APHAmphenol Corporation

APH — dossier, 2026-09-21

Slot names-ai-capex-3. Model-drafted unattended by the cloud research routine; nothing here is human-verified. Sources and locators are in relationships.yaml; figures below carry theirs inline. Anything from background knowledge is marked [background].

Price reference: $80.86 close, 2026-09-21 (ROIC.ai get_latest_stock_price NYSE:APH; volume 6,609,560 shares, so roughly $534m traded that day — liquidity is not a constraint at any size this workspace would contemplate).

Read the share count before anything else. Amphenol effected a two-for-one stock split on 2026-09-02 (8-K filed 2026-09-04, Item 8.01: "The two-for-one stock split referenced above has now been effected."). Every per-share figure in every filing and transcript quoted below predates it and is therefore pre-split. Diluted weighted-average shares were 1,289.2m for Q2 2026; post-split that is ~2,578.4m, giving a market capitalisation of ~$208.5bn at $80.86, not the ~$104bn a naive read of the last 10-Q would produce. The Q3 guide of $1.40–$1.42 adjusted diluted EPS is $0.70–$0.71 post-split. This is the single easiest way to get this name wrong by a factor of two and it is why the split is recorded as an edge.

S1 — Source

watchlist (owner, Apple Stocks, 2026-09-20), cluster ai_semis, coverage sec_domestic, cik 820313. No Tier 0 promotion; queue.json has never promoted APH.

S2 — Kill test

All figures from the XBRL company facts of the SEC filings (data.sec.gov/api/xbrl/companyfacts, CIK 0000820313). Calendar fiscal year.

Quarterly, the six most recent (10-Q / 10-K XBRL):

LTM (2025-07-01 → 2026-06-30): revenue $29,011.6m, gross profit $11,168.6m (38.5%), operating income $7,841.4m (27.0%), net income $5,143.4m, D&A $1,303.9m → EBITDA ~$9,145m.

Balance sheet at 2026-06-30 (10-Q): long-term debt and lease obligations $17,177.0m plus current portion $1,634.3m = $18,811.3m gross; cash $4,729.4m plus short-term investments $689.7m → net debt $13,392.2m, or 1.46x LTM EBITDA. Stockholders' equity $15,491.6m. Goodwill $17,554.7m — goodwill alone exceeds book equity, which is the arithmetic behind the company's own impairment risk factor.

Maturity wall (FY2025 10-K schedule, so pre-CommScope-drawdown): $937.2m due within twelve months, $1,950.9m in year two, $7,780.9m after year five. Liquidity is backstopped by a revolving credit facility, a US and a Euro commercial paper programme (nothing drawn at 2025-12-31 or 2024-12-31), and two delayed-draw term loans of which $1,534.1m each was drawn in January 2026 to fund CommScope.

Share count over five years — the thing that looks alarming and is not. 626m (2021) → 1,242m (2022) → 1,278m (2025) as filed, then a further doubling on 2026-09-02. Both jumps are two-for-one splits, not issuance. On a constant post-2026 basis the diluted count went from ~2,504m (2021) to ~2,555m (2025): ~2% over four years, against $3.2bn of buybacks and $2.7bn of dividends in that window. Amphenol does not fund itself with its own stock.

Analyst coverage count: not obtained this run (no source on this ladder gives a clean count). Six sell-side firms asked questions on the Q2 2026 call — Jefferies, J.P. Morgan, Evercore, Baird, Truist and TD Cowen appear in the turns read — which is a floor, not a count.

Verdict — survives, comfortably. A 27% operating margin, $4.4bn of FY25 free cash flow, net debt at 1.5x EBITDA and no dilution. The kill test does not fail; the risk here is the multiple and the durability of the demand, not solvency. The one caution the numbers themselves raise: FY25's margin expansion (20.7% → 25.4% operating) and Q2 2026's 40.5% gross margin are cycle-peak readings on a business that ran at ~32% gross margin for a decade, and $17.6bn of goodwill now sits against $15.5bn of equity.

S3 — Filings, gate-minimum

Read this run: the FY2025 10-K (Item 1A risk factors, Item 7 MD&A, Note 15 subsequent events), the Q2 2026 10-Q (MD&A delta, debt notes, acquisition note), the 2026 DEF 14A (ownership tables and the incentive-plan criteria), the Q2 2026 earnings call (turns 3–20 of 48, prepared remarks plus the first five analyst exchanges), and the complete SEC comment-letter exchange on the FY2024 10-K. Not read: the FY2024 10-K, older 10-Qs, the Q1 2026 call, the investor-day material.

The three biggest risks management itself discloses

1. Non-US exposure and trade policy — Item 1A, "Risks Related to our Global Operations", first risk factor: "During 2025, non-U.S. markets constituted approximately 65% of the Company's net sales, with China constituting approximately 16% of the Company's net sales. The Company employs approximately 90% of its workforce outside the United States." The same paragraph adds that "approximately 79% of the Company's long-lived assets were located outside of the United States". Q2 2026's margin was flattered by net tariff refunds — management said excluding them operating margin "still reached nearly 29%" against the reported 29.8%, and guided that further net tariff recoveries would be immaterial.

1. End-market cyclicality — Item 1A, "Risks Related to our End Markets": "The Company is dependent on end market dynamics to sell its products, and some of the Company's end markets are subject to cyclical and at times rapid periods of reduced demand." With IT datacom at 43% of Q2 2026 sales and the company attributing virtually all of that growth to AI, this generic risk factor is now concentrated in one place. It is already visible elsewhere in the mix: communications networks sales fell 6% organically year on year and are guided down in the mid-teens sequentially for Q3.

1. Acquisition integration and goodwill — Item 1A: "The Company has completed numerous acquisitions in recent years, including five in 2025 and two in 2024, some of which are large and complex. Additionally, on January 9, 2026, the Company closed the CommScope acquisition, which is the largest acquisition in the Company's history." And, separately: "On December 31, 2025, the total assets of the Company were $36.2 billion, which included $10.6 billion of goodwill ... and $2.2 billion of other intangible assets, net." Post-CommScope that goodwill is $17.6bn.

Latest quarter, MD&A delta

Q2 2026 revenue $8,758.1m, +55% in USD, +54% constant currency, +30% organic. Segment mix moved hard: Communications Solutions is now ~62% of sales (from ~52% for FY25), Harsh Environment ~21%, Interconnect and Sensor Systems ~17%. Management's own end-market split for the quarter, from the call: IT datacom 43%, industrial 20%, communications networks 11%, automotive 10%, defence 8%, commercial air 4%, mobile devices 4%.

On the call the CEO reported record orders of $10.7bn, a 1.23 book-to-bill, orders +94% year on year and +63% organically, and raised the CommScope contribution to $4.6bn of 2026 sales and $0.30 of accretion from $4.1bn and $0.15. Q3 2026 guidance: sales $9.3–9.4bn, adjusted diluted EPS $1.40–$1.42 (pre-split; $0.70–$0.71 post-split), i.e. 50–52% sales growth.

Proxy skim

All executive officers and directors as a group (16 persons) hold 17,469,159 shares, 1.42% of the class as of 2026-03-23; every individual director and officer is marked "less than one percent". Five-percent holders: Vanguard 10.2%, BlackRock 8.0%, FMR 7.0%.

The incentive plan is the interesting part. For the CEO and CFO the 2025 quantitative criteria were revenue growth of 7% and Adjusted Diluted EPS growth of 11% for a 100% multiplier, and "To achieve a 200% quantitative portion of the incentive plan multiplier in 2025 would have required Company revenue growth of at least 17.5% and Company Adjusted Diluted EPS growth of at least 27.5%." Delivered: 52% and 77%. Two things follow. The plan is capped far below what this cycle pays, so there is no marginal bonus incentive to chase the AI number — a mild positive. And plan revenue growth is not adjusted for acquisitions, in a company that completed five acquisitions in 2025 and a $10.5bn one in 2026 — which is where this joins the comment letter below.

Insider pattern

3spread (insiders APH --days 365): 91 Form 4 filings in the twelve months to 2026-09-21, clustered on 2026-05-27 (an annual grant/vest date) and in the first week of August 2026, i.e. the window immediately after the 2026-07-29 results. sales APH: 10 Form 144 proposed sales since 2026-03-25, all in the same two windows (three around 2026-05-01–05, seven from 2026-07-31 to 2026-08-05). Direction is one-way — there are proposed sales and no purchases on file — but the clustering is exactly what a 10b5-1 / post-earnings-window programme looks like, and the individual Form 4 amounts were not itemised this run. Recorded as a pattern to watch, not as a signal.

Red-flag sweep — one real finding, and it closed

The SEC's Division of Corporation Finance, Office of Manufacturing, reviewed the FY2024 Form 10-K and issued comments on 2025-02-07. Comments 3 and 4 asked about Amphenol's ASC 805 acquisition disclosure. Amphenol responded on 2025-05-16 and supplementally on 2025-06-20 over the CFO's signature, arguing that because the interconnect market "is highly fragmented and comprised mostly of small, niche manufacturers" it discloses acquisition details "even when those acquisitions are immaterial or in many cases trivial in size", and concluding that "we continue to believe that the 2024 Acquisitions and the 2025 Acquisitions were not material, individually or in the aggregate, to the Company". The staff closed the review on 2025-06-25: "We have completed our review of your filing."

No restatement, no amendment, no NT 10-K. The finding is not a defect — it is that the SEC pressed this exact question seven months before Amphenol closed the largest acquisition in its history, one that added $7.0bn of goodwill in a single quarter. The company's answer was "individually and in aggregate immaterial". That answer was about 2024 and 2025 and it does not carry to CommScope, which is separately disclosed — but the roll-up is now large enough that "immaterial in aggregate" is a claim worth re-testing at the next 10-K. The efts.sec.gov full-text sweep for "changes in accountants" / "restatement" was not run this session (time), so auditor change is unchecked beyond the absence of an NT 10-K or 8-K Item 4.01 in the filing index.

Equibles GetCustomerConcentration (3 most recent disclosing filings): the last figure Amphenol tagged is an 11%-of-revenue largest customer for FY2020, in the 10-K filed 2023-02-08. Nothing since. At 43% of sales in one end market, the absence of a disclosure is the disclosure.

S5 — Ownership

Top managers, 13F report date 2026-06-30 (threespread.py holders APH, CUSIP 032095101; 1,614 filers reporting 601,165,439 shares — pre-split basis):

⚠️ The quarter is incompletely ingested — 1,614 filers against 2,099 for 2026-03-31. The script names ten prior holders with no 13F in the quarter yet, including FMR (79.5m shares), JPMorgan (49.3m), Capital International (35.3m), Geode (31.3m), Janus Henderson (18.9m), Morgan Stanley (17.5m) and Northern Trust (13.5m). No exit may be read from any of those absences.

Two of those absences have a documented cause. Both "VANGUARD CAPITAL MANAGEMENT LLC" (159.8m shares) and "VANGUARD PORTFOLIO MANAGEMENT LLC" (35.9m) are listed as prior holders with nothing in the current quarter. The proxy explains why: "On March 26, 2026, The Vanguard Group filed an amendment to its Schedule 13G disclosing that, as a result of an internal realignment, it is no longer deemed to beneficially own any shares of Common Stock." Vanguard restructured which legal entities report. This is almost certainly the same cause behind the "both Vanguard entities absent" warning carried in the MRVL scorecard from slot names-ai-capex-2 — it is a reporting change across Vanguard, not a position change in any one issuer.

13D/G: nine filings on file, all 13G or 13G/A — passive. No 13D, no activist.

Insider ownership: 1.42% for the 16 officers and directors as a group.

Short interest (Equibles GetShortInterest, FINRA settlements): 63.95m (2026-06-30) → 66.03m → 61.67m → 59.09m (2026-08-14) → 27.12m (2026-08-31). The last step is a 54% fall in one settlement, alongside average daily volume halving from 22.5m to 11.5m. That boundary sits days before the 2026-09-02 two-for-one split, and the volume series halves with it, so the drop reads as a split-restatement artefact in the feed rather than a covering event. Do not read a squeeze into it. On either basis days-to-cover is 1.6–2.6 — there is no crowded short here. The tool's model estimate for the 2026-09-15 settlement (~28.3m) is explicitly a prediction, not FINRA data.

Shelf, ATM and buyback. A live automatic shelf: S-3ASR filed 2026-03-02 — Amphenol is a well-known seasoned issuer and can sell securities at will; it has funded every recent acquisition with debt and cash rather than stock, but the capacity is there and unlimited by construction. Buyback: $2.0bn authorised 2024-04-23, $411.1m remaining as of 2026-07-29, i.e. nearly exhausted and due for re-authorisation. Repurchases have run $585m–$935m a year for a decade — $665.2m in FY25 and $386.0m in H1 2026 — so this is a steady offset-the-grants programme, not a capital-return lever. Dividends were $802.2m in FY25 and $614.0m in H1 2026. No ATM programme was sought (the company is investment grade and does not run one) — not checked, and absence is not proof.

Who is on the other side: index funds and long-only growth managers, with T. Rowe the loudest active buyer of the quarter and two new positions above $1.3bn. No activist, no crowded short, no shelf overhang that has ever been used for equity. The other side of this trade is the multiple.

S8 — Valuation and sizing (proposed only; owner-only items OPEN)

At $80.86 (2026-09-21) with ~2,578.4m post-split diluted shares: market cap ~$208.5bn, net debt $13.4bn, EV ~$221.9bn.

Against LTM: 7.6x EV/sales, 24.3x EV/EBITDA, 40.5x trailing GAAP EPS (LTM GAAP diluted EPS $1.995 post-split).

Out-year model. Revenue is built from H1 2026 actual plus the Q3 guide midpoint plus a Q4 in line with the sequential trend; margins from the Q2 2026 run rate, which management calls a record.

The base case is +21% over three years from today's price, which is not much for the risk in the bear column. What the model is really saying: at 30x FY26 the market already pays for the AI mix, and the return from here depends on 2028 revenue being roughly $51bn — more than double FY2025.

⏳ Buy price and position size are owner-only and stay OPEN (user). For the record the liquidity constraint is not binding: ~$534m traded on 2026-09-21.

S9 — Monitoring

Next earnings: Q3 2026 results, expected in the last week of October 2026 — derived from filing cadence (Q3 2025 10-Q filed 2025-10-24; Q2 2026 call 2026-07-29, 10-Q 2026-07-31), not from a calendar feed. ROIC.ai's earnings calendar needs a paid plan on this account, the same limit slot names-ai-capex-1 hit.

Filing feed: the Tier 0 EDGAR sweep covers cik 820313 (ledger/research/tools/watch_edgar.py), so 8-K, 10-Q, S-3, Form 4 and any Item 4.01 arrive without a subscription.

The leading indicator: orders and book-to-bill. Amphenol reports both every quarter on the call and they lead revenue by roughly two quarters. Q2 2026 was a record $10.7bn of orders at 1.23 book-to-bill, +63% organically. A book-to-bill that falls below 1.0 — before any revenue miss — is the earliest disclosed number that would tell you the AI interconnect build has turned. Secondary: the organic IT datacom growth rate (+63% in Q2) and the communications networks line, which is already contracting organically and is guided down mid-teens sequentially in Q3.

Trigger that converts this to a buy: a quarter in which book-to-bill stays above 1.0 and organic IT datacom growth stays positive while the stock de-rates below ~20x the forward EPS on this model — i.e. the cycle is intact and the multiple has broken. On the FY2027E line that is roughly $67. The specific price and the size remain the owner's.