTM of 23.599% implies trailing EBITDA of $10.935B, which is $435 million above the sum of trailing operating income ($8.195B) and trailing depreciation and amortisation ($2.305B). The vendor's own quarterly ebitda fields reproduce that $10.935B, but they are not internally consistent either — the June 2026 quarter's $2.572B equals operating income plus D&A exactly, while the September 2025 quarter's $2.838B equals pre-tax income plus interest expense plus D&A. Two different formulas inside one series. We compute EBITDA as operating income plus D&A, consistently, and use $10.500B. We present EV/EBITDA as 23.55x and note that on the vendor's basis it is 22.71x; the choice does not change any conclusion.

Two balance-sheet facts that are risks rather than artefacts

One — the goodwill stack, and a specifically named impairment exposure. Goodwill is $49.362 billion at 31 December 2025, confirmed verbatim by the 10-K ("Goodwill totaled $49.36 billion at December 31, 2025"), which is 44.7% of total assets. Add $15.838 billion of intangibles and 59.1% of the balance sheet is acquisition accounting; tangible book equity is negative $11.793 billion. The annual impairment test in the tenth fiscal month of 2025 passed for all reporting units — but the 10-K then singles one out:

> "Given that the fair value of the reporting unit was not substantially in excess of its carrying value as of the annual 2025 assessment date, relatively small decreases in future cash flows versus anticipated results, decreases in peer trading multiples and/or increases in the weighted average cost of capital could result in impairment of goodwill. The reporting unit consisting of the filtration and separation business had $2.10 billion of goodwill, and an overall carrying value of $4.01 billion as of December 31, 2025."

That is the company itself telling you where the next write-down comes from if the cycle disappoints, on a business it bought four months before the test date.

Two — leverage, and what the covenant does and does not constrain. Net debt of $38.485 billion is 3.67x our trailing EBITDA. Interest coverage on the vendor's trailing measure is 5.59x (operating income $8.195B against interest expense of $1.467B). The only financial covenant in the $5.00 billion revolving credit facility is a minimum Consolidated Net Interest Coverage Ratio of 3.5:1.0 — an interest test, not a leverage test, computed on the facility's own definitions of Consolidated EBITDA and Consolidated Net Interest Expense, and with net interest expense running at roughly $426 million in FY2025 after $283 million of swap benefit, that covenant is nowhere near binding. The facility was entirely undrawn at 27 June 2026 and runs to 7 January 2028. So the leverage is real but it is not constrained by any lender. It is constrained only by the rating agencies and by the company's own appetite, and the record of the last four years is that the appetite is large.

A nuance worth recording: the 10-K states "The total estimated fair value of the company's debt at December 31, 2025 was $36.61 billion" against a carrying value of $39.385 billion. The market values Thermo Fisher's debt $2.78 billion BELOW book, because much of it was issued at coupons well under current rates — the register includes notes at 0.500%, 0.80%, 0.875%, 1.125%, 1.375%, 1.45%, 1.50%, 1.625%, 1.75%, 1.875% and 1.95%. On a fair-value-of-debt basis the enterprise value would be roughly $244.5 billion rather than $247.3 billion. We report EV on carrying value, as is standard, and note the alternative.

5. Cash flow and capital returns

Every one of the 2025 and 2024 capital figures is confirmed verbatim by the 10-K: "During 2025, issuance of debt provided $7.76 billion of cash. Repayment of debt used cash of $2.41 billion... the repurchase of $3.00 billion of the company's common stock (5.8 million shares) and the payment of $0.64 billion in cash dividends", and "During 2025, acquisitions used cash of $4.04 billion... $1.52 billion for the purchase of property, plant and equipment." Clean vendor-to-filing reconciliation on the entire capital-allocation line.

Free cash flow is high quality. Trailing operating cash flow of $9.028B against trailing net income of $6.978B gives an income-quality ratio of 1.29 — cash earnings comfortably exceed book earnings, which is the correct signature for a company carrying $2.3 billion of annual non-cash intangible amortisation. Capital intensity is modest at 3.7% of revenue, though rising: the 10-Q guides 2026 capital expenditure to $1.9-2.1 billion, up from $1.525 billion in 2025.

The uses of that cash tell the story of the business better than any margin. Across 2022 through the first half of 2026: approximately $19.7 billion on acquisitions, $17.0 billion on buybacks and $2.5 billion on dividends — roughly $39.3 billion of uses against roughly $30.1 billion of free cash flow over the same span. The gap was funded by debt and by drawing down the cash balance. In the first half of 2026 alone, $8.87 billion of acquisitions plus $4.00 billion of buybacks plus $0.34 billion of dividends is $13.2 billion of outflow against a first-half free cash flow the filing does not disclose but which cannot plausibly have exceeded $4 billion.

Buyback. $3.00 billion in 2025 (5.8 million shares) and $4.00 billion in the first half of 2026 alone (6.9 million shares), of which $3.00 billion landed early in the first quarter. On the current market capitalisation, $4 billion in six months is an 8 billion dollar annualised pace, or 3.8% of the company. But there is a hard constraint the filing states directly: the 6 November 2025 authorisation was for $5.00 billion, and "At July 31, 2026, $1.00 billion was available for future repurchases." Four fifths of the authorisation was consumed in seven months. Without a new authorisation the buyback stops in the fourth quarter of 2026. We do not assume the board declines to re-authorise — it has re-authorised repeatedly — but any forward shareholder-yield assumption above roughly 2% is a bet on a board action that has not yet happened, and we flag it as such.

Dividend, and a vendor labelling trap. The vendor reports lastDividend: 1.8 and dividendPerShareTTM: 1.80, a yield of 0.32%. That $1.80 is the trailing-twelve-month total, not the last declared quarterly dividend. From the filings: dividends paid were $0.34 billion in the first half of 2026 and $0.16 billion in the first quarter of 2026, which against roughly 370 million shares implies a quarterly rate of approximately $0.46 per share, or a forward run-rate near $1.84 (0.33% yield). Payout ratio is 9.5%. This is not an income stock and it should not be described as one.

Total shareholder yield is therefore roughly 2.2% — approximately 0.33% dividend plus a buyback we mark at a conservative $4 billion a year (1.9%) rather than the $8 billion annualised H1 pace, precisely because only $1.00 billion of authorisation remains.

6. Valuation — priced in or room?

At $564.75 (market capitalisation $208.815B on 369,747,286 shares, corrected enterprise value $247.300B):

Estimate coverage is good on revenue and adequate on the near forward years, and collapses beyond FY2028. Revenue carries 20 analysts on FY2026 and FY2027 and 17 on FY2028. Adjusted EPS carries 14 on FY2026 and 16 on FY2027 — but only 5 on FY2028, ONE on FY2029 and ONE on FY2030. The FY2029 and FY2030 rows rest on a single analyst each and are excluded from every conclusion in this dive. The FY2028 EPS row is used only as a cross-check, never as the primary anchor, and is labelled thin every time it appears.

Which basis are these struck on? Adjusted, and nothing in the payload says so. We verified it: est FY2025 epsAvg of $22.745 against a company that reported $17.74 of GAAP diluted EPS in 2025 — a 28.2% gap that only makes sense on the adjusted basis. Cross-checking the other way, the three reported adjusted quarters of 2025 in earn_cal (Q2 $5.36, Q3 $5.79, Q4 $6.57) sum to $17.72, implying a first quarter of approximately $5.02 and a full year of $22.74 — matching est.epsAvg of $22.745 to within a cent. Confirmed: every forward multiple in this dive is struck on ADJUSTED, non-GAAP EPS.

The GAAP-to-adjusted bridge, quantified

On the trailing twelve months, GAAP diluted EPS is $18.59 and adjusted EPS is $23.83. The gap is $5.24 per share — 22.0% of the adjusted figure, or approximately $1.96 billion of after-tax adjustments.

What is in it, per the 10-Q's own list: acquisition-related transaction costs including inventory step-up; restructuring costs, impairments of long-lived assets, and abandoned-facility expense; gains and losses on investments, business sales and pension settlements; and, the largest item, "the expense associated with the amortization of acquisition-related intangible assets because a significant portion of the purchase price for acquisitions may be allocated to intangible assets that have lives of up to 20 years."

Our position on that last exclusion is that it is defensible in principle and dangerous in practice for THIS company. Excluding intangible amortisation makes an acquisitive business comparable to a non-acquisitive one — that is the stated rationale and it is fair. But Thermo Fisher does not make acquisitions occasionally; it makes them continuously, at roughly $4 billion a year and $8.87 billion in the first half of 2026. A charge that recurs every year because the underlying activity recurs every year is a cost of doing business, not a one-off. The 10-Q even confirms the cash consequence: "Due primarily to the non-deductibility of intangible asset amortization for tax purposes, the company's cash payments for income taxes are higher than its income tax expense."

We therefore run forward valuation on the adjusted basis, because that is the basis on which consensus is struck and mixing bases produces nonsense — but we anchor the trailing multiple on GAAP, and we show a third figure below that we think is more honest than either.

The honest trailing multiple: a tax rate that is not a cash rate

The trailing effective tax rate is 7.85% — $597 million of tax on $7.609 billion of pre-tax income. That is not a typo and it is not sustainable, and the company says so from three directions:

• Discrete benefits, both years. The 10-Q discloses $175 million of deferred tax benefit in the first quarter of 2026 "resulting from the recognition of tax attributes related to domestication transactions" and $148 million in the second quarter "in jurisdictions where the deferred tax assets are now expected to be realized due to forecasted income." That is $323 million in six months, or approximately $0.87 per share — 9.5% of first-half GAAP EPS. The first half of 2025 carried $125M + $153M + $93M = $371 million of the same kind. These "discrete" items are not discrete; they arrive every year, at roughly $350 million a year.
• The company's own guidance. "The company expects its GAAP effective tax rate in 2026 will be between 9% and 11%" (10-Q, July 2026) — revised UP from "between 7% and 9%" in the February 10-K. "The company expects its adjusted tax rate will be approximately 11.5% in 2026."
• The cash rate. "the company's cash payments for income taxes... are expected to total approximately $1.5 billion in 2026." Against $597 million of trailing book tax charge, cash taxes are running at roughly two and a half times book. In 2025 cash taxes were $1.78 billion.

Restating trailing earnings at each of those rates:

Our arithmetic, clearly labelled as ours, using the company's own guided figures as inputs. The honest way to describe Thermo Fisher's trailing valuation is 30x to 34x depending on how you treat a tax line the company itself says will rise, not the 30.4x a screen shows.

Peer context

The vendor peer set contains three genuine comparables — Danaher ($136.9B), Agilent ($39.3B) and IQVIA ($37.6B) — alongside seven names that are not: Abbott, Amgen, AstraZeneca, Intuitive Surgical, Merck, Novo Nordisk and Novartis are pharmaceutical or medical-device companies, i.e. Thermo Fisher's customers, not its competitors. No peer multiples are supplied anywhere in the file, so no peer-multiple comparison is drawn. What is worth recording is scale: Danaher ($136.9B), Agilent ($39.3B) and IQVIA ($37.6B) together are worth $213.8 billion, and Thermo Fisher alone is $208.8 billion — 97.7% of the three of them combined. It is the dominant asset in the sector by a wide margin, which is why owning it is close to owning the sector.

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

At $564.75 — 20.6x FY2027 consensus adjusted EPS and 22.5x FY2026 — the price embeds the following falsifiable claims, each with a number and a date:

• Revenue reaches $47.845B in calendar 2026 (+7.4%) and $50.309B in calendar 2027 (+5.2%). (Consensus-derived; 20 analysts on each.) The nearest check is the 2026-10-28 print, where consensus wants $11.954B — up 7.5% on the September 2025 quarter's $11.122B. Note what this does and does not assume: it does NOT assume an inflection. Consensus models mid-single-digit growth forever. The price is not paying for a re-acceleration; it is paying for the absence of another decline.
• Adjusted EPS compounds from $23.83 trailing to $25.084 (FY2026), $27.404 (FY2027) and $30.208 (FY2028) — about 8.2% a year. (Consensus; 14, 16 and only 5 analysts respectively.) Since consensus revenue grows 5-7%, this requires continuous margin expansion plus continuous buyback. The margin half is credible on the record — the company has delivered "productivity improvements" every period for three years and has $220 million of further restructuring identified. The buyback half is the fragile leg: only $1.00 billion of the $5.00 billion authorisation remained at 31 July 2026.
• Net margin expands from 15.06% trailing to 20.6% by FY2028 — roughly 550 basis points. (Derived by us from the consensus revenue and net-income rows, which cross-check against epsAvg × share count to within 2%.) THIS IS THE MOST FRAGILE ASSUMPTION IN THE PRICE, and it is largely a tax assumption. Consensus net income of $11.008B on FY2028 revenue of $53.485B implies a tax rate close to the company's guided 11.5% adjusted rate applied to a much larger operating base. Against it stands the company's own disclosure that cash taxes will be approximately $1.5 billion in 2026 and that its GAAP rate guidance has already been revised UP from 7-9% to 9-11% within six months.
• The market keeps paying roughly 20x or better on the forward year at every date between now and then. (Our number.) At 18x FY2027E the stock is $493; at 23x it is $630. The stock has already traversed $438.34 to $639.45 inside twelve months, so both ends of that band are inside its recent range. The multiple, not the earnings, is the widest source of variance.
• And the assumption nobody is pricing explicitly: that the acquisitions keep working. (Our observation.) The price rests on a return on invested capital of 7.44% being an acceptable use of $4-9 billion a year. Watchable number: the goodwill balance at the next annual impairment test, and specifically the filtration-and-separation reporting unit that the 10-K says has a fair value "not substantially in excess of its carrying value" on $4.01 billion of carrying amount.

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

Expected return over the next twelve months decomposes as: adjusted EPS growth (+10.2%, from FY2027E $27.404 to FY2028E $30.208) + multiple drift + shareholder yield (+2.2%).

If the multiple simply HELD — that is, if in twelve months the market pays the same ~20.6x on the then-forward year (FY2028E $30.208) that it pays on FY2027E today — the price would be $622, a gain of 10.2% plus 2.2% of yield, or about +12.4%. That is the no-re-rating case and it is a perfectly reasonable outcome.

Our base of $575 assumes MILD COMPRESSION, and we should be explicit about why, because it is not a rolldown artefact. We assume the market pays roughly 19x on FY2028E in twelve months rather than 20.6x, for three stated reasons: (i) consensus revenue growth decelerates on its own path, from 7.4% in FY2026 to 5.2% in FY2027, so the growth being paid for is getting slower, not faster; (ii) the earnings growth that remains is increasingly buyback- and margin-driven at a company whose repurchase authorisation is 80% consumed; and (iii) the tax rate the earnings rest on is disclosed as rising, and the company's own GAAP guidance already moved up two percentage points in six months.

Almost all of the expected return under either construction is earnings growth, and it is a modest amount of earnings growth. There is no version of this thesis in which multiple expansion does the work — the bull case at $666 requires 23x on the FY2027 consensus HIGH, which is a re-rating of only 12% off today's forward multiple applied to an earnings beat. That is deliberately conservative construction. The bear case at $449 is the mirror: 17x applied to the FY2027 consensus LOW, which is a de-rating and a miss together, and lands almost exactly on the actual 52-week low of $438.34.

The asymmetry is the finding: 17.9% of upside against 20.6% of downside is 0.87 to 1 — pointing the wrong way.

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

• On the direction of the business we do not differ from the street at all, and that is itself the verdict input. The street shows 36 buy, 7 hold and ZERO sell across 43 analysts, with a consensus target of $592.11 — only 4.8% above spot, and a low target of $520 that is 7.9% below spot. Our base of $575 is 2.9% BELOW consensus. When our own arithmetic lands below the crowd's on a name where the knowledge base contributes literally nothing, the honest conclusion is that we have no edge here. We state that rather than manufacturing a differentiated view.
• We differ on how the trailing multiple should be quoted, and materially. Every screen shows Thermo Fisher at 30.4x trailing earnings. On the company's own guided cash-tax figure of approximately $1.5 billion for 2026, the comparable trailing multiple is 34.4x. That is not a small restatement on a stock whose entire investment case is "quality compounder at a reasonable price". Watchable number: the effective tax rate in the 2026-10-28 print. A third-quarter rate above 11% with no discrete benefit would confirm the normalisation; another $150 million discrete benefit would confirm that the reported rate is being managed.
• We differ on how much of the second-quarter acceleration to credit, and we are explicit that we cannot resolve it. The market read a +10.5% revenue quarter and a fifth consecutive beat as an inflection. The 10-Q attributes $441 million of clinical-research growth in that same quarter "primarily" to an acquisition that closed three months earlier for $8.87 billion of cash. We are not asserting the organic rate is weak — we are asserting that the number which would settle it is in a stripped table and is unavailable to us, and that paying a premium multiple for growth you cannot decompose is not a defensible act. Watchable number: the organic revenue growth rate disclosed on 2026-10-28. Above 5% and the inflection is real; at 2-3% and the last four years are still running.
• Positive surprise that would force a re-rate: a third-quarter organic growth rate above 5% with academic and government revenue growing rather than flat (the 10-Q says it was flat across the first half and grew in the second quarter — the trend matters); the microbiology sale closing at or above $1.075 billion; a new buyback authorisation materially above $5 billion; or clear evidence that the instrument cycle has turned, which after three consecutive down years would be the most important single datapoint about this company.
• Negative surprise that would break the thesis: a goodwill impairment against the filtration-and-separation reporting unit, which the 10-K has pre-announced as the exposure; a fourth consecutive annual decline in Instruments revenue; a GAAP tax rate above 13% without offsetting operating improvement; another $5-billion-plus acquisition funded with debt at 3.7x leverage; or a US federal research-funding outcome materially worse than the 2026 appropriation.

Synthos fair values

All three anchors are multiples of the FY2027 consensus adjusted-EPS distribution (mean $27.404, low $26.388, high $28.958, 16 analysts — the deepest EPS line in the file), cross-checked against the FY2028 mean of $30.208 (5 analysts, thin, used only as a cross-check).

• Bear ~$449 — 17x the FY2027 consensus LOW of $26.388. Cross-check: 14.9x FY2028E; and $449 is within 2.4% of the actual 52-week low of $438.34, i.e. the bear case is "back to where it traded three months ago". The scenario: organic growth proves to be 2-3%, the instrument cycle stays down, the tax rate normalises toward the cash rate, an impairment lands on the filtration reporting unit, and the multiple de-rates to what a low-single-digit compounder with 3.7x leverage deserves. −20.6%.
• Base ~$575 — 21x the FY2027 consensus MEAN of $27.404. Cross-check, and it is a strong one: 19x the FY2028 consensus mean of $30.208 gives $574 — the two constructions converge to within $1.50 from completely different years. Further cross-check: corrected EV of $251.1B on FY2027E revenue of $50.309B is 4.99x sales, against 5.34x trailing. Sensitivity, stated openly: 19x FY2027E gives $521 (−7.8%) and 23x gives $630 (+11.6%). The whole answer lives in a 19-23x band and we claim no more precision than that. The scenario: mid-single-digit revenue growth as consensus models, continued productivity-driven margin gains, a tax rate drifting up toward guidance, a buyback that continues but at a slower pace, and a multiple that drifts modestly lower as the growth rate is revealed to be what it is. +1.8%, plus roughly 2.2% of shareholder yield.
• Bull ~$666 — 23x applied to the FY2027 consensus HIGH of $28.958. Cross-check: 22.0x FY2028E; and $666 is 4.1% above the 52-week high of $639.45, so the bull case requires a new high. This bull case needs only a 12% re-rating off today's forward multiple plus an earnings beat — it is not a heroic construction. The scenario: the organic rate proves to be 5-6%, the instrument cycle turns, Clario integrates cleanly and the pharma and biotech end market keeps strengthening as the 10-Q describes. +17.9%.

Base is 1.8% above spot; asymmetry is 0.87:1 — against us (20.6% down, 17.9% up). A base case one point eight percent above the price, below the street, with negative asymmetry and an empty conviction lane, is the textbook definition of a fairly valued security. That is a Hold, and it is a Hold by arithmetic rather than by hesitation.

7. Knowledge base — zero claims name this company

Raw hits across six sweeps: 79. Entity matches naming Thermo Fisher: 0. Text-only matches naming Thermo Fisher: 0. Used as name-level conviction: 0. Retained as labelled end-market context: 6. Discarded: 73.

The collision problem was exactly as anticipated and it was total. Every one of the 33 text-only matches in the primary sweep is a false positive on a string, not a company. The prefix "thermo" appears in thermodynamics (six claims about entropy, Landauer limits and thermodynamic computing), thermoregulation (five claims from health-science channels about sauna, sleep and core body temperature), thermometer (used metaphorically about Bitcoin and literally about a manufacturing inspection at an unrelated biotech), and atmospheric / atmosphere (five claims about lunar industry, jet streams, microwave latency and nuclear reactor design, which match on "atmo-" inside "atmosphere"). Not one of the 33 is about a company, let alone this company. This is precisely the failure mode the four-lane attribution policy exists to catch, and it is why text-only matches require individual triage rather than aggregation.

Sweep 5 is the decisive one. A search across the full 51,928-claim corpus for Thermo Fisher Scientific, Fisher Scientific, PPD, Patheon, Applied Biosystems and Invitrogen — the company's own legal name plus five of its largest brands, several of which were multi-billion-dollar public companies in their own right — returned zero matches of any kind, entity or text. The knowledge base does not contain the string "Thermo Fisher Scientific" anywhere.

A search on the chief executive's name, Marc Casper, returned zero hits.

What we do retain, labelled as end-market context and nothing more

Six claims bear on markets Thermo Fisher sells into. None of them mentions the company. None is used to justify a conviction premium, a multiple, or a single dollar of the fair value. They appear here because refusing to use them at all would be as dishonest as pretending they are company claims.

> 2026-01-23 · bullish · conviction 65 · horizon: fact · entities: NIH, CDC · channel: biotech_hangout · no named speaker

> "Congress moderately increased NIH (~$49B) and kept CDC flat, avoiding the administration's proposed ~40% cuts—relief for the sector."

> 2025-09-30 · bearish · conviction 60 · horizon: fact · entities: NASA, NSF, NIH · channel: all_in · no named speaker

> "US research funding is being cut sharply — NSF graduate fellowships halved this year, NIH/NASA reductions — creating a serious headwind for academic science hiring."

These two point in opposite directions on the same question — the academic and government market that the 10-K says drove Thermo Fisher's 2025 revenue decline. The bullish one is fresher (January 2026 versus September 2025) and it is a fact claim about an enacted appropriation rather than a forecast. Both carry speaker: null and are attributed only to a channel, which under the four-lane attribution policy is the weakest sourcing we admit. Net: mildly supportive of the 10-Q's observation that academic and government revenue grew in the second quarter of 2026 after declining through 2025.

> 2025-12-15 · bearish · conviction 70 · horizon: thesis · entities: WuXi · channel: biotech_hangout · no named speaker

> "Watered-down Biosecure Act becomes law, barring federally-funded firms from working with Chinese 'companies of concern' (CDMOs/CROs), pressuring China supply chains."

If accurate, this is a tailwind for the pharma-services and clinical-research businesses inside Laboratory Products and Biopharma Services, because the named Chinese contract manufacturers and research organisations are the direct alternative to Patheon and PPD. We note the direction and assign it no weight, because the claim is about a legislative outcome, not about Thermo Fisher, and no filing in this archive mentions the legislation.

> 2025-08-02 · bullish · conviction 70 · horizon: principle · entities: Danaher, TransDigm, Constellation Software, Berkshire Hathaway · channel: cognitive_revolution · no named speaker

> "Compounders that buy companies, improve them, and redeploy free cash flow into more acquisitions have been outstanding public-market performers and hedge-fund favorites."

This is the serial-acquirer archetype, and it is the closest thing in the entire corpus to a claim about Thermo Fisher's business model — except that it names Danaher, not Thermo Fisher, and it is a claim about a category rather than a company. We record it because it is the frame a bull would use here. We also record the obvious tension: the archetype's defining feature is redeploying free cash flow into acquisitions, and Thermo Fisher's acquisitions in the first half of 2026 exceeded its free cash flow and were funded with debt.

The conclusion, stated as a finding rather than dressed up

The Synthos knowledge base has no view on Thermo Fisher Scientific. Not a thin view, not a sector-level view standing in for a company view — no view at all. Zero claims name it, zero claims name its brands, zero claims name its chief executive, and the 33 apparent matches on its name are string collisions with the physics of heat.

This is not a defect in the knowledge base; it is information. The corpus is dense on semiconductors, artificial-intelligence capital expenditure, crypto, macro and clinical-stage biotech, and it is silent on the industrial suppliers underneath biotech. What it means for this dive is simple: every judgement in it comes from the filings and the vendor payload, with no independent expert corroboration of any kind. Conviction is None, breadth is 0, claim count is 0, and the absence is a direct input to the Hold verdict — under house policy, no coverage pushes toward Watch or Hold, and here it lands on Hold because the arithmetic independently says the stock is fairly priced.

8. Data integrity — what we rejected from the vendor file and why

Fourteen findings. Listed rather than silently corrected. Thermo Fisher's vendor file is average for this batch: the revenue-type and geographic tables reconcile essentially exactly, the capital-return series matches the 10-K verbatim, and the reporting currency is clean — but the balance sheet is seven months stale, the annual and quarterly income series contradict each other by $800 million, and one technical field cannot be reconciled with the others.

1. bal_a[0] is dated 2025-12-31 — SEVEN MONTHS STALE — and misses the largest capital events in the company's recent history. REJECTED as a current balance sheet. Between that date and this dive Thermo Fisher completed the Clario acquisition for $8.87 billion of cash (24 March 2026), issued $5.24 billion of debt, repaid $1.41 billion, repurchased $4.00 billion of stock and paid $0.34 billion of dividends — all confirmed in the 10-Q filed 2026-07-31. The 10-K had already flagged $3.80 billion of the note issuance and $3.00 billion of the buyback as first-quarter-2026 subsequent events. No balance-sheet figure from bal_a[0] is presented in this dive as a current figure. We rebuild June-2026 gross debt at approximately $43.2 billion from the filing's cash-flow narrative and use the vendor's km_ttm-implied net debt of $38.485 billion, stating openly that the June balance sheet itself is unavailable because the 10-Q's numeric tables are stripped by the extraction.

2. All three 8-Ks scanned for post-statement financings and acquisitions — NONE FOUND, and two are bare cover pages. The 2026-04-23 and 2026-07-23 filings are Item 2.02 earnings announcements whose entire substance is in Exhibit 99.1, which was not extracted; both files contain no financial figures whatsoever. Stated verbatim as a finding, per the extraction-limits rule. The 2026-05-26 filing is a complete Item 5.07 annual-meeting report and is the only 8-K in this archive with usable content — see Finding 13.

3. km_ttm.netDebtTTM omits short-term investments — CONFIRMED, quantified, immaterial at the current date. At 2025-12-31 the vendor's netDebt of $29.533B nets total debt against cash and equivalents only, ignoring $253 million of short-term investments; the correct figure is $29.280B. At 2024-12-31 the omission was $1.561 billion. We cannot test the June-2026 figure because that balance sheet is unavailable, but on a $253 million year-end balance the likely error on $38.485B is under 1%. Tested, real, small, and stated rather than suppressed.

4. The annual and quarterly income series contradict each other by $803 million on gross profit — the FY2025 gross-margin "compression" is REJECTED as an artefact. The four FY2025 quarters sum to cost of revenue $26.975B and gross profit $17.581B; inc_a FY2025 reports $27.778B and $16.778B. The same year's SG&A is $912 million lower in the annual row ($7.272B) than in the quarterly rows ($8.184B). Read the annual series alone and gross margin appears to collapse from 41.34% (FY2024) to 37.66% (FY2025) — a 368 basis point compression that did not happen. The trailing ratios confirm the quarterly series is the consistent one: grossProfitMarginTTM of 0.40372 reproduces exactly from the quarterly figures. We use operating margin as the primary margin lens for annual comparisons (annual $8.109B against a quarterly sum of $8.002B — a 1.3% difference rather than a 4.8% one) and we draw no gross-margin trend conclusion from the annual series at all.

5. Operating expenses carry an unlabelled residual in 2026 quarters that is absent in 2025 quarters — year-on-year expense comparisons REJECTED. In every FY2025 quarter, operatingExpenses equals R&D plus SG&A exactly. In Q1 2026 it exceeds R&D plus SG&A by $479 million and in Q2 2026 by $583 million — $1.062 billion in the first half that the vendor does not name. The filings identify the plausible components (restructuring and other costs, transaction-related costs for Clario, impairment of long-lived assets including a specific $60 million charge in Laboratory Products and Biopharma Services), but the vendor supplies no label. The effect is that Q2 2026 operating margin of 17.4% is not computed on the same basis as Q2 2025's 17.7%, and we do not present that comparison as a margin change.

6. Share count: the vendor market capitalisation is struck on the MARCH cover count, not the current one — CORRECTED. Market capitalisation of $209,872,959,750 divided by $564.75 gives 371,621,000 shares. The 10-Q for the first quarter states 371,621,465 shares outstanding at 2026-03-28 — a match to within 465 shares. But the current 10-Q states 369,747,286 shares outstanding at 2026-06-27, and the FY2025 10-K cover states 371,484,244 at 2026-01-31. The company repurchased 6.9 million shares in the first half of 2026, and the vendor has not picked up the second-quarter portion. We use 369,747,286, giving a market capitalisation of $208.815 billion against the vendor's $209.873 billion — an overstatement of $1.058 billion, or 0.51% — and the enterprise value, EV/Sales and EV/EBITDA figures are recomputed on it. No share-class complication: there is a single class of common stock, CUSIP 883556102.

7. est.epsAvg is ADJUSTED (non-GAAP) and nothing in the payload says so — basis stated explicitly throughout. Verified two ways. First, est FY2025 epsAvg of $22.745 against reported GAAP diluted EPS of $17.74 — a 28.2% gap that only exists on the adjusted basis. Second, the three reported adjusted quarters of 2025 in earn_cal sum to $17.72, implying a first quarter of ~$5.02 and a full year of ~$22.74, matching est.epsAvg to within a cent. Every forward multiple in this dive is struck on adjusted EPS and labelled as such; the trailing multiple is shown on GAAP (30.4x), adjusted (23.7x) and our cash-tax restatement (34.4x).

8. est.ebitdaAvg and est.ebitAvg are internally coherent but on a different basis from reported GAAP — NOT USED for valuation, per the data contract. Tested against actuals: FY2025 ebitAvg of $8.938B is 10.2% ABOVE actual GAAP operating income of $8.109B, and ebitdaAvg of $12.089B is 11.0% above our computed FY2025 EBITDA of $10.889B. No arithmetic impossibility — ebitdaAvg exceeds ebitAvg in every forward year, no negative rows, no revenue-mirroring signature, and no year where revenue falls below the prior year. The rows are simply adjusted-basis while the actuals are GAAP. This file does not trip any of the four detection signatures in the data contract; it is a clean est block on that test, and the deviation is a basis mismatch rather than a defect. All forward valuation still runs on epsAvg and revenueAvg.

9. Analyst counts collapse after FY2028 — FY2029 and FY2030 EXCLUDED from every conclusion. Revenue coverage: 20 (FY2026), 20 (FY2027), 17 (FY2028), 14 (FY2029), 8 (FY2030). EPS coverage: 14, 16, 5, ONE, ONE. The FY2029 and FY2030 adjusted-EPS rows rest on a single analyst each and are not used anywhere. The FY2028 EPS row (5 analysts) is used only as a fair-value cross-check and is labelled thin at every appearance. The FY2027 row (16 analysts) is the primary anchor.

10. seg_prod is the revenue-BY-TYPE table, not the segment table, and the segment rows it does contain use a SUPERSEDED taxonomy — corrected in presentation, accepted in substance. The three FY2018-2025 lines (Consumables, Instruments, Service) are the 10-K Note 6 disaggregation; the four reportable segments appear only in the FY2017 and earlier rows, and under old names — "Laboratory Products and Services" (renamed Laboratory Products and Biopharma Services) and "Analytical Technologies" (a segment that no longer exists). The 2010-2012 rows contain only "Intersegment Eliminations" and "Excluded from Reportable Segment Operating Income". We verified the sums on the usable years: FY2025 totals $44,557M against reported revenue of $44,556M (+$1M), FY2024 $42,878M against $42,879M (−$1M), and FY2023, FY2022 and FY2021 EXACT. No double-counting. The table is used, correctly labelled. The four-segment revenue and margin split is NOT AVAILABLE from any source in this file — the 10-K and 10-Q Note 11 tables are numeric and were stripped — and no segment-level figure appears in this dive.

11. seg_geo changes label basis between FY2023 and FY2024 — the cross-year comparison of the largest line is REJECTED. FY2023 and earlier report "North America" and "Other Regions"; FY2024 and FY2025 report "UNITED STATES" and "All Other Countries". North America includes Canada and Mexico. The apparent decline from $22.764B (FY2023) to $22.504B (FY2024) is not like-for-like and no conclusion is drawn from it. The sums are otherwise clean: $44,557M (FY2025), $42,878M (FY2024) and $42,857M (FY2023) against reported revenue of $44,556M, $42,879M and $42,857M. The FY2024-to-FY2025 comparison is used.

12. Multiple zeroed fields, one of which the filing directly contradicts. Scanning every series: (a) cf_a FY2025 reports incomeTaxesPaid: 0 and interestPaid: 0 — the 10-K states "Cash payments for income taxes were $1.78 billion during 2025." THE FILING WINS: the vendor reports $0; the 10-K shows $1.78 billion; we use $1.78 billion, and it is load-bearing for the tax discussion in Section 6. (b) bal_a FY2025 capitalLeaseObligations: 0 against $1.239B (FY2024) and $1.244B (FY2023). Thermo Fisher has not eliminated its finance leases; the FY2024-to-FY2025 total-debt comparison is therefore not on a consistent basis, and the true increase is either +$6.61B or +$7.85B depending on where the leases went. (c) bal_a FY2025 otherCurrentLiabilities: 0 against $2.938B (FY2024), with accruedExpenses simultaneously jumping from $1.988B to $5.324B — a reclassification, not an economic event. (d) bal_a FY2025 deferredRevenueNonCurrent: 0 against $1.138B (FY2024). (e) bal_a FY2025 longTermInvestments: 0 against $560M (FY2024). We use no year-on-year comparison of any of these five lines.

13. Currency — TESTED AND CLEAN on the reporting basis, but the DEBT is multi-currency and the filings say so. inc_a[0].reportedCurrency, inc_q[0].reportedCurrency and bal_a[0].reportedCurrency are all USD; profile.currency is USD; the market capitalisation reconciles to price times shares in USD. There is no ASML- or TSM-style currency mixing in this file and every multiple in this dive is struck USD against USD. But the underlying debt is not dollar-denominated. The 10-Q discloses that a wholly-owned finance subsidiary, Thermo Fisher Scientific (Finance I) B.V., has issued a series of registered public "Euronotes" — the Floating Rate Notes due 2027 and the 0.80% due 2030, 1.125% due 2033, 3.628% due 2035, 1.625% due 2041 and 2.00% due 2051 — that "Interest is payable annually on the euro and public Swiss franc-denominated fixed rate senior notes, quarterly on the euro-denominated floating rate senior note", that there are Japanese yen-denominated and Swiss franc-denominated private placement notes hedged with cross-currency swaps, and that the euro commercial paper programme may issue in "euro, U.S. dollars, Japanese yen, British pounds sterling, Swiss franc, Canadian dollars or other currencies." This matters for one specific number. In FY2025 the vendor's total debt rose from $32.775B to $39.385B (+$6.61B) while the 10-K's cash-flow narrative reports net debt issuance of only $5.35 billion ($7.76B issued less $2.41B repaid) — a residual of roughly $1.3 billion, and up to $2.5 billion if the vanished finance-lease line is stripped from the base. That residual is currency translation on the euro, Swiss franc and yen borrowings plus discount accretion. It is disclosed in the filings and appears in no vendor field. The 10-K quantifies the broader exposure: "A 10% depreciation in year-end 2025 functional currencies, relative to the U.S. dollar, would result in a reduction of shareholders' equity of approximately $2.93 billion" — 5.5% of total equity — and confirms the deliberate design: "The company also uses foreign currency-denominated debt to partially hedge its net investments in foreign operations." Reporting currency clean; debt stack materially non-dollar; stated because the check was run and the answer is more interesting than "clean".

14. tech.max_dd_from_peak of −15.36% cannot be reconciled with tech.hi52 and tech.lo52 — REJECTED and not used. The 52-week range is $438.34 to $639.45, which is a peak-to-trough decline of 31.45% inside the window. A maximum drawdown from peak of 15.36% is arithmetically inconsistent with that range under any ordering of the high and the low that also satisfies the reported return series. We reject the field, do not quote it anywhere, and instead describe the path from the return series directly (see Section 9). Related, and separately reported: quote.yearHigh/yearLow ($643.99 / $435.27) disagree with tech.hi52/lo52 ($639.45 / $438.34) by 0.71% and 0.70%, and quote.priceAvg50/priceAvg200 ($504.81 / $530.72) disagree with tech.sma50/sma200 ($509.56 / $531.15). We use the tech block throughout, per the data contract, because it is computed from the same close series that produces the moving averages and relative returns, and mixing sources inside one section would make the percentages inconsistent.

Not defects — correctly reported and independently confirmed. The 10-K's "short-term obligations and current maturities of long-term obligations totaled $3.53 billion" matches bal_a shortTermDebt of $3.533B exactly. The 10-K's "Goodwill totaled $49.36 billion" matches bal_a goodwill of $49.362B. The entire FY2025 and FY2024 capital-allocation series — $3.00B and $4.00B of buybacks, $0.64B and $0.58B of dividends, $4.04B and $3.13B of acquisitions, $1.52B and $1.40B of capital expenditure — matches the 10-K verbatim. ratios_ttm.grossProfitMarginTTM, researchAndDevelopementToRevenueTTM, salesGeneralAndAdministrativeToRevenueTTM, operatingProfitMarginTTM, effectiveTaxRateTTM and priceToEarningsDilutedRatioTTM all reproduce exactly from the quarterly series. incomeQualityTTM of 1.290 correctly reflects trailing operating cash flow of $9.028B against net income of $6.978B.

Non-equity tripwire — checked and passed. TMO is common stock, NYSE-listed, CUSIP 883556102, ISIN US8835561023, single class, IPO 17 March 1980. The price of $564.75 is not par-like ($25, $50 or $1,000); beta is 0.868, well above the sub-0.3 typical of a fixed-income-like instrument; the dividend is variable, small (0.32% yield) and only 9.5% of earnings rather than a fixed coupon; volume was 1,855,417 shares (roughly $1.05 billion of turnover); and the 52-week band of $438.34 to $639.45 is a 45.9% range — the opposite of the narrow band a baby bond or preferred shows. This is common equity.

One thing we looked for and did not find. Following the off-balance-sheet rule, we grepped the 10-K and both 10-Qs for commitments, unrecorded, not yet commenced, guarantee and purchase obligations. The 10-K states that the company "has unconditional purchase obligations in the ordinary course of business" and the 10-Q states that commitments "did not change materially subsequent to December 31, 2025, except in connection with the completion of the Clario acquisition" — but the quantified table is numeric and was stripped. There is a guarantee disclosure — Thermo Fisher "has fully and unconditionally guaranteed all of Thermo Fisher International's obligations under the Euronotes" — but that is a parent guarantee of consolidated debt already on the balance sheet, not an off-balance-sheet exposure. We found no META- or ORACLE-scale hidden commitment, and we state that we looked.

9. Technicals

• Price $564.75. −11.68% below the 52-week high of $639.45; +28.84% above the 52-week low of $438.34. Position within the annual range: 62.9th percentile — upper-middle.
• Moving averages, and the configuration is worth stating precisely. The price is +10.83% above a 50-day average of $509.56 and +6.33% above a 200-day average of $531.15. So the price is above BOTH. What is unusual is the relationship between the two: the 50-day average sits 4.06% BELOW the 200-day average. That is a chart in which the short-term average has not yet caught up to the long-term average even though the price has passed both — the signature of a sharp recovery out of a deep hole, where the recent lows are still dragging the fifty-day mean down. It is a constructive configuration, not a broken one, but it is not the same as a clean uptrend.
• RSI 62.4 — firmly positive, not overbought. MACD +19.05 — positive and the strongest technical reading in this file.
• The path, reconstructed from the return series (because tech.max_dd_from_peak is rejected at Finding 14). Twelve-month return +21.97%, six-month −2.93%, three-month +22.08%. Chaining those: the price twelve months ago implies roughly $463; six months ago roughly $582; three months ago roughly $463. That gives a precise shape: up approximately 25.6% over the first six months of the window, then DOWN approximately 20.5% in the following three months, then up 22.1% in the last three. The 52-week high of $639.45 and low of $438.34 sit inside that path. This is a violent V, and the reader is buying it four-fifths of the way back up the right-hand side.
• Relative performance, and it is the least flattering table in the dive. 3-month +22.1% against SPY +7.6% and QQQ +7.7% — an outperformance of 14.5 points. 6-month −2.9% against SPY +11.1% and QQQ +15.7% — an underperformance of 14.0 points. 12-month +22.0% against SPY +24.3% and QQQ +30.8% — Thermo Fisher has UNDERPERFORMED the S&P 500 over twelve months despite the last three months. Every point of the twelve-month gain and more was made in the last quarter.
• Sentiment: 36 buy, 7 hold, zero sell across 43 analysts; consensus target $592.11 (+4.8%), median $600, high $650 (+15.1%), low $520 — 7.9% BELOW spot. A consensus target less than 5% above the price, with no sell ratings and a low target below the price, is the street saying the same thing our arithmetic says: fairly valued.
• Vendor composite rating: B / 3 out of 5, with sub-scores of 1 on debt-to-equity, 2 on price-to-earnings and 2 on price-to-book, against 4 on return on equity and return on assets. The debt-to-equity score of 1 is directionally correct here and we do not reject it — unlike KLAC, where buybacks had artificially suppressed book equity, Thermo Fisher's leverage is genuine: $38.485 billion of net debt at 3.67x EBITDA against negative tangible book. The price-to-book score is the one we set aside, because a 3.98x price-to-book on a balance sheet that is 59% goodwill and intangibles is not measuring anything.

Today's move and what it does to the entry

Thermo Fisher closed 2026-08-04 at $564.75, DOWN 1.62% or $9.28 from a previous close of $574.03. It opened at $573.57 — essentially flat — traded UP to a day high of $582.31, then reversed and closed at $564.75, which is 3.02% below the day's high and only 0.40% above the day's low of $562.50. Volume was 1,855,417 shares against a 2,410,000 average — 77% of normal.

That is a weak close: an intraday reversal from a 1.4% gain to a 1.6% loss, finishing on the lows, on light volume.

It happened on a day the broad market and the semiconductor complex rose sharply. The batch context for 2026-08-04 records MRVL +12.8%, INTC +10.9%, LRCX +7.9%, MU +7.6%, KLAC +7.0%, AMD +7.0%, AMAT +5.5% and ASML +4.2%.

We are not going to tell you why Thermo Fisher fell, because this file does not say. There is no company-specific news in any source available to this dive for 2026-08-04. The last company events were the second-quarter earnings release on 2026-07-23, twelve days earlier, and the 10-Q filing on 2026-07-31, four days earlier — and the earnings release beat on both lines while the 10-Q described strengthening demand across all end markets. A one-day divergence between a defensive healthcare name and a semiconductor melt-up is a common pattern and rotation is the obvious candidate explanation, but "obvious candidate" is not evidence and we are not presenting it as a finding. What we will say is what is verifiable: the move was not accompanied by heavy volume (77% of average), it was not accompanied by any disclosed company event, and it followed a fifth consecutive earnings beat by twelve days.

What the day does to the entry is almost nothing, and that is the point. A 1.62% decline against a base fair value that sits 1.8% above the closing price moves the gap from 0.1% to 1.8%. You are not being offered a discount; you are being offered the same fair price you were offered yesterday.

The honest read on the chart. This is a large-capitalisation defensive compounder four-fifths of the way back up from a 20.5% drawdown, with RSI at 62.4, MACD positive, price above both moving averages, and a fifty-day average still 4.1% below the two-hundred-day. In pure chart terms that is constructive. What it is not is cheap: even after the drawdown and the recovery, the stock trades at 30.4x trailing GAAP earnings, 34.4x on a cash-tax basis, 20.6x the FY2027 consensus and 5.34x sales, and our base fair value of $575 is only 1.8% above the closing price and 2.9% below the street's.

The setup that would change this is a single number: a price near $500. That is 18.2x the FY2027 consensus and 16.6x the FY2028 consensus, it would put our base of $575 at a 15% premium, and it is only 11.5% below today's close — comfortably inside a range this stock has traversed twice in the last six months. At $500 the risk and reward genuinely change. At $564.75 they do not.

10. Insiders — nothing to see, and one governance vote that says more than the trades

Eight transactions in the file, spanning 2026-05-20 to 2026-07-27.

Read carefully, this table contains almost no information and we will not pretend otherwise.

Six of the eight transactions are director compensation. Five are quarterly phantom stock unit credits — deferred-compensation bookkeeping entries at zero price, totalling 338.67 units across five directors, which is dividend-equivalent accrual, not a decision. The 20 May 2026 award of 499 common shares to Dion Weisler is the annual director equity grant, and the F-InKind transaction eight days later is 80.287 shares withheld by the issuer to satisfy the tax on it — not an open-market sale.

There is exactly one open-market transaction in the entire file: a sale of 400 shares by President and Chief Operating Officer Gianluca Pettiti on 27 July 2026 at $565.00, for $226,000, leaving him holding 24,650.83 shares. That is 1.6% of his position, and it is a rounding error against a $208.8 billion company. It carries no signal and we score it neutral. The one detail worth noting is the price: $565.00, which is within 0.05% of where the stock closed on the day of this dive — so an officer transacted at today's level eight days ago.

There are zero open-market purchases in the file.

The disclosure that actually matters is in the 8-K, and it is not a trade

At the annual meeting held on 20 May 2026, shareholders VOTED DOWN the company's say-on-pay proposal. From the 8-K filed 2026-05-26, Item 5.07, verbatim:

> "2. A non-binding, advisory proposal on the compensation of the Company's named executive officers was not approved."

A failed say-on-pay vote at a company of this size is uncommon and it is a real governance event. The vote is advisory and non-binding, and the 8-K does not disclose the margin (the tabulated results are in a stripped numeric table), so we cannot say how close it was. But the direction is unambiguous: a majority of voting shareholders declined to endorse executive compensation at a company whose revenue has been flat for four years and whose earnings per share has risen 0.6% over three years while it spent $19.7 billion on acquisitions. In the same meeting, the full board slate was elected and PricewaterhouseCoopers was ratified as auditor — so this was a targeted rejection of the pay proposal, not general shareholder revolt.

And one further item from the 10-Q, Item 5, disclosed under Rule 408(a):

> "Mr. Casper modified his Rule 10b5-1 trading arrangement, which was originally adopted on November 11, 2025."

A chief executive modifying a pre-existing selling plan is a routine and entirely lawful administrative act, and the disclosure does not say whether the modification increased or decreased the shares authorised for sale — only that "the aggregate number of shares reported in this table represents the shares authorized for potential sale following the modification", and that table is numeric and stripped. We report it at face value, assign it no weight, and note it only because it is the single piece of chief-executive share-disposition information in the entire archive, and because it sits eleven days before a failed say-on-pay vote.

Net insider read: neutral on the trades, mildly negative on the governance.

11. Verdict, kill-criteria and flip conditions

Hold.

This is a fairly valued security with no edge attached to it, and that sentence is doing all the work.

What is genuinely good here, and none of it is in dispute: the most complete supplier position in the physical infrastructure of biological science, spanning four segments, 125,000 employees and roughly 70 countries with no customer concentration disclosed anywhere in the 10-K; a revenue base that is 83.6% consumables and services with only 16.4% in one-time instrument sales; $29.70 billion of remaining performance obligations at 27 June 2026, two-thirds of a year's revenue already contracted; $7.318 billion of trailing free cash flow converting at 105% of net income on capital expenditure of just 3.7% of revenue; stock-based compensation of only 0.70% of revenue; a $5.00 billion revolver, undrawn, whose only covenant is an interest-coverage test that is nowhere near binding; and a second quarter of 2026 that grew revenue 10.5% year on year and beat on both lines for the fifth consecutive time, with the 10-Q describing strengthening demand across all four end markets and every major geography including China.

What we are declining to pay for: four consecutive calendar years with no revenue growth — $44.915B, $42.857B, $42.879B, $44.556B, a compound rate of minus 0.27%; diluted GAAP EPS that rose 0.6% over three years and only because the share count fell 4.3%; roughly $19.7 billion of cash acquisitions and $5.3 billion of added goodwill to bridge that gap, taking goodwill to 44.7% of total assets and tangible book equity to negative $11.8 billion; net debt of $38.485 billion at 3.67x EBITDA; a trailing effective tax rate of 7.85% against guided 2026 cash taxes of approximately $1.5 billion, which takes the honest trailing multiple from 30.4x to 34.4x; a return on invested capital of 7.44%; a knowledge base that contains zero claims naming this company; and a base fair value of $575 that is 1.8% above spot and 2.9% below the street's $592.11, with 20.6% of downside against 17.9% of upside.

And the thing we cannot check, which is why this is a Hold and not something more definite in either direction. The organic revenue growth rate — the company's own primary non-GAAP measure, the number that separates a franchise from a roll-up — is disclosed only in tables that the filing extraction strips, and it appears in no vendor field. We know that Clario closed on 24 March 2026 for $8.87 billion of cash and that the 10-Q attributes $441 million of second-quarter clinical-research growth "primarily" to it, against total company revenue growth of $1.139 billion. We cannot resolve how much of the recent acceleration is earned. Paying a premium multiple for growth you cannot decompose is not defensible, and neither is refusing to own a genuinely important business because a table did not survive a text extraction. Hold is the correct expression of that state.

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

• A goodwill impairment against the filtration-and-separation reporting unit, which the 10-K has effectively pre-announced: $2.10 billion of goodwill on a $4.01 billion carrying value with a fair value "not substantially in excess" of it. An impairment there would confirm that the acquisition programme is overpaying.
• Organic revenue growth disclosed at or below 2% on 2026-10-28 or 2027-01, which would mean the four flat years are still running and the second quarter of 2026 was Clario.
• A fourth consecutive annual decline in Instruments revenue (2023, 2024 and 2025 are already down; the FY2026 figure is the test). Instruments lead the rest of the portfolio.
• A GAAP effective tax rate above 13% without offsetting operating improvement, which would remove roughly two turns from the honest earnings base.
• Another acquisition above $5 billion funded with debt at current leverage, which would take net debt through roughly 4x EBITDA on a company growing revenue at 5%.
• A break below the 200-day moving average of $531.15 on company-specific news rather than sector rotation.

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

• Price below approximately $500 — 18.2x the FY2027 consensus, 16.6x FY2028 — with no deterioration in the fundamental picture. This is the single most likely path to a Buy and it is only 11.5% below today's close, well inside the range this stock has traversed twice in six months. At $500 our base of $575 is a 15% premium and the asymmetry flips from 0.87:1 against to roughly 1.7:1 in favour.
• An organic revenue growth rate above 5% disclosed on 2026-10-28, with academic and government revenue growing rather than flat. That would settle the central question in the bull's favour and justify paying up.
• A confirmed turn in the instrument cycle — Instruments revenue growing year on year for the first time since 2022, which would signal that pharma capital spending, academic budgets and China have all normalised together.
• A new share-repurchase authorisation materially above $5 billion, which would remove the constraint on the buyback leg of the return bridge (only $1.00 billion remained at 31 July 2026).
• Any name-level knowledge-base coverage appearing at all — a single claim from a credible independent voice about this company rather than about thermodynamics. At present there is none, across six sweeps and 51,928 claims, and that alone is worth roughly one tier.

Where TMO fits in the Synthos Framework Portfolio. The life-sciences infrastructure sleeve, at 0% today with a 2-3% target on a fill near $500 or on a disclosed organic growth rate above 5%. This is the natural anchor holding for the diseases-of-aging thesis — more of that thesis runs through Thermo Fisher than through any single therapeutic name, and it does so without single-asset clinical risk. That is exactly why the entry price matters: a position you intend to hold through a decade of biological progress should not be opened at fair value with negative asymmetry. On the batch overlap question, note that Danaher, Agilent and IQVIA in the vendor peer set are the same trade in smaller form; Thermo Fisher at $208.8 billion is worth 98% of the three of them combined, so owning it is owning the sector, and owning it alongside them is concentration dressed as diversification. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $564.75.

Single biggest risk: that the growth is bought, not earned — and that we cannot check. Four calendar years of flat revenue bridged with $19.7 billion of acquisitions, $8.87 billion of it in the first six months of 2026, funded with $5.24 billion of new debt at 3.67x leverage, producing a second quarter that grew 10.5% with $441 million of one segment's growth attributed "primarily" to the acquisition that closed three months earlier. If the organic rate underneath is 2-3%, then the entire earnings-per-share growth path in consensus — 8.2% a year through FY2028 — depends on margin expansion and buyback at a company that has 80% consumed its repurchase authorisation and whose tax rate is disclosed as rising. At 20.6x forward and 34.4x on a cash-tax trailing basis there is very little in the valuation to absorb the discovery.

Most fragile assumption in the price: that net margin expands from 15.06% trailing to 20.6% by FY2028. Every other assumption in Section 6a is a volume assumption, and volume is observable quarter by quarter. This one is substantially a tax assumption, and it is made directly against the company's own disclosures — a GAAP effective-tax-rate guide that was already revised UP from 7-9% to 9-11% within six months; a cash-tax guide of approximately $1.5 billion against a trailing book charge of $597 million; and roughly $350 million a year of "discrete" deferred-tax benefits that have arrived in each of the last two years and are therefore not discrete. Operating leverage from the PPI programme is real and has delivered before — approximately $0.3 billion of restructuring in each of 2024 and 2025 for approximately $0.2 billion and $0.5 billion of annual savings, with $220 million more identified. It is being asked to deliver again, from a higher base, while the tax line works against it. If the tax rate lands at 12% rather than the ~9% consensus implies, FY2028 adjusted EPS is nearer $29 than $30.21, and at 19x that is $551 — below the price today.

Provenance & disclosures

• Traceability: ZERO tagged knowledge-base claims name Thermo Fisher Scientific (raw hits 79 across six sweeps; entity matches naming the company 0; text-only matches naming the company 0; discarded 73; retained as labelled end-market context 6; breadth 0, net conviction none, claim count 0). Searches run: (1) TMO, Thermo Fisher, Thermo — 0 entity, 33 text-only, all 33 discarded as homograph collisions on "thermodynamics", "thermoregulation", "thermometer" and "atmospheric"; (2) life sciences tools, bioprocessing, CDMO — 0 entity, 3 text-only, 3 retained as sector context; (3) Marc Casper, Danaher, DHR — 3 entity, 1 text-only, zero hits on the chief executive's name, 1 retained as serial-acquirer archetype context; (4) pharma capex, NIH funding, NIH — 9 entity and 27 text-only, all about the NIH as an agency rather than about Thermo Fisher, 2 retained as end-market context, 34 discarded; (5) Thermo Fisher Scientific, Fisher Scientific, PPD, Patheon, Applied Biosystems, Invitrogen — 0 entity and 0 text-only, completely empty; (6) Illumina, Agilent, Waters Corp, Revvity, Sartorius — 2 entity and 1 text-only, all about Illumina and Grail, all discarded. All six retained claims carry speaker: null and are attributed only to a channel — the weakest sourcing the four-lane attribution policy admits — and none carries a management speaker_role, so the half-weight management discount does not apply anywhere. All quotes are verbatim from the stored claim text. The knowledge base having no view on this company is reported as a finding and is a direct input to the Hold verdict, not padded into false coverage.
• Data as-of: income statement, balance sheet and cash flow for fiscal 2025 ended 2025-12-31 (10-K filed 2026-02-26); quarterly income statements through 2026-06-27; most recent filing-verified period 2026-06-27, from the 10-Q filed 2026-07-31 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873603 = 2026-08-04T20:00:03Z ($564.75, −1.62%; 50-DMA $509.56; 200-DMA $531.15; RSI 62.4; MACD +19.05) · knowledge-base claims 2026-08-04. Fiscal note: Thermo Fisher's fiscal year is the calendar year, but its quarters end on the nearest Saturday — the second quarter of 2026 ended 27 June 2026 and the first ended 28 March 2026. All figures come from the Synthos vendor data file for TMO or from the SEC filings in the TMO archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents and its gaps: 10-K filed 2026-02-26 (fiscal 2025); 10-Q filed 2026-05-01 (first quarter, ended 2026-03-28); 10-Q filed 2026-07-31 (second quarter, ended 2026-06-27); 8-K filed 2026-04-23 (first-quarter results); 8-K filed 2026-05-26 (annual meeting, Item 5.07); 8-K filed 2026-07-23 (second-quarter results). The 2026-04-23 and 2026-07-23 8-Ks are present as cover documents ONLY — Exhibit 99.1, the press release carrying the financials, was not extracted and neither file contains a single financial figure. Numeric tables are stripped from all extracted filing text, which in this file specifically removes: the four-segment revenue and margin tables (10-K and 10-Q Note 11), the ORGANIC REVENUE GROWTH tables that are the company's primary non-GAAP disclosure, the June-2026 balance sheet, the debt schedule, the Clario purchase-price allocation, the share-repurchase detail table, the contractual-obligations table and the annual-meeting vote tallies. Every one of those gaps is stated in the body where it bites rather than estimated around.
• Where the filings contradicted or corrected the vendor (detailed in Section 8): the vendor's FY2025 cash-flow statement reports incomeTaxesPaid: 0 while the 10-K states "Cash payments for income taxes were $1.78 billion during 2025" — the filing wins and we use $1.78 billion, which is load-bearing for the tax analysis; the vendor's balance sheet is 2025-12-31, seven months stale, and the 10-Q's cash-flow narrative supplies the $8.87B acquisition, $5.24B debt issuance, $1.41B repayment and $4.00B buyback that post-date it, from which we rebuild gross debt at approximately $43.2 billion; the vendor's market capitalisation is struck on the March 10-Q share count of 371,621,465 rather than the June count of 369,747,286, overstating it by $1.058 billion and the enterprise value by the same, both corrected; the vendor's FY2025 capitalLeaseObligations of $0 against $1.239B a year earlier makes the total-debt comparison invalid; the annual and quarterly income series contradict each other by $803 million on gross profit and $912 million on SG&A, so the apparent 368 basis point FY2025 gross-margin compression is rejected as an artefact; tech.max_dd_from_peak of −15.36% is arithmetically irreconcilable with a 52-week range implying a 31.45% peak-to-trough decline, and is rejected outright; quote.yearHigh/yearLow ($643.99/$435.27) and quote.priceAvg50/priceAvg200 ($504.81/$530.72) are set aside in favour of the tech block ($639.45/$438.34 and $509.56/$531.15); seg_prod is identified as the revenue-by-TYPE table rather than the segment table, with the only segment rows it contains dated FY2017 and earlier and using superseded names; seg_geo changes label basis from "North America" to "United States" between FY2023 and FY2024, invalidating that comparison. Where vendor and filing AGREED — worth recording: short-term obligations and current maturities of $3.53 billion at 2025-12-31 (10-K) against shortTermDebt of $3.533B; goodwill of $49.36 billion (10-K) against $49.362B; the entire FY2025 and FY2024 capital-allocation series (buybacks $3.00B and $4.00B, dividends $0.64B and $0.58B, acquisitions $4.04B and $3.13B, capital expenditure $1.52B and $1.40B) verbatim; and every revenue-by-type and geographic line summing to reported revenue within $1 million.
• Basis note: analyst estimates in est are ADJUSTED (non-GAAP) and the payload nowhere says so — verified against FY2025, where epsAvg of $22.745 sits 28.2% above reported GAAP diluted EPS of $17.74 and matches the sum of reported adjusted quarters to within a cent. All forward multiples in this dive are struck on adjusted EPS. The trailing multiple is shown on three bases: GAAP 30.4x, adjusted 23.7x, and our cash-tax restatement 34.4x (computed by applying the company's own guided ~$1.5 billion of 2026 cash taxes to trailing pre-tax income — our arithmetic, labelled as ours). The trailing GAAP-to-adjusted bridge is $5.24 per share, 22.0% of adjusted EPS, or approximately $1.96 billion after tax, consisting principally of acquisition-related intangible amortisation, restructuring and transaction costs.
• Estimate coverage: 20 analysts on FY2026 and FY2027 revenue and 17 on FY2028; 14, 16 and only 5 on the corresponding adjusted-EPS lines. FY2029 and FY2030 adjusted EPS rest on ONE analyst each and are excluded from every conclusion. The FY2028 EPS row is used only as a fair-value cross-check and is labelled thin at every appearance. est.ebitdaAvg and est.ebitAvg were tested against actuals and trip none of the four defect signatures in the data contract — no negative rows, no impossible ordering, no revenue-mirroring, no declining out-year revenue — but run 10-11% above GAAP actuals because they are adjusted-basis, and are therefore not used for valuation.
• Peer note: the vendor peer set contains three genuine comparables — Danaher ($136.9B), Agilent ($39.3B) and IQVIA ($37.6B) — alongside seven names that are Thermo Fisher's customers rather than its competitors: Abbott, Amgen, AstraZeneca, Intuitive Surgical, Merck, Novo Nordisk and Novartis. No peer multiples are supplied anywhere in the file, so no peer-multiple comparison is drawn. The 10-K declines to name competitors at all, stating only that "Because of the diversity of our products and services, we face many different types of competitors and competition."
• Fair-value caveat: the $449 / $575 / $666 anchors are multiples of the FY2027 consensus adjusted-EPS distribution — 17x the low of $26.388, 21x the mean of $27.404, and 23x the high of $28.958 — with the base independently cross-checked at 19x the FY2028 mean of $30.208, which gives $574, converging to within $1.50 from a different year. Stated arithmetic, not a discounted cash flow. The base is deliberately sensitivity-disclosed: 19x FY2027E gives $521 and 23x gives $630. The whole answer lives in a 19-23x band and no greater precision is claimed. The bear case assumes a de-rating and a miss together and lands within 2.4% of the actual 52-week low; the bull case requires only a 12% re-rating plus an earnings beat.
• Timing: second-quarter results were released 2026-07-23, twelve days before this dive, and the 10-Q was filed 2026-07-31, four days before it, so the entire first half of 2026 is reflected in the data. The next print is 2026-10-28, 85 days away — there is no near-term company-specific catalyst, which is an explicit input to the Hold verdict rather than an oversight. One unscheduled item is live: the sale of the microbiology business to Astorg for approximately $1.075 billion is expected to close in the third quarter of 2026. 2026-08-04 was a day on which the broad market and the semiconductor complex rose sharply and TMO fell 1.62%, closing 3.02% below its intraday high on 77% of average volume; there is NO company-specific news in this file for that date and we assign no reason to the move.
• 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.