Amphenol APH

Reference table

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

Amphenol describes itself in the profile as a global designer, manufacturer and distributor of electrical, electronic and fibre-optic connectors, interconnect systems, antennas, sensors and sensor-based products. Chief executive Richard Adam Norwitt. Approximately 170,000 employees worldwide as of 2025-12-31, of which approximately 15,000 are in the United States; less than 10% of the U.S. workforce is unionised. Approximately 6,400 employees support research, development and engineering. NYSE-listed, Class A Common Stock, $.001 par value, CIK 0000820313, CUSIP 032095101, IPO 1991-11-08.

The organisational structure is itself the disclosure. The 10-K states that Amphenol runs a "flat organizational structure" with more than 140 general managers running unique, independent businesses, each with income-statement and balance-sheet responsibility, grouped into three divisions that are the reportable segments. This is a federation, not a monolith, and it is why the acquisition programme works: acquired businesses are bolted on as new general-manager units rather than absorbed.

The three reportable segments (from the 10-K and the 2026-07-31 10-Q)

• Communications Solutions — connector and interconnect systems including high-speed, radio-frequency, power and fibre-optic products; coaxial, fibre-optic and high-speed cable; antennas. Serves IT datacom, mobile devices, industrial, communications networks, automotive, commercial aerospace and defense. This is where CommScope and Andrew went.
• Harsh Environment Solutions — ruggedised interconnect products, connectors, specialty cable, printed circuits and printed-circuit assemblies for industrial, defense, commercial aerospace, automotive, communications networks and IT datacom. This is where Trexon went.
• Interconnect and Sensor Systems — sensors, sensor-based systems, connectors and value-add interconnect systems for automotive, industrial, IT datacom, communications networks, defense and commercial aerospace.

The segment table, and the two things the vendor gets wrong

seg_prod is a genuine segment table — unlike several names in this batch, it is not a product table masquerading as one. The 10-K is explicit that no product breakdown exists: "As the Company is not organized by product or group of products, it is impracticable to disclose net sales by product or group of products." So the vendor has the right disclosure. It has two problems with it.

Problem one: the fiscal 2025 line is on a superseded taxonomy label. The vendor's third segment for fiscal 2025 is called "Interconnect Products And Assemblies" — which is the pre-2022 legacy label, the one that appears in the vendor's own fiscal 2010 through fiscal 2021 rows alongside "Cable Products And Solutions". For fiscal 2024, 2023 and 2022 the same vendor file correctly calls it "Interconnect and Sensor Systems", which is what both the 10-K and the 10-Q use. Read naively, the fiscal 2025 row implies a segment change that did not happen. We relabel to the filing's vocabulary throughout.

Problem two: the fiscal 2024 and fiscal 2025 lines do not sum to reported revenue, and the earlier years do. Here is the reconciliation, printed in full.

The 10-K explains the excess. Note 13 says the segment tables "summarize, by segment, total sales, intersegment sales and external net sales" and that "Intersegment net sales and operating expenses have been eliminated in the computation of consolidated net sales and operating income." The vendor appears to have taken the total-sales column for fiscal 2024 and 2025 and the external-net-sales column for fiscal 2022 and 2023. That is a basis change between adjacent years, and it means the fiscal 2024-to-fiscal 2025 segment growth rates from this table are computed on a consistent basis with each other but not with the earlier years.

We can recover the external figures independently, and they cross-check. The 10-K gives segment operating income and segment operating margin for fiscal 2025: Communications Solutions $3,746.6M at 31.1%, Harsh Environment $1,541.4M at 26.2%, Interconnect and Sensor Systems $1,005.1M at 19.5%. Dividing gives external net sales of approximately $12,046M, $5,883M and $5,154M — summing to $23,083M against reported $23,094.7M, a 0.05% residual from rounded margin percentages. The implied intersegment sales are therefore roughly $115M in Communications, $119M in Harsh Environment and $62M in Interconnect and Sensor Systems, totalling approximately $296M against the vendor's $283.9M excess. The reconciliation holds. We use the vendor's lines for shape and the 10-K's stated segment shares (Communications ~52%, Harsh Environment ~26%, Interconnect and Sensor Systems ~22% of fiscal 2025 net sales) for levels.

Segment operating margins — expanding everywhere, filing-verified

All three segments expanded in both periods, and the consolidated GAAP operating margin went from 20.7% to 29.5% in eighteen months. These figures come from the 10-K and 10-Q directly, not from the vendor. Note the ordering: Communications Solutions is both the largest and the highest-margin segment, and it is the one exposed to AI datacom.

End markets — the diversification is real, and the concentration is new

Read the second column, because it is the whole story. Of the $7.872B of fiscal 2025 revenue growth, $4.594B — 58% of it — came from one end market. Communications networks added another $1.374B and the 10-K says that was "driven primarily by contributions from acquisitions, in particular the acquisition of Andrew". Amphenol is a diversified company that grew in a concentrated way. Both halves of that sentence are true and both matter.

Customer concentration — the cleanest fact in the file

> "No single customer accounted for 10% or more of the Company's net sales during the years ended December 31, 2025, 2024 and 2023."

That is a materially better risk profile than anything else in this batch. KLA's largest customer was 19% of revenue in fiscal 2025. Amphenol sells to "thousands of original equipment manufacturers" plus electronic manufacturing services companies, original design manufacturers and service providers.

Geography (seg_geo, verified line-by-line against reported revenue)

The vendor's labels are mangled — "C [N]" is China and "U [S]" is the United States, an XBRL extraction artefact. The substance is clean: the lines sum exactly to reported net sales in every year we checked.

China's share fell from 23.0% to 15.9% in two years while China dollars still grew 27.4% — the denominator simply grew faster. The 10-K states net sales by geographic area are based on "the customer location to which the product is shipped", so this is a shipment geography, not an end-demand geography, and should not be read as an exposure map.

2. The single most important thing to understand about these numbers: the growth was mostly earned, not bought

A 51.7% revenue year at a serial acquirer is guilty until proven innocent. Here is the proof, and it clears.

The 10-K, verbatim:

> "Net sales were $23,094.7 for the year ended December 31, 2025 compared to $15,222.7 for the year ended December 31, 2024, representing an increase of 52% in U.S. dollars, 51% in constant currencies and 38% organically (excluding both currency and acquisition impacts...)"

The decomposition, computed from the filing's own figures. Fiscal 2024 base $15,222.7M; fiscal 2025 $23,094.7M; increase $7,872.0M (+51.71%).

Precision note, stated openly: the 38% is a rounded percentage, so each half-point of rounding is approximately $76M of organic revenue and the acquisition residual carries that uncertainty in the opposite direction. The conclusion — that roughly three-quarters of the increase was organic — survives the rounding comfortably.

The same exercise for the first half of 2026, from the 2026-07-31 10-Q, and it cross-checks beautifully. First-half 2026 net sales $16,378.2M against $10,461.3M — an increase of $5,916.9M, described as "an increase of 57% in U.S. dollars, 55% in constant currencies and 32% organically", with currency stated at $136.4M.

And the residual is independently confirmable. The 10-Q's internal-control section states that CommScope accounted for approximately 13% of consolidated net sales for the six months ended June 30, 2026 — which is $2,129.2M. Add the two smaller 2026 acquisitions and the Trexon stub and the $2.433B acquisition residual is exactly what it should be. This is the cleanest independent verification of an organic-growth decomposition we have been able to perform in this batch.

Now the second derivative, because it is the part the headline hides.

Organic growth has decelerated for two consecutive quarters — 38%, then 33%, then 30% — while the headline dollar growth stayed above 50% because acquisitions filled the gap. That is not a criticism; a 30% organic rate is spectacular. It is a warning that the headline and the underlying are diverging, and that the reported number will fall hard the moment the acquisition contribution annualises out, which for CommScope is 2027-01-09.

By segment, organic, from the filings:

Communications Solutions is the engine and it is the one decelerating. Harsh Environment and Interconnect and Sensor Systems are actually accelerating modestly against their fiscal 2025 rates, driven by defense, industrial and commercial aerospace. That mix shift is a stabiliser and it is under-discussed.

The point of this section is to establish that the growth is real — so that when the rest of this dive says the price is too high and the balance sheet has changed shape, that is not a disguised argument that the growth is fake. It is not fake. It is the best organic growth number in this batch by a wide margin. It is an argument about what has already been paid for it, and about what was borrowed to get the rest.

3. Growth and margin trajectory — and the three things flattering the margin

Annual, from the vendor income series with GAAP operating income restored from the filings:

Five-year revenue compound growth +21.9%; five-year diluted EPS compound growth +27.8%. Note the fiscal 2023 dip — revenue actually fell 0.5% in 2023. This company is not immune to the cycle; it just had a very small last one.

Quarterly, from the vendor income series with GAAP figures restored from the 10-Qs where they differ:

Six consecutive beats, and the last four were large — the June quarter delivered revenue of $8,758.1M against $8,258.2M expected (+6.1%) and adjusted EPS of $1.35 against $1.18 (+14.4%). That is not a rounding beat.

Gross margin has gone from 33.6% to 40.5% in eight quarters. That is a 690-basis-point expansion in a components business, and it is the most impressive operating fact in this file. Two adjustments make it slightly less impressive and both are disclosed:

• Q1 2026's 36.75% gross margin is depressed by $132.0M of non-cash amortisation of acquisition-related inventory step-up costs from CommScope, which the 10-Q says sit in cost of sales. Adding it back gives 38.48%.
• Q2 2025's 36.34% is depressed by $16.9M of the same thing from Andrew. Adding it back gives 36.64%.

On a step-up-adjusted basis the June-quarter year-on-year gross margin expansion is +387 basis points, not +417. Still very large.

Now the three things flattering the operating margin, in descending order of importance.

First, and disclosed by the company itself: an $80.0M tariff recovery. The 10-Q attributes part of the June-quarter margin improvement to "a net benefit of $80.0 related to the recovery of International Emergency Economic Powers Act ('IEEPA') tariffs". This is NOT excluded from the company's adjusted figures. Stripping it, June-quarter operating income is $2,504.6M and operating margin is 28.60%, not 29.51% — a 91-basis-point flatter. Any reader modelling forward margin off the 29.5% print is modelling a one-off.

Second, and running the other way: the acquisitions are dilutive to margin, by the company's own statement. The 10-Q says the margin expansion was "partially offset by the effect of acquisitions, which currently have higher Operating expenses as a percentage of net sales compared to the Company average." So the underlying business expanded margin by more than the reported figure shows, and the reported figure will improve further if CommScope is brought to the corporate average. We flag this as management framing and half-weight it accordingly — but it is a structural statement about mix, not a forecast, and it is consistent with the segment tables.

Third, and the one that will bite: the tax rate. From the filings:

The adjusted rate has risen 300 basis points in eighteen months, on the company's own reconciliation, and the company says why: the China tax matter plus "a continued shift in income to higher-tax jurisdictions". The 1H 2026 GAAP rate of 32.4% carries $169.0M of income-tax accruals, of which $130.0M relates to unfavourable determinations from Chinese tax authorities and $160.0M to a reassessment of prior-year rate assumptions. The $230.0M of tax payment notices was paid in full during the second quarter of 2026. Every 100 basis points of tax rate is roughly 1.35% of net income; 300 basis points is roughly 4%, permanently.

Research and development. The vendor reports $0 for fiscal 2025 and for every quarter except 2024-12-31, where it dumps the entire fiscal 2024 figure of $453.0M. This is a zeroed field, not a fact — the 10-K says research, development and engineering efforts were "supported by approximately 6,400 of our employees" at the end of 2025. The fiscal 2025 R&D figure is not available from any source in this file and no R&D-intensity conclusion appears in this dive. See Section 8.

4. Balance sheet and the enterprise-value rebuild — the seven-month problem

This is the single most consequential data-integrity issue in the file, and it is defect class 1 in its purest form.

bal_a[0] is dated 2025-12-31. It was filed with the 10-K on 2026-02-11. It is the most recent annual balance sheet in the payload, and it predates every material financial event of 2026:

The 8-K of 2026-05-12 is exactly the financing the defect-1 protocol exists to catch. It is a real underwritten public offering of €1.1B of senior notes, syndicated across ten banks, and it post-dates every balance sheet in the vendor annual file. We read it in full. Proceeds were used "to repay borrowings under its U.S. commercial paper program and its 364-day unsecured delayed draw term loan credit agreement" — so it refinanced rather than added, which is a materially better fact than the raw issuance suggests.

The 8-K of 2026-05-22 is NOT a financing. It is the annual-meeting Item 5.07 filing. It is nonetheless the most useful document in the archive for one purpose: it states that 1,229,430,709 shares of Class A Common Stock were outstanding as of the 2026-03-23 record date, which is our share-count check (Section 8, finding 12).

The balance sheet, old and new

Every fiscal 2025 line in the vendor annual file reconciles to the 10-K exactly — cash and short-term investments $11,434.2M, accounts receivable $4,717.1M, inventories $3,424.9M, PP&E $2,305.6M, goodwill $10,575.4M, intangibles $2,241.4M, accounts payable $2,661.9M. The fiscal 2025 balance sheet is not wrong. It is simply seven months out of date, and the seven months contain the largest transaction in the company's history.

Building net debt at 2026-06-30

The 10-Q's Note 4 debt table was stripped in extraction, so we do not have a single filing-stated total-debt figure. We build it two ways and report both.

Route one — the vendor's own trailing block, which IS current. km_ttm.enterpriseValueTTM is $224,906,415,100 against a market capitalisation of $210,824,515,100. The difference is $14,081.9M of net debt. We verified this block is struck on the June 2026 basis rather than the December 2025 basis: evToSalesTTM of 7.7523 implies trailing revenue of $29,011.6M, which is the exact sum of the four quarters to 2026-06-30 ($6,194.4 + $6,439.0 + $7,620.1 + $8,758.1). The trailing block is current even though the annual balance sheet is not.

Route two — roll the filing-verified December balance sheet forward on the filing-verified first-half cash flows.

> Total debt at 2025-12-31 $15,502.0M

> + net cash proceeds from borrowings, 1H 2026 $4,927.2M (10-Q)

> − repayments of Senior Notes and other long-term debt, 1H 2026 $1,535.7M (10-Q)

> = approximately $18,893.5M before currency translation

> + translation on approximately €2.7B of euro-denominated senior notes into a weaker dollar, plus lease additions — unquantified, positive

> ≈ $19.5B total debt, against filing-verified cash and short-term investments of $5,419.1M

> ≈ $14.1B net debt — which is what route one says.

The two routes agree. We use net debt of approximately $14,081.9M and label it as reconstructed rather than filing-stated, because the debt table was not extracted.

The rebuilt enterprise value:

> Market cap $210,824.5M (1,230,283,915 shares × $171.37)

> + net debt $14,081.9M

> = enterprise value $224,906.4M

Defect class 2 — tested, and the answer is split. The bal_a netDebt field DOES exhibit the classic omission: at 2025-12-31 it reports $4,371.4M, which is total debt of $15,502.0M less cash of $11,130.6M, ignoring $303.6M of short-term investments. The corrected December figure is $4,067.8M. But the km_ttm enterprise-value path does not exhibit it, because it is built on the June 2026 balance sheet where cash and short-term investments of $5,419.1M are the filing's own stated liquidity figure. We report both: the annual net-debt field is wrong by $303.6M; the enterprise value is right.

The debt stack, from the filings

Senior Notes outstanding, enumerated from the 10-K and the 2026-07-31 10-Q (this is not exhaustive — the notes table was stripped, so older U.S. series are not fully listed):

The November 2025 issuance alone was $7,500.0M across seven series, and it was raised specifically to fund CommScope. Interest expense went from $80.9M in Q2 2025 to $213.7M in Q2 2026 — an annualised run rate of approximately $855M against fiscal 2025's $367.8M. That is a permanent $490M reduction in pre-tax income relative to the 2025 base, roughly $0.28 of annual EPS at a 27% tax rate.

Facilities, and the commercial-paper question the brief flagged — VERIFIED CLEAN.

• $3,000.0M revolving credit facility, maturing March 2029. No outstanding borrowings at 2026-06-30 or 2025-12-31. In compliance with financial covenants.
• U.S. Commercial Paper Program, maximum $3,000.0M. No USCP Notes outstanding at 2026-06-30 or 2025-12-31, though the company "borrowed under the U.S. Commercial Paper Program throughout much of the first six months of 2026" and repaid all of it with cash on hand before quarter-end.
• Euro Commercial Paper Program, maximum $2,000.0M. No ECP Notes outstanding at 2026-06-30 or 2025-12-31; not drawn at all in the first half of 2026.
• Board authorisation caps commercial paper plus revolver borrowings at $3,000.0M combined. Programmes rated A-2 / P-2.
• Delayed-draw term loans: $934.1M (364-day) and $1,534.1M (three-year) outstanding at 2026-06-30. Reduced to approximately $400.0M and $1,534.1M by 2026-07-31 after $300.0M and $234.1M repayments on 2 and 27 July.

The brief's hypothesis — that Amphenol's large commercial-paper programme would be missing or misclassified in the vendor debt fields — is FALSE for this file, and the reason is that the balance was zero at every reporting date. That is a genuinely useful verified-clean finding: the programme is real, large and actively used intra-quarter, and it happens to be flat at each measurement point. A reader should know that Amphenol manages to zero at quarter ends, which is a disclosure choice worth being aware of, not a defect.

Note that the July term-loan repayments do not improve net debt. They were made "using cash on hand", so gross debt fell $534.1M and cash fell $534.1M. Net debt at 2026-07-31 is the same $14.1B it was at 2026-06-30.

The intangibles problem, stated plainly

Just over half of Amphenol's balance sheet is unamortised purchase price, and shareholders' equity of approximately $15.6B is entirely accounted for by goodwill and intangibles with $7.2B left over on the wrong side. The 10-K names the risk directly: "if the financial performance of the Company's businesses were to decline significantly, the Company could incur a material non-cash charge to its income statement for the impairment of goodwill and other intangible assets." Amphenol has never taken one — the 10-K states no goodwill impairment resulted from the July 2025 or July 2024 annual assessments, and none was recognised in 2025, 2024 or 2023. That is a 25-year clean record, and it is also a record that has never been tested by a hard downcycle at this level of goodwill.

Two consequences. First, priceToBookRatioTTM of 13.60x and returnOnEquityTTM of 37.2% are both distorted upward by an equity base that is mostly intangible; returnOnInvestedCapitalTTM of 15.1% is the honest measure and it is a respectable but not extraordinary number. Second, the vendor's own composite rating scores Amphenol 1 out of 5 on both price-to-book and debt-to-equity, dragging the overall grade to B / 3 — an artefact of the acquisition strategy, not a judgement about the business, and rejected as a lens.

Working capital, and it is deteriorating. Days sales outstanding 69 days at 2026-06-30 against 66 days at 2025-12-31. Inventory days 76 against 77. Payable days 69 against 60 — the company is stretching its suppliers, which offsets the receivables build. The 10-Q reports that first-half working capital increased $1,176.4M excluding acquisitions and translation, on receivables up $1,310.7M and inventories up $579.0M. Cash conversion cycle 96.5 days. This is what 55% revenue growth costs in cash.

Purchase commitments and off-balance-sheet items — hunted for, and modest. The 10-K discloses purchase commitments of $2,006.9M in 2026, $72.1M across 2027-2028 combined and $4.2M beyond 2028 — $2,083.2M in total, which is 9.0% of fiscal 2025 revenue. Leases "executed but which have not yet commenced" were "not material", and the 10-K states the company "does not generally enter into leases involving the construction or design of the underlying asset" and has no residual-value guarantees or external subleases. After several names in this batch carried enormous unrecorded lease and guarantee obligations, Amphenol's are genuinely small, and we report that as a clean finding. The one item worth naming: redeemable noncontrolling interests from recent acquisitions carry put options that could require Amphenol to purchase the holders' interests at a multiple-of-earnings redemption value. These sit in temporary equity; the amount was not extracted.

5. Cash flow and capital returns

The operating cash flow figures reconcile to the 10-K verbatim — "Net cash provided by operating activities ('Operating Cash Flow') was $5,374.7 in 2025, compared to $2,814.7 in 2024 and $2,528.7 in 2023." The fiscal 2025 buyback of $665.2M for 7.4 million shares also matches the 10-K exactly.

The most important row in that table is "Acquisitions". Across fiscal 2022 through the first half of 2026, Amphenol spent approximately $17.9B on acquisitions against approximately $14.9B of free cash flow over the same span. This company does not fund acquisitions from free cash flow. It funds them from debt, and the 10-Q says so: "The Company has funded all of its recent acquisitions entirely with a combination of cash on hand and net proceeds from its debt instruments, including the CommScope acquisition." That is a perfectly legitimate strategy with a long track record. It is also the reason net debt tripled, and it is the reason a reader should stop treating Amphenol as a self-funding compounder.

Free cash flow margin is falling as growth accelerates — 18.96% for fiscal 2025 against 16.25% trailing — because working capital is consuming cash faster than earnings are arriving. incomeQualityTTM of 1.132 (operating cash flow divided by net income) is still comfortably above 1.0, which is the reassuring counterpoint: the earnings are cash.

Capital expenditure has stepped up. $996.6M in fiscal 2025 (4.3% of revenue) against $665.4M and $372.8M in the prior two years. The 10-K says capex has "historically been in the range of 3% to 4% of net sales", so 4.3% is at the top of the historical band. First-half 2026 capex net of disposals was $642.0M. Note the 10-K reports gross PP&E purchases of $1,040.3M for fiscal 2025 against the vendor's capex line of $996.6M — the difference is disposals; we use the vendor's net figure and flag the discrepancy.

Dividend, and the same labelling trap seen elsewhere in this batch. The vendor reports lastDividend: 0.915 and dividendPerShareTTM: 0.915, yielding 0.534%. That $0.915 is the trailing-twelve-month total, not the last declared quarterly dividend. The filings give the sequence: the Board raised the quarterly rate from $0.11 to $0.165 on 2024-07-23, then from $0.165 to $0.25 on 2025-10-21, effective with dividends declared in the fourth quarter of 2025. Three quarters at $0.25 plus one at $0.165 is $0.915 — the vendor's figure, correctly computed but mislabelled. The forward run rate is $0.25 per quarter, $1.00 annualised — a 0.583% forward yield, not 0.534%. Payout ratio 19.7%.

Buyback — and this is not a buyback story. The 2024 Stock Repurchase Program authorised $2,000.0M over three years to 2027-04-28. Fiscal 2025 usage was $665.2M; first-half 2026 was $386.0M (2.8 million shares); a further $74.0M (0.5 million shares) between 1 and 28 July. Remaining authorisation as of 2026-07-29 is $411.1M — 0.195% of the market capitalisation.

And the share count is going UP, not down.

Diluted shares rose 2.0% in eighteen months and 4.8% since fiscal 2020, because option exercises outrun repurchases. Fiscal 2025 stock issuance from option exercises was $553.0M against $665.2M of buyback — a net $112.2M of shrinkage on a then-$150B-plus market cap. First-half 2026 option proceeds were $200.6M against $386.0M of buyback. Total shareholder yield is therefore approximately 0.58% dividend plus a gross buyback of roughly 0.37%, of which most is absorbed by dilution — call the net yield well under 1%. This is not an income story and it is not a shrinking-share-count story, and it should not be sold as either.

One further consequence of the option programme that matters for earnings quality. The 10-K discloses excess tax benefits from stock option exercises of $246.6M in fiscal 2025 and $142.6M in fiscal 2024. That $246.6M is 5.8% of fiscal 2025 net income, and it exists because the share price rose. It is excluded from the company's adjusted figures — correctly — but it is what allowed GAAP diluted EPS of $3.34 to exactly equal adjusted diluted EPS of $3.34 in fiscal 2025, with $148.8M of after-tax acquisition expense and a $100.0M China tax charge offset by the option benefit. A reader looking only at the GAAP number would conclude fiscal 2025 had no adjustments. It had several; they netted.

6. Valuation — priced in or room?

At $171.37 (market cap $210,824.5M, 1,230,283,915 shares, rebuilt enterprise value $224,906.4M):

Estimate coverage is adequate but not deep, and it thins fast. Eleven analysts on fiscal 2026 and fiscal 2027 revenue; only eight on fiscal 2026 EPS and ten on fiscal 2027 EPS; five on fiscal 2028 EPS; ONE on fiscal 2029 EPS. The fiscal 2029 row is discarded entirely and no conclusion in this dive touches it. The fiscal 2028 EPS row is used only as a cross-check, never as an anchor.

The estimate basis — verified arithmetically rather than assumed, which resolves defect class 7 cleanly. The fiscal 2026 epsAvg of $5.238 is struck on the company's ADJUSTED (non-GAAP) basis, and here is the proof: first-half 2026 adjusted diluted EPS was $2.42 (10-Q). The Q3 consensus is $1.42 (earnings calendar). $2.42 + $1.42 leaves $1.398 implied for the fourth quarter — a coherent sequence. Run the same test on GAAP: first-half GAAP diluted EPS was $2.10, which would require $3.14 in the second half, a step no analyst is modelling. epsAvg is adjusted. Every forward multiple in this dive is therefore struck on an adjusted basis, and the trailing multiple is shown on both (39.8x adjusted, 42.8x GAAP).

Per the data contract, est.ebitdaAvg and est.ebitAvg are treated as unreliable and are not used. In this file they are demonstrably broken: the fiscal 2029 row reports netIncomeAvg of $10,865.6M against ebitAvg of $10,766.9M — net income above operating income for a company with roughly $855M of annualised interest expense and a 27% tax rate. That is arithmetically impossible. The fiscal 2028 row implies a combined interest-and-tax burden of 3.3%. All EBIT, EBITDA and net-income estimate rows are rejected; epsAvg and revenueAvg carry the entire forward valuation.

epsAvg itself was verified against actuals and is sound, if slightly conservative. The fiscal 2025 row estimated $3.30749 against an actual of $3.34 on both GAAP and adjusted bases — 1.0% low. The fiscal 2024 row estimated $1.8393 against $1.92 GAAP and $1.89 adjusted — 2.7% to 4.2% low. No split corruption; no 10x errors; a mild and consistent conservatism.

Peer context — and there is essentially none available. The vendor peer set is Accenture, Applied Materials, Arista Networks, Arm, Intel, Intuit, KLA, Lam Research, Sony and Texas Instruments. Not one of these is an interconnect or connector company. There is no TE Connectivity, no Sensata, no Molex. The only arguably adjacent name is Arista Networks ($239.9B market cap, $190.55), which is AI networking equipment rather than components. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn — and the reader should know that Amphenol has been benchmarked here against semiconductor capital equipment and enterprise software, which is not informative.

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

At $171.37 — 32.7x fiscal 2026 consensus adjusted EPS and 26.9x fiscal 2027 — the price embeds the following falsifiable claims, each with a number and a date:

• Revenue reaches $35.121B in fiscal 2026 (+52.1%) and $41.196B in fiscal 2027 (+17.3%). (Consensus-derived; 11 analysts each.) First-half 2026 actual is $16.378B, so the second half must deliver $18.743B — up 14.4% sequentially over the first half and 45.5% year on year. The nearest check is the 2026-10-28 print: consensus revenue of $9.296B for the September quarter is +50.1% year on year and +6.1% sequentially on a quarter that itself grew 14.9% sequentially.
• Organic growth stops decelerating somewhere in the twenties. (Our number, derived from the company's disclosed organic series.) Organic was 38% for fiscal 2025, 33% in March, 30% in June. The CommScope contribution annualises out on 2027-01-09, at which point reported growth converges to organic growth plus whatever is bought next. If organic settles at 20%, fiscal 2027 revenue is roughly $39.5B before further acquisitions — 4% below consensus. If it settles at 15%, it is roughly $37.7B — 8.5% below. This is the most fragile assumption in the price, and it is fragile precisely because 36% of the company depends on one end market's capital-spending cycle.
• Net margin expands from 18.5% (fiscal 2025 actual) to roughly 19.4% in fiscal 2026 and 20.6% by fiscal 2028 — about 210 basis points. (Derived by us from consensus revenue and epsAvg × share count.) This must happen while the adjusted effective tax rate rises from 25.5% to 27.0% on the company's own reconciliation, while interest expense runs at roughly $855M annualised against fiscal 2025's $367.8M, and while acquisitions the company itself describes as carrying higher operating expenses than the corporate average are integrated. It also must happen without the $80.0M tariff recovery that added 91 basis points to the June-quarter operating margin. It is not an outrageous ask — 210 basis points is modest, and segment margins are genuinely expanding — but every tailwind in the June quarter has to be repeated and several were one-offs.
• Goodwill of $17.555B and intangibles of $5.289B are never written down. (Filing-confirmed balances; our assumption on the direction.) The 10-K warns explicitly that a significant decline in business performance could produce "a material non-cash charge". Amphenol has never taken one — but it has also never carried this much.
• The market keeps paying roughly 27x or better on the forward year at every date between now and then. (Our number.) At 22x fiscal 2027E the stock is $140; at 32x it is $204. The multiple, not the earnings, is the widest source of variance in the outcome — and the multiple is being asked to hold while EPS growth decelerates from +57% to +21.5% to +14.1%.

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

Expected return over the next twelve months decomposes as: EPS growth (+21.5%, from FY2026E $5.238 to FY2027E $6.366) + multiple drift (COMPRESSION, from 32.7x to 28x on the respective forward year, −14.4%) + shareholder yield (+0.58% dividend, with the gross buyback of roughly 0.37% largely absorbed by option dilution) ≈ +4% to +8%.

Our base case assumes multiple COMPRESSION, and we should be explicit that this is not a rolldown artefact. Today the market pays 32.7x on the current fiscal year's estimate. In twelve months, if it pays 28x on the then-current year's estimate, that is a genuine de-rating of about 14%, not the mechanical effect of the calendar advancing. We assume it because growth decelerates hard on the consensus path itself: revenue +52.1%, then +17.3%, then +13.0%; EPS +56.8%, then +21.5%, then +14.1%. A multiple appropriate to 57% earnings growth is not appropriate to 14%, and the market normally begins discounting the deceleration a year ahead of it. The additional reason specific to this name: the debt-funded balance sheet. A company at 0.3x net leverage deserves a different multiple from the same company at 1.5x with negative tangible book, and the market repriced from the former to the latter over the same six months in which the earnings were rising.

Almost all of the expected return is earnings growth, and the multiple is fighting it. If the multiple simply HELD at 32.7x on FY2027E the price would be $208 (+21.5%); if it compressed to 24x the price would be $153 (−10.7%). The entire spread between a good and a bad outcome sits in a multiple we do not control and have no edge on — and the knowledge base offers no help whatsoever in forming a view about it.

The bull case at $229 does not require multiple expansion at all — it holds today's 32.7x and applies it to the fiscal 2027 consensus high of $6.99. That is a deliberately conservative construction: the bull case is an earnings beat, not a re-rating, and Amphenol has beaten six consecutive quarters. The bear case at $125 is the mirror image: a de-rating to 21x applied to the fiscal 2027 consensus low of $5.958.

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

• We differ on the QUALITY of the growth, and this is the one place we are more positive than the headline. The market sees +51.7% revenue growth at a serial acquirer and discounts it. The 10-K says 38% of it was organic and the 10-Q's own internal-control disclosure independently confirms the acquisition residual to within rounding. That is a better number than a casual reader gets, and it is the reason this dive rates growth quality an 8 rather than a 5. Watchable number: organic growth disclosed in the 2026-10-28 release. A print above 28% would say the deceleration has stalled; below 24% would say it has not.
• We differ on the balance sheet, and materially, and it runs the other way. Every screen built off the annual file shows Amphenol with $4.371B of net debt and roughly 0.4x leverage. The correct figures at 2026-06-30 are approximately $14.082B and 1.51x, with tangible book equity of NEGATIVE $7.222B and goodwill plus intangibles at roughly 51% of total assets. That is not a rounding error on the risk assessment — it moves Amphenol from "unlevered compounder" to "moderately levered acquirer mid-integration", and it is why the risk score is a 6 rather than a 4.
• We differ on the earnings-quality picture, and it cuts both ways. Fiscal 2025 GAAP diluted EPS of $3.34 exactly equalled adjusted EPS of $3.34 — which looks like a company with no adjustments and is in fact a company where $148.8M of after-tax acquisition expense plus a $100.0M China tax charge were offset by $246.6M of excess tax benefits from option exercises. That option benefit is 5.8% of fiscal 2025 net income and it is a function of the share price. In a flat or down year it shrinks and GAAP earnings fall for a reason unrelated to operations. Meanwhile the June quarter's record 29.5% operating margin includes an $80.0M IEEPA tariff recovery that is NOT excluded from adjusted figures; ex that it is 28.6%.
• We differ on shareholder yield and we think it is widely misunderstood. Amphenol is frequently described as a dividend grower — the quarterly rate went $0.11 to $0.165 to $0.25 in two years, a 127% increase. But the diluted share count is RISING (1,263.6M to 1,289.4M in eighteen months), remaining buyback authorisation is $411.1M or 0.195% of market cap, and fiscal 2025 option issuance of $553.0M nearly matched the $665.2M buyback. Net shareholder yield is under 1% and the dividend is 0.58%. Nobody should own this for the return of capital.
• Positive surprise that would force a re-rate: September-quarter revenue materially above $9.296B with organic growth at or above 30% and operating margin above 29% excluding any further tariff recovery — which would validate both the volume leg and the margin leg of 6a at once. Or a completed CommScope purchase-price allocation that lands materially more value in identifiable intangibles than in goodwill, which would reduce the impairment overhang.
• Negative surprise that would break the thesis: organic growth below 24% in the October release; a September-quarter guide implying fiscal 2027 revenue below approximately $38B; any goodwill impairment; an adjusted effective tax rate above 28%; or evidence that rack-scale interconnect is shifting from copper to optics faster than the CommScope fibre position can absorb. Any of these hits a stock at 32.7x forward earnings, 7.75x sales and negative tangible book, sitting 2.8% below its 52-week high.

Synthos fair values

All three anchors are multiples of the fiscal 2027 consensus adjusted EPS distribution (mean $6.366, low $5.958, high $6.99, 10 analysts — the deepest forward EPS coverage in the file after the part-completed current year), cross-checked against fiscal 2026E of $5.238 and fiscal 2028E of $7.263.

• Bear ~$125 — 21x the fiscal 2027 consensus LOW of $5.958. Cross-check: 23.9x FY2026E; 31.3x trailing GAAP EPS of $4.00, so still not "cheap" in absolute terms; and 12.6% above the 52-week low of $107.93, which this stock traded at inside the last twelve months. The scenario: organic growth decelerates to the mid-teens as the AI datacom cohort digests, CommScope integration costs run longer than planned, the adjusted tax rate goes past 27%, and the multiple de-rates to reflect 1.5x leverage and negative tangible book. −27.1%.
• Base ~$178 — 28x the fiscal 2027 consensus MEAN of $6.366. Cross-check: 34.0x FY2026E; 24.5x FY2028E; rebuilt enterprise value of $233.2B on FY2027E revenue of $41.196B is 5.66x sales. Sensitivity, stated openly: 26x gives $165 — BELOW spot — and 30x gives $191, essentially the street's $192.33. The entire answer lives in a 26-30x band and we are not going to pretend to more precision than that. The scenario: revenue lands roughly on consensus, margins expand about as modelled but with the tax rate taking a slice, and the multiple compresses modestly as growth decelerates from 52% to 17% to 13%. +3.9%, plus roughly 0.6% of dividend.
• Bull ~$229 — today's 32.7x forward multiple, HELD, applied to the fiscal 2027 consensus HIGH of $6.99. Cross-check: 43.7x FY2026E. This bull case requires no re-rating, only an earnings beat — organic growth holds near 30%, CommScope reaches the corporate operating-margin average faster than guided, the AI datacom cycle extends, and Amphenol executes one more accretive deal. Six consecutive beats say this is not fanciful. +33.6%.

Base is 3.9% above spot; asymmetry roughly 1.24:1 to the upside (27.1% down, 33.6% up). That ratio is thin. What makes it a Watch rather than a Buy is that the base case sits essentially at the price, sits 7.5% below the street, depends on a consensus revenue path we cannot independently corroborate, receives zero support from the knowledge base, and is being offered on a day the stock rose 4.92% at 2.8% below its 52-week high.

7. Knowledge base — zero claims, and the search that proves it

Raw hits on the ticker: 636. Entity matches: 0. Text matches naming the company: 0. Used as name-level conviction: 0. Used as supporting colour: 0. Discarded for collision: 636.

This is the emptiest knowledge-base lane we have published against a company of this size, and the mechanism is worth stating precisely because it is the textbook homograph collision the attribution policy warns about.

The ticker search on APH returned 636 text-only matches and zero entity matches. Every one we inspected is a substring collision on the three letters "aph" inside ordinary English words — demographics, graph, paraphrase, sapphire, prophylactic, biography. The top-ranked hits were claims about global fertility rates, quantum cryptography, the loneliness epidemic, Venezuelan oil and Chinese robotics. Not one of the 636 is about Amphenol Corporation, and the entity-match count of zero is the tell: the tagger never once identified Amphenol as a subject.

The confirming test is decisive. A search on the full company name "Amphenol" returned ZERO matches of any kind — zero entity, zero text. In a corpus of 51,928 distilled claims, the word "Amphenol" does not appear.

The supplementary sweeps, and what they returned:

The "interconnect" sweep deserves its own paragraph, because it is the closest thing to relevance in the corpus and it still fails. Of 72 text matches, the substantial majority are attributed to a channel slug bearing the name of a large accelerator designer's chief executive, and they are product-architecture claims about that company's own roadmap — scale-up switch fabrics, co-packaged optics, copper versus optical scaling within its own rack designs. Under the 4-lane attribution policy these are simultaneously (a) management/promoter-class voice, (b) about a different company, and (c) product claims rather than equity claims. They carry no weight on Amphenol. The nearest genuinely adjacent claim is a neutral, unattributed observation dated 2026-11-28 about AI-infrastructure suppliers diverging by end-customer exposure, which names five tickers, none of them APH.

Attribution quality across everything that came back: uniformly the weakest form the policy admits. Of the sector-level AI-capex and datacentre-capex claims we scanned, essentially all carry speaker: null and are attributed only to a channel slug. A channel-level attribution tells you where a claim was said, not who said it or what their record is. We found no named speaker on any AI-capex or datacentre-capex claim in the sample, and no claim in any sweep carried a speaker_role of management for Amphenol — because there are no Amphenol claims at all, so the half-weight management discount has nothing to attach to.

Conclusion, stated as a finding rather than dressed up.

> The Synthos knowledge base has no view on Amphenol Corporation. Not a weak view, not a sector-proxy view — no view.

It has a large, unattributed, internally contradictory body of opinion about whether datacentre capital expenditure is ahead of demand or behind it. That body of opinion is genuinely relevant to Amphenol's fate — 36% of the company rides on it — but it is not a claim about this equity, it carries no named speakers, and it points in both directions simultaneously. Nothing in it justifies a conviction premium in the fair value, nothing in it argues for owning Amphenol rather than any other AI-infrastructure supplier, and the absence is itself a verdict input. Conviction rating None, breadth 0, claim count 0, and this alone is worth roughly one tier against the verdict.

A note on what this means for the collision warning in general. The brief anticipated that AI-capex claims would be basket calls listing many tickers, and that those are sector colour rather than name-level conviction. The reality here was worse than anticipated: there are not even basket calls naming APH. A reader should treat any screen or tool that reports "636 knowledge-base mentions of APH" as reporting a substring frequency, not a coverage statistic.

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

Sixteen findings. Listed rather than silently corrected. Amphenol's file has one very large defect — a stale annual balance sheet across the biggest transaction in company history — sitting alongside an unusually clean set of everything else. Both facts are reported.

1. bal_a[0] is SEVEN MONTHS STALE and predates the largest acquisition in company history — CONFIRMED, defect class 1, and it is the headline finding. The most recent annual balance sheet in the payload is dated 2025-12-31. Between that date and this dive Amphenol closed the CommScope Connectivity and Cable Solutions acquisition for approximately $10,684.0M net of cash acquired (2026-01-09), drew $3,068.2M of delayed-draw term loans, issued €1,600.0M of euro senior notes across three transactions, and repaid approximately $1,535.7M. The vendor reports net debt of $4,371.4M; the 2026-07-31 10-Q and the vendor's own trailing block show approximately $14,081.9M at 2026-06-30; we use $14,081.9M. Goodwill went from $10,575.4M to $17,554.7M and intangibles from $2,241.4M to $5,288.9M over the same span. Any screen or ratio built off the annual balance sheet understates Amphenol's leverage by roughly 3.2x.

2. The 8-K financing sweep — CONDUCTED, and it found a real note offering that post-dates every balance sheet in the annual file. The 8-K of 2026-05-12 records the issuance of €600,000,000 of 3.375% Senior Notes due 2029 and €500,000,000 of 3.875% Senior Notes due 2034, sold in an underwritten public offering across ten banks, with net proceeds of approximately €1,093.1M used "to repay borrowings under its U.S. commercial paper program and its 364-day unsecured delayed draw term loan credit agreement." This is refinancing rather than net new leverage, which is a materially better fact than the raw issuance size. The 8-K of 2026-05-22 is NOT a financing — it is the annual-meeting Item 5.07 report, and it is used in this dive only for the share count.

3. bal_a netDebt omits short-term investments — CONFIRMED, defect class 2, at the annual level only. At 2025-12-31 the field reports $4,371.4M, which is total debt of $15,502.0M less cash of $11,130.6M, ignoring $303.6M of short-term investments. The 10-K's own liquidity statement counts them: "the Company had cash, cash equivalents and short-term investments of $11,434.2." Corrected December net debt is $4,067.8M. The km_ttm enterprise-value path does NOT exhibit the defect, because it is built on the June 2026 balance sheet where the filing's stated liquidity of $5,419.1M already includes short-term investments. We report both.

4. km_ttm and ratios_ttm are CURRENT even though bal_a is not — verified, and this is why the enterprise value is usable. evToSalesTTM of 7.7523 against an enterprise value of $224,906.4M implies trailing revenue of $29,011.6M, which is the exact sum of the four quarters to 2026-06-30. netIncomePerShareTTM of $4.1910 × the implied 1,229.7M shares gives $5,153.7M, which is the exact sum of the four quarterly net-income figures. The trailing block is struck on June 2026 data. The payload mixes a stale annual file with a current trailing file, and a reader who does not check will build ratios from two different balance-sheet dates.

5. Trailing EBITDA — recomputed and it agrees with the vendor to within 0.4%. We compute operating income of $7,888.8M (sum of the four quarters) plus depreciation and amortisation of $1,448.3M = $9,337.1M (32.2% margin). The vendor's ebitdaMarginTTM of 32.294% implies $9,369.2M. We use $9,337.1M and note the difference is immaterial. Unlike several names in this batch, the vendor's EBITDA here is not internally contradictory — a verified-clean check worth recording.

6. seg_prod does not sum to reported revenue in fiscal 2024 or fiscal 2025, and DOES in fiscal 2022 and fiscal 2023 — a basis change, defect class 8. Fiscal 2025 sums to $23,378.6M against reported $23,094.7M (+$283.9M, +1.23%); fiscal 2024 to $15,394.8M against $15,222.7M (+$172.1M, +1.13%); fiscal 2023 and fiscal 2022 tie exactly. The 10-K's Note 13 states the segment tables carry "total sales, intersegment sales and external net sales" with intersegment eliminated in consolidation, so the vendor appears to have taken the total-sales column for the recent years and the external column for the earlier ones. We independently recovered the external figures from the 10-K's segment operating income and margin percentages (Communications ≈$12,046M, Harsh Environment ≈$5,883M, Interconnect and Sensor Systems ≈$5,154M, summing to $23,083M against reported $23,094.7M) and the implied intersegment sales of approximately $296M reconcile the $283.9M excess. Reconciliation printed in full in Section 1. The lines are used for shape; the 10-K's stated segment shares are used for levels.

7. seg_prod labels the third segment on a SUPERSEDED taxonomy in fiscal 2025 only — defect class 8. Fiscal 2025 reads "Interconnect Products And Assemblies", which is the pre-2022 legacy label that also appears in the same file's fiscal 2010-2021 rows. Fiscal 2024, 2023 and 2022 correctly read "Interconnect and Sensor Systems", which is what the 10-K and the 10-Q both use. Read naively the fiscal 2025 row implies a segment reorganisation that did not occur. Relabelled to the filing's vocabulary throughout this dive. The current three-segment taxonomy — Harsh Environment Solutions, Communications Solutions, Interconnect and Sensor Systems — is confirmed in both the 10-K and the 2026-07-31 10-Q.

8. seg_geo labels are mangled but the substance is exact — verified clean. The vendor emits "C [N]" for China and "U [S]" for the United States, an XBRL extraction artefact. The lines sum EXACTLY to reported net sales in fiscal 2025 ($23,094.7M), fiscal 2024 ($15,222.7M) and fiscal 2023 ($12,554.7M). Labels corrected in presentation; figures used as supplied.

9. Vendor operating income is CONSISTENT year to year but INCONSISTENT quarter to quarter — defect class 4, half confirmed. The vendor's annual operating income exceeds the 10-K's GAAP figure by exactly the separately-presented acquisition-related expense line in every year: fiscal 2023 $2,594.2M vs GAAP $2,559.6M (+$34.6M), fiscal 2024 $3,284.3M vs $3,156.9M (+$127.4M), fiscal 2025 $5,972.0M vs $5,868.6M (+$103.4M). Each add-back is exactly the amount the 10-K states. Annual margins are therefore comparable across years — a verified-clean result on the most dangerous defect class in this batch. But the quarters are not. Q2 2025 vendor operating income of $1,430.8M exceeds the 10-Q's $1,418.8M by exactly the $12.0M of separately-presented acquisition expense, and first-half 2025 by exactly $56.0M — while Q2 2026's $2,584.6M matches GAAP exactly despite $23.5M of such expense existing, and first-half 2026's $4,416.4M matches GAAP exactly despite $140.4M existing. The distortion is small (Q2 operating margin expansion reads +418bp on vendor data against +440bp on GAAP) but the basis genuinely changed. All operating margins in this dive are computed on the filings' GAAP figures where available.

10. researchAndDevelopmentExpenses is $0 for fiscal 2025 and for eight of nine quarters — REJECTED, defect class 11. The field reads $453.0M for fiscal 2024, $342.2M for 2023, $323.6M for 2022 — and $0 for fiscal 2025, with the entire fiscal 2024 figure dumped into the 2024-12-31 quarter and every other quarter zero. researchAndDevelopementToRevenueTTM is likewise 0. Amphenol did not stop doing research: the 10-K states R&D and engineering was supported by approximately 6,400 employees at the end of 2025. The fiscal 2025 figure appears to have been folded into selling, general and administrative expenses, which jumped from $1,402.4M to $2,545.7M. No R&D-intensity conclusion appears anywhere in this dive.

11. stockBasedCompensation is $0 in the fiscal 2025 cash-flow statement — REJECTED, defect class 11. The field reads $89.5M (2022), $99.0M (2023), $109.5M (2024) and $0 (2025). The vendor's own stockBasedCompensationToRevenueTTM of 0.5415% against trailing revenue of $29,011.6M implies $157.1M — which we use. The zero is a missing value, not a fact.

12. Share count — CHECKED and CLEAN, defect class 6, with one disclosure worth flagging. Market capitalisation of $210,824,515,100 divided by $171.37 implies 1,230,283,915 shares. The 8-K of 2026-05-22 states 1,229,430,709 shares of Class A Common Stock outstanding as of the 2026-03-23 record date — agreement to 0.07%, comfortably explained by four months of option exercises net of 3.3 million shares repurchased. No share-class error, no partial-unit error, no split. The disclosure worth flagging is the gap to diluted: Q2 2026 diluted weighted-average shares are 1,289.4M — 59.1M, or 4.8%, ABOVE the outstanding count. That is a large in-the-money option overhang, and its scale is visible in the insider table: the chief executive exercised options struck at $22.3725 against a $171 stock. Not a defect; a real dilution fact that a market-cap-based screen will miss.

13. Vendor Q2 2026 diluted EPS of $1.38 contradicts the 10-Q's $1.37 — the filing wins. The 10-Q states "Net income attributable to Amphenol Corporation and ... Diluted EPS were $1,769.2 and $1.37, respectively, for the second quarter of 2026." $1,769.2M ÷ 1,289.4M = $1.372. We use $1.37 and every trailing EPS figure in this dive is built on the filing's number. The same one-cent discrepancy appears in Q1 2026 (vendor $0.72, filing-implied $0.73).

14. est EBIT, EBITDA and net-income rows are arithmetically impossible — REJECTED in full, defect class 9. The fiscal 2029 row reports netIncomeAvg of $10,865.6M against ebitAvg of $10,766.9M — net income above operating income for a company carrying roughly $855M of annualised interest expense at a 27% tax rate. The fiscal 2028 row implies a combined interest-and-tax burden of 3.3%. Every EBIT, EBITDA and net-income estimate row is discarded; epsAvg and revenueAvg carry the forward valuation. epsAvg itself was verified against actuals and is sound: fiscal 2025 estimated $3.30749 against an actual $3.34 (1.0% low), fiscal 2024 estimated $1.8393 against $1.92 GAAP (4.2% low). No split corruption, no 10x errors.

15. Analyst coverage thins fast and the far years are discarded. Eleven analysts on fiscal 2026 and fiscal 2027 revenue; eight on fiscal 2026 EPS, ten on fiscal 2027 EPS, five on fiscal 2028 EPS and ONE on fiscal 2029 EPS. The fiscal 2029 row is excluded from every conclusion. The fiscal 2028 row is used only as a cross-check. All fair values anchor on fiscal 2027, the deepest forward EPS year.

16. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — we use tech and say so, defect class 12. The quote block reports a 52-week high of $178.52 and low of $104.71; the computed technical block reports $176.32 and $107.93. The discrepancies are 1.25% and 3.08%. We use the tech figures throughout, because they are computed from the same six-year close series that produces the moving averages, drawdown and relative-return figures. Note that tech.max_dd_from_peak equals tech.pct_from_hi at −2.81% — that is not a defect, it is what a stock sitting at its high looks like.

Other defect classes tested and found CLEAN — each stated because a clean check is a finding.

• Defect class 3 (one-off items capitalised as run-rate) — tested, and the answer is nuanced rather than clean. Fiscal 2025 GAAP diluted EPS of $3.34 exactly equals adjusted diluted EPS of $3.34, which superficially says "no adjustments". In fact $148.8M of after-tax acquisition-related expense and a $100.0M China tax charge were offset by $246.6M of excess tax benefits from stock-option exercises. That option benefit is 5.8% of net income and depends on the share price. Separately, the June-quarter operating margin of 29.5% includes an $80.0M IEEPA tariff recovery which is NOT excluded from the company's adjusted figures; ex that, operating margin is 28.60%. Both are disclosed by the company and both are flagged in this dive rather than smoothed. Our clean trailing adjusted EPS is $4.31 (39.8x); ex the tariff recovery it is approximately $4.27 (40.1x).
• Defect class 5 (currency mixing) — CLEAN. inc_a[0].reportedCurrency is USD; profile.currency is USD; the quote is in USD on the NYSE. No mixing. Worth noting nonetheless that the company now carries approximately €2.7B of euro-denominated senior notes, and translation added $84.6M to fiscal 2025 sales and $136.4M to first-half 2026 sales — real but disclosed, and the multiples are unaffected.
• Defect class 10 (off-balance-sheet obligations) — hunted and CLEAN. Total purchase commitments are $2,083.2M ($2,006.9M in 2026, $72.1M across 2027-2028, $4.2M beyond) — 9.0% of fiscal 2025 revenue and modest for a manufacturer of this size. Leases executed but not yet commenced were "not material"; the 10-K states the company does not generally enter construction-type leases, has no residual-value guarantees and does not sublease to external parties. After several names in this batch carried enormous unrecorded lease commitments, this is a clean result. The one item on the watch list is redeemable noncontrolling interests from recent acquisitions carrying put options at multiple-of-earnings redemption values; these sit in temporary equity and the balance was not extracted.
• Defect class 13 (non-equity tripwire) — CLEAN. APH is Class A Common Stock, $.001 par value, NYSE-listed, ISIN US0320951017, CUSIP 032095101. The price of $171.37 is not par-like; beta is 1.243, far above the sub-0.3 typical of a fixed-income-like instrument; the dividend is variable and has been raised from $0.11 to $0.25 per quarter in two years; the 52-week band of $107.93 to $176.32 is a 63% range; volume was 6.83M shares (approximately $1.17B of turnover). This is common equity. Note the 8-Ks reference both a CommonClassAMember and a CommonClassBMember in their XBRL headers; the 2026-05-22 8-K reports only Class A shares outstanding (1,229,430,709) and the annual-meeting quorum was struck solely on Class A, so no dual-class voting complication is present in the file.
• Vendor-to-filing agreements worth recording. Fiscal 2025 revenue $23,094.7M, fiscal 2024 $15,222.7M and fiscal 2023 $12,554.7M all match the 10-K exactly. Operating cash flow of $5,374.7M, $2,814.7M and $2,528.7M match verbatim. The fiscal 2025 buyback of $665.2M matches. Cash and short-term investments of $11,434.2M matches. Goodwill $10,575.4M, intangibles $2,241.4M, receivables $4,717.1M, inventories $3,424.9M, PP&E $2,305.6M and payables $2,661.9M all match. Vendor interest expense of $207.9M (Q1) plus $213.7M (Q2) equals the 10-Q's stated first-half figure of $421.6M exactly. The fiscal 2025 balance sheet is accurate. It is simply seven months old.
• One item flagged as suspicious rather than rejected. inc_q depreciation and amortisation reads exactly $485.8M in BOTH the 2026-03-31 and 2026-06-30 quarters — identical to the decimal. That is more likely a first-half total split evenly than two coincidentally identical quarters. It affects our EBITDA build by an unknown amount within the half; the annual and trailing totals are unaffected. Flagged, used, and noted.
• capitalLeaseObligations drops to $0 at 2025-12-31 from $285.6M (2024) and $212.1M (2023) — the year-on-year total-debt comparison is not meaningful. Amphenol has not eliminated its leases; the 10-K discusses operating lease right-of-use assets increasing in 2025 from new and renewed agreements. The fiscal 2024-to-fiscal 2025 totalDebt comparison in the vendor file is broken by this alone, before the staleness problem is even considered.

Filing-archive gap, stated verbatim as a finding. The 8-K of 2026-07-29, which announced the June-quarter and first-half results, is present in the archive as the cover document ONLY. Exhibit 99.1 — the press release carrying the financials — was not extracted, and the file contains no numbers whatsoever. Unusually for this batch, nothing is lost: the 10-Q filed 2026-07-31, two days later, covers the identical period and IS in the archive with full MD&A prose. This is the only name in the batch where the freshest balance sheet is filing-verified rather than vendor-supplied. Separately, numeric tables are stripped from all extracted filing text — the 10-Q's Note 4 debt table, the segment tables, the non-GAAP reconciliation tables and the contractual-obligations table are all present as headings with the numbers removed. Every figure in this dive sourced to a filing comes from MD&A or footnote PROSE, where the numbers survive extraction. That is why total debt at 2026-06-30 is reconstructed rather than quoted.

9. Technicals

• Price $171.37. −2.81% from the 52-week high of $176.32, and +58.78% above the 52-week low of $107.93. Position within the annual range: the 93rd percentile. This stock is at its high.
• Moving averages, both rising, both below: +10.09% above a 50-day average of $155.66; +20.21% above a 200-day average of $142.56. The 50-day sits above the 200-day. This is a clean, intact uptrend with no technical damage anywhere in the structure.
• RSI 62.95 — firm but not overbought (the conventional threshold is 70). MACD +1.42 — positive.
• Maximum drawdown from peak over the trailing year: −2.81%. The stock has not had a meaningful correction in twelve months.
• Relative performance: 3-month +21.51% vs SPY +7.59% and QQQ +7.67%; 6-month +18.24% vs SPY +11.09% and QQQ +15.71%; 12-month +64.29% vs SPY +24.26% and QQQ +30.80%. A 40-point outperformance against the S&P over twelve months, achieved without a drawdown.
• Beta 1.243. Average daily volume 9.62M shares.
• Sentiment: 0 strong buy, 16 buy, 13 hold, 1 sell across 30 analysts. Consensus target $192.33 (+12.2%), median $198.50 (+15.8%), high $215 (+25.5%), low $165 — 3.7% BELOW spot. A well-dispersed set, but note that 14 of 30 analysts are not buyers on a stock that has just gone up 64%.

Today's move and what it does to the entry

APH closed 2026-08-04 at $171.37, up 4.92% or $8.03 from a previous close of $163.34. It opened at $169.86 — a 3.99% gap — traded a $165.715 to $172.88 range, and closed 0.87% below the day's high on 6,831,544 shares.

Two things about that move.

First, it was sector beta and Amphenol was at the back of the pack. The whole semiconductor and AI-hardware complex ran: MRVL +12.81%, INTC +10.92%, LRCX +7.85%, MU +7.62%, AMD +7.00%, KLAC +6.95%, AMAT +5.48%, APH +4.92%, ASML +4.22%. Amphenol's move was the second smallest of that group, which is what you would expect from a components manufacturer rather than a semiconductor company — its beta is 1.243 and its business is not chip fabrication. No company-specific news for 2026-08-04 appears anywhere in this file. The last company event was the earnings release six days earlier on 2026-07-29.

Second, and more useful: it happened on BELOW-average volume. 6.83M shares against a 9.62M average is 29% below normal. A 4.92% advance on 71% of normal volume is a lower-conviction move than the percentage suggests — it is consistent with a broad, index-driven bid rather than concentrated accumulation in this name.

The honest read, and it is the opposite of KLA's in this batch. Where KLA gapped 7% from 35% below its high into overhead resistance, Amphenol gapped 4.92% from 2.8% below its high, at the 93rd percentile of its annual range, 20% above a rising 200-day average, with RSI at 63 and no drawdown worth the name in twelve months. In pure chart terms this is a breakout attempt within an established uptrend, and the technical picture is genuinely strong. It is also, unambiguously, paying up.

So here the gap IS part of the reason to decline the entry — but only part. The larger part is that our base fair value of $178 is 3.9% above the closing price and 7.5% below the street's $192.33, and that the price sits at 32.7x fiscal 2026 consensus, 7.75x sales and 44.7x free cash flow for a business whose net debt tripled six months ago and whose organic growth has decelerated in two consecutive quarters.

What we are therefore telling the reader, plainly: we are not asking you to chase a 4.92% day into a 52-week high, and we are not asking you to buy at all today. The setups that would change this are the falsifiers:

• A price near $150 — roughly 23.6x fiscal 2027E and 28.6x fiscal 2026E, and roughly 4% below the 50-day average — which would put our base $178 at an 18.7% premium and make the risk/reward genuinely asymmetric. That is 12.5% below today's close and well inside the range this stock has traversed in a quarter.
• Or the 2026-10-28 print landing above $9.296B of revenue with disclosed organic growth at or above 28% and operating margin above 29% excluding any further tariff recovery, which would validate both legs of 6a at once. At that point paying up is defensible, because the uncertainty being paid for has been resolved.

Buying today means paying a 4.92% one-day advance, at a 52-week high, for a 3.9% base-case return, with no company-specific catalyst for 85 days, against zero knowledge-base support. That is not a trade with an edge in it.

10. Insiders — one very large chief-executive sale, and it needs reading carefully

Eight transactions in the file, and unlike most names in this batch they are not all the same thing.

The arithmetic. Norwitt exercised 686,104 options at $22.3725 — a cost of $15,349,341 — and sold all 686,104 shares the same day at $161.5748 and $162.2712, for gross proceeds of $110,987,343. Net of the strike, approximately $95.6M, on 2026-07-31 — two days after the June-quarter earnings release.

How to read this, carefully, because it is genuinely the most significant insider event in this batch.

What it is NOT. It is not a straightforward liquidation of a long-held position. Every share sold was created that day by exercising a deep-in-the-money option struck at $22.3725 — roughly 87% below the current price, which places the grant a very long time ago. Exercise-and-immediately-sell is the standard way an executive monetises an option that is approaching expiry, and it carries far less signal than a sale of already-owned stock. The indirect stock-option holding went to zero, which is consistent with a final tranche being cleared.

What it IS. It is still $111M of stock hitting the market from the chief executive, two days after a print, with no F-InKind tax-withholding component anywhere in the file. And there is a specific disclosure gap: the 10-Q for the June quarter states that "no director or officer of the Company adopted or terminated a 'Rule 10b5-1 trading arrangement' or 'non-Rule 10b5-1 trading arrangement'" during the three months ended 2026-06-30. The sale occurred on 2026-07-31, in the third quarter. We therefore cannot determine from this file whether the sale was executed under a pre-existing 10b5-1 plan, and we do not assume it was. That determination would require the Form 4 footnotes, which are not in the payload.

What the residual holdings say. After the transaction Norwitt retains 1,927,507 shares directly and 864,177 indirectly — 2,791,684 shares, worth $478.4M at the closing price — plus 399,788 remaining direct stock options. That is a very large continuing stake and it materially softens the reading. He sold roughly 20% of his post-exercise economic position and kept 80%.

The two May grants are awards, not purchases. Luc Walter and Peter Straub, presidents of two of the three divisions, each received 108,809 stock options at a $132.06 strike on 2026-05-22. These are compensation grants; they signal a retention decision by the board, not a market view by the recipients. The $132.06 strike against a $171.37 price means they are already 29.8% in the money after ten weeks.

Net reading: neutral-to-mildly-negative, and materially less benign than a table of pure tax withholding. There are ZERO open-market purchases in the file. The single largest transaction is a nine-figure chief-executive sale immediately after a strong print, of uncertain plan status, at a price 6% below where the stock closed today. We do not treat it as a bearish signal — the option economics explain it and the retained stake is enormous — but we do not treat it as a neutral event either, and we note the option overhang it reveals: 4.8% of the diluted share count sits in in-the-money options struck at a fraction of the current price.

11. Verdict, kill-criteria and flip conditions

Watch.

This is the best-growing large business in this batch and the one where our fair value sits closest to the price.

What is genuinely excellent, and none of it is in dispute: 38% organic revenue growth in fiscal 2025 and 30% in the June quarter, verified against the filings and independently cross-checked through the 10-Q's own internal-control disclosure; consolidated GAAP operating margin from 20.7% to 29.5% in eighteen months with all three segments expanding in both periods; gross margin from 33.6% to 40.5% in eight quarters; backlog from $6.1B to $8.9B; no single customer above 10% of net sales in any of the last three years; seven end markets none larger than 36%; a decentralised model of more than 140 general managers that has absorbed dozens of acquisitions; six consecutive earnings beats, the last four of them large; $4.71B of trailing free cash flow; a $3.0B undrawn revolver; zero commercial paper at every balance-sheet date; and total purchase commitments of only $2.08B.

What we are declining to pay for: 32.7x fiscal 2026 consensus, 26.9x fiscal 2027, 7.75x sales, 44.7x free cash flow, 24.1x rebuilt trailing EBITDA. A base fair value of $178 that is 3.9% above spot and 7.5% below the street's $192.33. A balance sheet where net debt tripled to approximately $14.1B in six months, where goodwill plus intangibles are roughly 51% of total assets, and where tangible book equity is NEGATIVE $7.222B. An adjusted effective tax rate the company itself discloses rising from 24.0% to 27.0% in eighteen months. A diluted share count that is going up, not down, with only $411.1M of buyback authorisation left. Organic growth that has decelerated in two consecutive quarters — 38%, 33%, 30%. A knowledge base with literally zero claims about this company. And an entry price 2.8% below the 52-week high, at the 93rd percentile of the annual range, including a 4.92% one-day advance on below-average volume.

The distinction that matters. We are not saying Amphenol is a bad business or that the CommScope acquisition will fail. We are saying we have no independent basis for believing the forward path more than the market does, no valuation cushion if we are wrong, no asset floor beneath a business that is half goodwill, and no company-specific catalyst for 85 days. In the Synthos frame, a name where our fair value sits below consensus and the conviction lane is empty rather than merely thin is a Watch by construction, not by hesitation. Manufacturing a differentiated view here would be dishonest.

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

• Disclosed organic net sales growth below 20% in any quarter, or a September-quarter guide implying fiscal 2027 revenue below approximately $38B. That is the growth engine stalling, and there is nothing under a 32.7x multiple to catch it.
• Any goodwill or intangible impairment charge. With $22.8B of goodwill and intangibles against $15.6B of book equity, a 30% write-down eliminates the equity. The 10-K names this risk explicitly.
• Adjusted effective tax rate above 28%, or a further material accrual on the China tax matter beyond the $230.0M paid and $160.0M reserved.
• Operating margin below 26% in any quarter, which would say the acquisition dilution the company describes is structural rather than transitional.
• Net debt rising above roughly $18B — for example on another large debt-funded acquisition before CommScope is integrated — or a credit-rating downgrade, which the 10-K flags as a specific risk to the cost of capital.
• A break below the 200-day moving average of $142.56 on company-specific news rather than sector beta.

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

• Price below approximately $150 (roughly 23.6x FY2027E, 28.6x FY2026E) with no deterioration in the fundamental picture. This is the single most likely path to a Buy and it is 12.5% below today's close. At $150 the base case is +18.7% and the risk/reward becomes genuinely asymmetric.
• The 2026-10-28 print beating $9.296B of revenue with disclosed organic growth at or above 28% and operating margin above 29% excluding tariff recoveries — the volume leg and the margin leg of 6a validated together, at which point paying 30x is defensible because the uncertainty has been resolved.
• A completed CommScope purchase-price allocation that lands materially more value in identifiable intangibles than in goodwill, reducing the impairment overhang, combined with a first disclosure of CommScope operating at or near the corporate margin.
• Net debt below 1.0x EBITDA on organic deleveraging, without a new acquisition.
• Genuine name-level knowledge-base conviction appearing — any claim about Amphenol at all, from any named speaker. At present there is not one, and that alone is worth roughly one tier.

Where APH fits in the Synthos Framework Portfolio. The AI-infrastructure supply-chain sleeve, at 0% today with a 2-2.5% target on a fill near $150 or a confirmed October print. On the batch overlap question: Amphenol is a genuinely different exposure from the wafer-fab-equipment complex in this batch (KLAC, LRCX, AMAT, ASML), which is one trade with four expressions. Amphenol sits downstream of the fab, not inside it — it sells into the assembled system, not into the process of making the chip. Its revenue is more diversified (seven end markets, no customer over 10%), its cyclicality is shallower (revenue fell only 0.5% in the 2023 downturn), and its growth in 2025 was faster than any of them. The trade-off is that it is a lower-margin, higher-leverage, more acquisitive business with half its balance sheet in goodwill. For a portfolio wanting AI-datacentre exposure without semiconductor capital-equipment cyclicality, Amphenol is the most natural single expression in this batch — at a price. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $171.37.

Single biggest risk: the IT datacom concentration reverses. IT datacom is 36% of revenue and supplied approximately $4.59B of the $7.87B of fiscal 2025 revenue growth — 58% of it. Communications Solutions, the segment that houses it, grew 71% organically in fiscal 2025 and is the highest-margin of the three at 33.6%. The entire valuation architecture — the 32.7x forward multiple, the $17.6B of goodwill, the $14.1B of net debt raised to buy more of the same exposure — rests on that one end market's capital-spending cycle continuing. The 10-K acknowledges it in the risk factors: "some of our customers are making significant investments in AI, and these investments are driving robust demand for certain of the Company's products. The continued growth of this market will be dependent upon many factors, including our go-forward market share for such products, the demand for our customers' products and services, the amount and mix of capital spending by our customers..." There is no valuation cushion to absorb a pause. At 32.7x forward earnings and 7.75x sales, a 15% revenue disappointment does not produce a 15% price decline; it produces a de-rating on top of it, which is what the $125 bear case is.

Most fragile assumption in the price: that organic growth stops decelerating. Every other assumption in 6a is observable quarter by quarter or is a margin assumption of modest size. This one is a rate-of-change assumption that has already moved against us twice in a row — 38% for fiscal 2025, 33% in March, 30% in June — and it becomes visible in the reported number only on 2027-01-09, when the CommScope contribution annualises out and headline growth converges to organic growth. Until then the acquisition is masking the trend, and consensus of +17.3% for fiscal 2027 quietly assumes both that the deceleration stops in the low twenties and that Amphenol buys something else. If organic settles at 15% and nothing is acquired, fiscal 2027 revenue is roughly $37.7B against consensus $41.2B, and at a 20.2% net margin and 1,310M shares that is approximately $5.81 of EPS against $6.37 — 8.8% below. At 26x that is $151 — 12% below today's close. That is the arithmetic of the base case failing, and it does not require anything to go wrong operationally.

Provenance & disclosures

• Traceability: ZERO tagged knowledge-base claims name Amphenol Corporation (raw hits on the ticker 636, entity matches 0, text matches naming the company 0, discarded for collision 636; breadth 0, net conviction none, claim count 0). The 636 raw hits are substring collisions on the letters "aph" inside ordinary English words — demographics, graph, paraphrase, sapphire — and the entity-match count of zero confirms the tagger never identified Amphenol as a subject of any claim. Searches run across all 51,928 distilled claims: APH (636 text, 0 entity, all discarded); Amphenol (0 of any kind); Adam Norwitt and Norwitt (0); CommScope (0); copper cable, cabling, electrical components, copper to optical (0); connectors (5 text, all about software API connectors, discarded); interconnect (72 text, all discarded — overwhelmingly product-architecture claims attributed to a channel bearing a large accelerator designer's chief executive's name, which under the 4-lane policy is simultaneously management/promoter-class voice, about a different company, and product rather than equity claims); AI capex (451 text, sector colour only, none names Amphenol); datacenter capex (80 text, same). Attribution quality across everything returned was uniformly the weakest form the policy admits: speaker: null with channel-only attribution on essentially every AI-capex and datacentre-capex claim sampled. No claim carries a management speaker_role for Amphenol, because there are no Amphenol claims. This is stated as a finding, not padded, and it is an explicit input to the Watch verdict. No claim is quoted in this dive because none exists to quote.
• Data as-of: fundamentals — income statement for the quarter and six months ended 2026-06-30 and the fiscal year ended 2025-12-31, sourced from the vendor payload and cross-checked line by line against the 10-K filed 2026-02-11 and the 10-Q filed 2026-07-31; balance-sheet data as of 2026-06-30 taken from the 10-Q, because the vendor's most recent annual balance sheet (2025-12-31) predates the CommScope acquisition · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873599 = 2026-08-04T19:59:59Z ($171.37, +4.92%; 50-DMA $155.66; 200-DMA $142.56; RSI 62.95; MACD +1.42) · knowledge-base claims 2026-08-04. Fiscal/calendar note: Amphenol's fiscal year ends 31 December, so fiscal and calendar years coincide. All figures come from the Synthos vendor data file for APH or from the SEC filings in the APH archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents and its gaps: 10-K filed 2026-02-11 (fiscal 2025); 10-Q filed 2026-05-01 (March 2026 quarter); 10-Q filed 2026-07-31 (June 2026 quarter — the freshest filing-verified balance sheet in this dive); 8-K filed 2026-05-12 (issuance of €600.0M of 3.375% Senior Notes due 2029 and €500.0M of 3.875% Senior Notes due 2034; net proceeds ≈€1,093.1M used to repay commercial paper and the 364-day delayed-draw term loan); 8-K filed 2026-05-22 (annual meeting, Item 5.07 — 1,229,430,709 Class A shares outstanding at the 2026-03-23 record date); 8-K filed 2026-07-29 (June-quarter results). The 2026-07-29 8-K is present as the cover document ONLY — Exhibit 99.1, the press release carrying the financials, was not extracted and the file contains no numbers. Nothing is lost, because the 10-Q filed two days later covers the same period. Numeric tables are stripped from all extracted filing text, including the 10-Q's Note 4 debt table, the segment tables, the non-GAAP reconciliations and the contractual-obligations table. Every filing-sourced figure in this dive comes from MD&A or footnote prose. Total debt at 2026-06-30 is RECONSTRUCTED, not quoted, and is labelled as such throughout.
• Where the filings contradicted or corrected the vendor (detailed in Section 8): the annual balance sheet was rejected as seven months stale and net debt rebuilt from the vendor annual file's $4,371.4M to approximately $14,081.9M at 2026-06-30, using the 10-Q's stated cash and short-term investments of $5,419.1M, its stated first-half borrowing and repayment flows ($4,927.2M and $1,535.7M) and the vendor's own current trailing block, which we verified is struck on June 2026 data; goodwill restated from $10,575.4M to $17,554.7M and intangibles from $2,241.4M to $5,288.9M per the 10-Q; the vendor's annual netDebt field additionally rejected for omitting $303.6M of short-term investments at 2025-12-31; GAAP operating income restored from the filings for every period ($5,868.6M for fiscal 2025 against the vendor's $5,972.0M; $3,156.9M for fiscal 2024 against $3,284.3M; $2,559.6M for fiscal 2023 against $2,594.2M; $1,418.8M for Q2 2025 against $1,430.8M) after establishing that the vendor consistently adds back the separately-presented acquisition-related expense line at the ANNUAL level but inconsistently at the QUARTERLY level; Q2 2026 diluted EPS taken as the 10-Q's $1.37 rather than the vendor's $1.38; seg_prod identified as summing $283.9M ABOVE fiscal 2025 reported revenue and $172.1M above fiscal 2024, on a total-sales rather than external-net-sales basis, with the reconciliation printed and the external figures recovered independently from the 10-K's segment operating margins; seg_prod's fiscal 2025 label "Interconnect Products And Assemblies" rejected as a superseded pre-2022 taxonomy label and restored to the filings' "Interconnect and Sensor Systems"; seg_geo's mangled "C [N]" and "U [S]" labels corrected to China and United States; researchAndDevelopmentExpenses of $0 for fiscal 2025 and stockBasedCompensation of $0 for fiscal 2025 both rejected as missing values; capitalLeaseObligations dropping to $0 noted as breaking the year-on-year debt comparison; all est EBIT, EBITDA and net-income rows rejected as arithmetically impossible (fiscal 2029 net income exceeds EBIT); quote.yearHigh/yearLow ($178.52/$104.71) rejected in favour of tech.hi52/lo52 ($176.32/$107.93); lastDividend of $0.915 identified as the trailing-twelve-month total rather than the last declared dividend, with the filings giving a forward run-rate of $1.00 (0.583% yield). Where vendor and filing AGREED — worth recording, because it was most of the file: fiscal 2025, 2024 and 2023 revenue; all three years of operating cash flow; the fiscal 2025 buyback of $665.2M and 7.4 million shares; cash and short-term investments of $11,434.2M; goodwill, intangibles, receivables, inventories, PP&E and payables at 2025-12-31; quarterly interest expense summing exactly to the 10-Q's first-half figure of $421.6M; and the share count (market cap implies 1,230,283,915 against the 8-K's 1,229,430,709 at an earlier date — 0.07%).
• Basis note: analyst estimates are struck on the company's ADJUSTED (non-GAAP) basis, and this was verified arithmetically rather than assumed — first-half 2026 adjusted diluted EPS of $2.42 plus the $1.42 Q3 consensus leaves $1.398 for Q4 against a fiscal 2026 epsAvg of $5.238, a coherent sequence; the same test on GAAP (first-half $2.10) would require an implausible $3.14 second half. Forward multiples in this dive therefore use consensus adjusted EPS, and the trailing multiple is shown on both bases (39.8x adjusted, 42.8x GAAP). The company's adjustments exclude acquisition-related expenses, refinancing costs, excess tax benefits from stock-based compensation and certain discrete tax items. They do NOT exclude the $80.0M IEEPA tariff recovery recorded in the June quarter, which added 91 basis points to operating margin, and this dive flags that separately.
• Estimate coverage: 11 analysts on fiscal 2026 and fiscal 2027 revenue; 8 on fiscal 2026 EPS, 10 on fiscal 2027 EPS, 5 on fiscal 2028 EPS and ONE on fiscal 2029 EPS. The fiscal 2029 row is excluded from every conclusion in this dive. Fiscal 2028 is used only as a cross-check. All fair values anchor on fiscal 2027, the deepest forward EPS year. epsAvg was verified against actuals (fiscal 2025 estimate $3.30749 vs actual $3.34; fiscal 2024 estimate $1.8393 vs GAAP $1.92) and is sound with a mild, consistent conservatism.
• Peer note: the vendor peer set — Accenture, Applied Materials, Arista Networks, Arm, Intel, Intuit, KLA, Lam Research, Sony, Texas Instruments — contains no interconnect or connector comparable whatsoever. No TE Connectivity, no Sensata, no Molex. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn, and the reader should be aware that Amphenol has been benchmarked here against semiconductor capital equipment and enterprise software, which is not informative. The nearest adjacent name in the set is Arista Networks ($239.9B, $190.55), which is AI networking equipment rather than components.
• Fair-value caveat: the $125 / $178 / $229 anchors are multiples of the fiscal 2027 consensus adjusted EPS distribution — 21x the low of $5.958, 28x the mean of $6.366, and 32.7x (today's forward multiple, held) the high of $6.99. Stated arithmetic, not a discounted cash flow. The base is deliberately sensitivity-disclosed: 26x gives $165, which is BELOW spot; 30x gives $191, essentially the street's $192.33. The whole answer lives in a 26-30x band and no greater precision is claimed. The bear case assumes a de-rating and an estimate cut; the bull case assumes no re-rating at all, only an earnings beat.
• Timing: June-quarter results were released 2026-07-29, six days before this dive, and the 10-Q covering them was filed 2026-07-31, three days before — so the entire first half is in the data and filing-verified. 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 Watch verdict rather than an oversight. 2026-08-04 was a violent sector-wide semiconductor and AI-hardware up-day (MRVL +12.81%, INTC +10.92%, LRCX +7.85%, MU +7.62%, AMD +7.00%, KLAC +6.95%, AMAT +5.48%, APH +4.92%, ASML +4.22%); Amphenol's 4.92% was the second smallest of that group, carried no company-specific news in this file, and occurred on 6.83M shares against a 9.62M average — 29% below normal volume. It is treated as sector beta.
• 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.