APH — Amphenol Corporation

Sentiment is bullish but not euphoric, which is worth noting because the headlines oversell it. As of 2026-09-13 the stock had returned +25.33% over six months and +9.13% over three months, and popped 4.57% on the day, yet RSI was only 57.4 - buyers are in control, not frothing. The sell-side is more balanced than the 'strongly bullish' narrative suggests: 16 Buy, 13 Hold, 1 Sell. The consensus price target of $175.67 is almost certainly stale pre-split data (halve it to ~$88), and recent-average targets (~$170.5, or ~$85 post-split) plus TD Cowen's $90 Hold imply analysts see roughly 5-10% upside, not the 100% the raw target number implies. Drivers: the Citi TMT conference reaffirmation of order momentum, record Q2 orders, and APH's status as a picks-and-shovels supplier to AI data center buildouts.

Revenue has gone from $3.20B in Q3 2023 to $8.76B in Q2 2026 - up 174% in under three years - and the trend is still accelerating, not decelerating. YoY growth ran 26-30% through 2024, then stepped up to 48-59% across every quarter of 2025 and 2026 (Q1'26 +58.4%, Q2'26 +55.0%). Sequential growth is also unbroken: Q2'26 was +14.9% QoQ after +18.3% in Q1'26. Two important caveats. First, only ~30 points of Q2'26's 55% YoY growth was organic per management; the rest is acquired (CommScope's CCS business, El.Com, Wilder Technologies), so roughly half the headline growth is purchased, not earned. Second, the margin story is the genuinely impressive part: gross margin expanded from 33.6% in Q2'24 to 40.5% in Q2'26, and operating margin from 21.3% to 29.5% - roughly 820bps of operating leverage in two years, driven by AI datacom mix and post-acquisition scale. Earnings have followed (Q2'26 net income $1.77B, +62% YoY; diluted EPS $0.69, +60%), with one visible blemish: Q1'26 revenue rose 58% YoY but net income rose only 28%, because the debt raised for CommScope pushed long-term debt from $7.1B (Q3'25) to $16.6B, and interest expense plus amortization ate the operating beat. That disconnect largely resolved in Q2'26. Free cash flow is compounding too - TTM FCF of $4.71B versus ~$2.1B in 2024 - and capex is rising ($355.5M in Q2'26, up 20% YoY), which is a genuine signal of capacity being built for demand rather than a warning sign. Bottom line: this is an accelerating, structurally improving trend, with the honest asterisk that acquired growth flatters the headline.

Expensive on trailing numbers, defensible-to-slightly-rich on forward numbers - and that distinction matters. As of 2026-09-13 the stock carried a 39.96x trailing P/E against a 5-year average of 32.07x (a ~25% premium to its own history), a 7.13x P/S against a 5-year average of 5.15x (+38%), 13.32x book, 24.6x EV/EBITDA, and a 2.3% TTM free-cash-flow yield. The deterministic score's own valuation overlay lands at 10.6/100 and labels it 'Rich vs. its own history' - that is directionally correct. The reconciliation is forward earnings: on consensus FY2026E EPS of $2.641 the forward P/E is 31.8x, and on my FY2027E EPS of ~$2.97 it is ~28x. So the trailing 40x is really a trough-EPS artifact of acquisition amortization and financing costs that are still working through the P&L, and the PEG of 0.676 - while flattered by M&A-inflated trailing growth - is not meaningless. Verdict: not a bubble, but no margin of safety either. You are paying near the top of APH's own historical band for growth that is only half organic.

The market is pricing in continued high-teens EPS growth and a successful CommScope integration - it is not pricing in a stumble. At ~32x FY2026E EPS, the multiple assumes the 30% organic AI datacom growth persists into 2027 and that the 820bps of margin expansion holds. The asymmetry is unfavorable: if organic growth decelerates to high single digits as 2027 comps get harder, earnings might still grow while the multiple compresses from 32x to the mid-20s, and you could see a $70-75 stock despite an intact P&L. Conversely, estimate revisions have been persistently upward and the $10.7B backlog with 1.23x book-to-bill is real forward visibility, so the path of least resistance stays up while orders stay above 1.0x. Practical read: this is a 'buy on evidence, not on narrative' name right now - the single most informative number each quarter is book-to-bill. Below 1.0 would be the first credible crack in the thesis and should be treated as a sell signal rather than a buying opportunity.

As of 2026-09-13, RSI(14) stood at 57.42 - constructive but nowhere near overbought, leaving room before momentum becomes a problem. Price was 4.1% above the 20-day (approx. $80.6), 4.54% above the 50-day (approx. $80.3), and 14.68% above the 200-day (approx. $73.2), a clean and orderly uptrend rather than a parabolic one. The stock sat ~6.0% below its 52-week high of $89.26, and the one-month return of -1.25% against a six-month +25.33% says the recent action is shallow consolidation inside a strong trend, not distribution. Entry-timing implication: this is not an oversold entry and it is not an overbought chase - it is a 'wait for a tag of the 50-day' setup. The $78-81 zone (20/50-day cluster) is the logical add level; a decisive break below the 200-day near $73 would invalidate the trend and should be respected. Note these levels are split-adjusted post the August 2026 two-for-one split and are anchored to 2026-09-13, not live.

• AI/hyperscaler data center capex cycle - IT Datacom is roughly 41% of revenue and is the highest-growth, highest-multiple part of the story; any digestion pause at a major cloud customer hits both earnings and the multiple simultaneously.
• Interest rates and the Fed path - APH is now a levered borrower ($17.18B long-term debt vs $4.73B cash as of 2026-06-30), so refinancing costs matter to EPS conversion, while the discount rate also drives the multiple on long-duration AI growth.
• Industrial and automotive end-market softness plus tariff/trade policy - connectors cross borders constantly, and a broad industrial production downturn would pressure the non-AI half of the portfolio.
• Defense and aerospace budget tailwinds - geopolitical spending provides a stable, counter-cyclical offset to datacom cyclicality and supports margin mix.
• Serial M&A integration risk - the balance sheet was transformed by CommScope CCS and several bolt-ons; execution and goodwill quality now matter as much as the core business.

• Q3 2026 earnings on October 28, 2026 - the key tells are book-to-bill staying above 1.0x, operating margin resilience near 29%, and any upward revision to full-year guidance.
• CommScope CCS integration progress and realized cost synergies - evidence of accretion above plan would justify the current multiple; a shortfall would not.
• Hyperscaler capex guidance updates from the major cloud buyers, which the entire AI interconnect complex trades off in real time.
• Further bolt-on acquisitions funded from free cash flow, or alternatively visible deleveraging as the $17.2B debt stack is paid down.
• Post-split retail/index flow - the August 2026 two-for-one split broadens the shareholder base and can mechanically support demand, though it changes nothing about intrinsic value.

• Customer and end-market concentration in AI datacom - roughly 41% of revenue and the majority of incremental growth now depends on a handful of hyperscaler capex programs; a pause there would hit both earnings and the premium multiple at once.
• Balance-sheet transformation - long-term debt went from $7.13B (Q3 2025) to $17.18B (Q2 2026) against $4.73B cash, taking debt-to-equity to 1.214 versus roughly 0.5x two years ago. Interest expense already visibly compressed Q1'26 earnings conversion (revenue +58%, net income +28%). Net debt/EBITDA near 1.5x and $4.7B TTM FCF keep this manageable, not dangerous, but it is a genuine change in the risk profile.
• Multiple compression - 39.96x trailing earnings and 13.3x book leave essentially zero cushion. A pure de-rating to the mid-20s forward P/E would take the stock into the $70s even with earnings fully intact.
• ROIC dilution from serial M&A - reported ROIC of 15.1% is mediocre for a company priced at these multiples, because goodwill and intangibles from CommScope and the bolt-ons inflate the capital base; if returns on that deployed capital disappoint, the premium is unjustified.
• Cyclicality - Amphenol is a high-quality compounder but not immune to industrial downturns, and the connector industry has historically been cyclical even when management execution is excellent.

Amphenol is a genuinely excellent business doing genuinely excellent things - 30% organic growth, a record $10.7B order book, and 820bps of operating margin expansion since 2024 - but the market has noticed, and at 39.96x trailing earnings and 31.8x FY2026E EPS the shares already discount a lot of it. My calculated fair value of $92 implies roughly 10% upside as of 2026-09-13, which is a real but unexciting return for a levered, half-acquired growth story. I would hold this and add on weakness toward $75-80, but I would not chase at $84 near the 52-week high, and I would treat a book-to-bill print below 1.0x as the signal to reassess rather than to buy more.

Primary method: apply 31x to my own FY2027E EPS of $2.97, built from FY2027E revenue of ~$40.3B (+14% off the consensus FY2026E figure of $35.38B, reflecting deceleration from ~50% to mid-teens as AI comps harden) at an 18.2% net margin on ~2.47B shares - 31x being a discount to the 39.96x trailing P/E and roughly in line with APH's 5-year average trailing P/E of 32.07, justified by the demonstrated margin expansion but capped by the post-CommScope leverage. Cross-check with EV/EBITDA: ~20x FY2027E EBITDA of ~$12.3B less ~$14B net debt gives roughly $94 per share, confirming the ~$92 level. Both approaches point to modest single-digit-to-low-teens upside from the $83.92 price as of 2026-09-13, not the near-doubling implied by the stale pre-split analyst consensus target of $175.67.