$830M Above the Ceiling, Datacom at 48% of Sales, and a Stock Back Where It Was in July: Maintaining Outperform

Per-share figures throughout are on the post-split basis. Amphenol's 2-for-1 split took effect on September 3, 2026, and prior-period EPS, guidance and share prices are restated accordingly (the July guide of $1.40 to $1.42 becomes $0.70 to $0.71; the July 31 close of $160.70 becomes $80.35).

Results vs. Consensus

Q3 2026 Scorecard

Year-Over-Year and Sequential

Q2 2026 per-share and share-count figures are split-adjusted (Q2 GAAP net income of $1,769.2M and adjusted net income of $1,745.7M over 2,578.4M diluted shares). The nine-month adjusted EPS of $2.01 equals $0.53 + $0.68 + $0.80. Book-to-bill for Q3 2025 is orders of $6.111B over sales of $6,194.4M.

Quality of Beat

Revenue. The headline +65% splits into 37 points of organic growth and 28 points of acquisitions, with currency neutral. Sequentially the company grew 17% in dollars and 16% organically, against a July guide that implied 6% to 7%. The composition is the point to dwell on. IT datacom grew 30% sequentially against a mid-teens expectation, mobile devices 32% against roughly 20%, automotive 12% against flat, industrial 7% against roughly flat, and commercial air 9% against a modest increase. Defense grew 12% in line with a low-double-digit guide, and communications networks fell 13%, in line with a guided mid-teens decline. Five of seven markets beat their own sequential outlook. On the year-over-year view, five of seven markets grew more than 20% organically. The single largest contributor was datacom at +77% organic, and management said AI drove the vast majority of the sequential gain there.

Margins. Adjusted operating margin of 29.7% fell 10bp from Q2's record as reported, entirely because Q2 carried about $80M of net tariff recoveries and Q3 carried about $15M. Ex-tariff, the margin rose from roughly 28.9% to roughly 29.5%, the approximately 70bp improvement the CFO cited. The path there is different from Q2. Gross margin fell 125bp sequentially to 39.3%. If the tariff recoveries sit in cost of sales, as tariff items usually do, ex-tariff gross margin still fell about 50bp, from 39.6% to 39.1%. SG&A fell from 10.7% to 9.6% of sales, 115bp of leverage, which more than covered it. We read the gross-margin dip as mix: CommScope grew about 31% sequentially on our arithmetic at a low-20s operating margin, mobile devices grew 32%, and a newly acquired automotive assembly business entered the Interconnect and Sensor Systems segment in September. Management called the cost environment "challenging" but did not discuss gross margin at all. As-reported sequential incremental conversion was 29.1%; ex-tariff it was 33.5%, still above the roughly 30% framework but below Q2's roughly 39%.

EPS. Adjusted EPS of $0.80 beat by 9.6% and is clean at the tax line. The adjusted rate was 27.0%, unchanged for a third quarter, and GAAP EPS of $0.79 sits a penny below it. The GAAP figure absorbs a $60.0M accrual for "settlements of foreign tax matters" and $15.5M of acquisition costs, partly offset by $52.2M of excess stock-compensation tax benefits. Interest expense was flat sequentially at $212.3M after the company repaid $937.9M of senior notes in the quarter. The $15M tariff benefit is worth less than half a cent. The share count rose only 0.2% sequentially, so per-share growth of 18% tracked adjusted net income growth of 17.8%. The one composition point to hold: the company has now excluded $389.0M of discrete tax items from adjusted EPS in nine months, $0.15 per share. China items make up $290M of it (a $130M assessment accrual and a separate $160M reassessment of prior-year rates), and the $230M assessment was paid in full in Q2. But Q2 and Q3 each added a new accrual for unnamed "foreign tax matters," $39.0M and $60.0M.

Segment Performance

Communications Solutions: 64% of the Company, and a Margin That Slipped While Sales Rose 22%

Communications Solutions nearly doubled in dollars to $6.583B and grew 51% organically, accelerating from 42% in Q2. It is now 64.4% of Amphenol, up from 61.5% last quarter and 53.4% a year ago. The 10-Q attributes the growth to "outsized organic growth in the IT datacom and industrial markets," strong mobile and automotive, and acquisitions, offset by the organic decline in communications networks. Segment operating income rose $372M sequentially on $1.2B of extra sales, a 31% incremental margin, and the margin itself slipped 50bp to 33.1%. That is the arithmetic of CommScope growing faster than the rest of the segment: on our derivation from the 10-Q, CommScope's sales rose from just over $1.2B in Q2 to roughly $1.58B in Q3 at a margin the CFO put in the low 20s.

"Accordingly, we now expect CommScope sales in the IT Datacom market to more than double for the full year 2026 and we expect that those sales should represent roughly half of their sales for the full year and that compares to around a third of their sales."
— R. Adam Norwitt, Chairman and CEO

Assessment: The segment is doing exactly what the thesis needs and is getting more concentrated while it does it. A 31% incremental margin on $1.2B of sequential growth, with the lowest-margin piece growing fastest, is a good result. The cost of that result is that the segment's margin now depends on CommScope's convergence path, and the segment's revenue depends on one end market for well over half its sales. We would watch the segment margin rather than the corporate one next quarter: a second decline would say the mix headwind is outrunning CommScope's improvement.

Harsh Environment Solutions: Defense at 30% Organic, Margin Held at 30.1%

Harsh Environment grew 34.7% to $2.042B, 29% organically, an acceleration from 22% in Q2, and held its 30.1% margin with a 30% sequential incremental margin. Defense, 8% of company sales, grew 42% in dollars and 30% organically, with growth "across virtually all segments of the defense market." Commercial air, 3% of sales, grew 22% organically and 9% sequentially against a guide for a modest increase. The 10-Q also credits IT datacom within this segment, a reminder that AI demand now shows up in all three reporting segments rather than in one.

"With our expanded product offering, as well as the significant capacity expansions we have made in recent years, we are positioned better than ever to capitalize upon this long term demand trend."
— R. Adam Norwitt, Chairman and CEO

Assessment: This is still the ballast, and it is better ballast than it was a year ago: 27.1% margin then, 30.1% now, at 29% organic growth. Defense is now guided up more than 35% for the full year, and the pending AdvancedPCB acquisition (roughly $270M of 2026 sales, printed circuit boards for defense and industrial) lands here. If AI demand pauses, this is the segment that keeps the margin above 29%.

Interconnect and Sensor Systems: Growth Slowed and the Margin Gave Back 100bp

The smallest segment grew 17.2% to $1.605B, 11% organically, a deceleration from 13% in Q2, and its margin fell from 21.0% to 20.0%. Segment operating income rose only $1.8M on $87M of sequential sales growth. Two things happened at once. Empire Electronics, an automotive interconnect assembly business with about $320M of expected 2026 sales, closed at the beginning of September and entered this segment. And automotive grew 12% sequentially against a flat guide, partly on that acquisition.

"Sequentially, our sales grew by 12% from the second quarter which was much better than our expectations. Reflecting the ability of our team to quickly execute on a range of opportunities together with some impact from the addition of Empire in the quarter."
— R. Adam Norwitt, Chairman and CEO

Assessment: A flat sequential profit on rising sales is the weakest line in the segment table, and the most likely explanation is an acquired assembly business entering below the segment's margin, which is how Amphenol's acquisitions usually start. Humanetics Sensors (about $70M of sales) will also land here. We would expect the margin to recover toward 21% over a few quarters, as it did after earlier deals, but this segment has run between roughly 19% and 21% since mid-2025 and has the most self-help still available.

End-Market Detail

Three features of this table matter more than the datacom row. First, breadth held: five of seven markets grew more than 20% organically, five of seven beat their own sequential expectation, and management said that without AI the company grew organically in the high teens. Second, the one negative market got worse, not better. Communications networks went from −6% organic in Q2 to −28% in Q3 on broadband and operator softness, and it is guided down again in Q4. At 8% of sales it is small, but it is the only market where CommScope's legacy exposure works against the thesis. Third, book-to-bill was positive in all but one market, against every market in Q1 and Q2. Management did not name the exception, and the table makes communications networks the likely candidate. That is our inference, not a disclosure.

Key KPIs

The Q3 2025 IT datacom share is implied from 100% U.S.-dollar growth to roughly $4.9B. Days sales outstanding is period-end receivables over quarterly sales times days in the quarter.

Key Topics & Management Commentary

Overall Management Tone: As assured as at any point in this coverage, and more expansive than in July: management spent much of the question period on culture, defense and the decentralized operating model rather than on defending the numbers, and volunteered the first sizing of AI inside datacom. The answers turned general exactly where the questions were most specific, on the next rack architecture, on the scale of power within AI, and on what a datacom-heavy mix means for capital allocation. Confidence held at the Q2 level; the new element was a cautious Q4 datacom guide delivered without much explanation.

1. IT Datacom Up 30% Sequentially, and AI Sized at Three-Quarters of It

IT datacom grew 100% in dollars and 77% organically and represented 48% of sales, roughly $4.9B on our arithmetic. That is well above the $4.3B our July signpost asked for, and it took the market's share of the company up five points in a single quarter. Sequential growth of 30% was double the mid-teens expectation management set in July, and because CommScope was inside the company for all of Q2 as well, the sequential figure is close to a pure organic reading.

"I would say that the vast majority of our sequential growth was related to AI, which now does represent approximately 3-quarters of our IT Datacom sales."
— R. Adam Norwitt, Chairman and CEO

That sentence is the most important new disclosure of the quarter. Three-quarters of $4.9B is about $3.7B of AI-related revenue in the quarter, roughly 36% of Amphenol, and an annualized run-rate near $14.7B. The remaining quarter of datacom, the base IT business, is about $1.2B a quarter. CommScope's datacom sales are expected to more than double for the year and to be about half of CommScope's total.

"On a sequential basis, sales increased by 30% from the second quarter which was substantially better than our expectation for a mid teens increase."
— R. Adam Norwitt, Chairman and CEO

Assessment: The disclosure confirms the scale of the bull case and puts a number on the concentration bear at the same time. In July we called the absence of any AI revenue line "a real transparency gap"; that gap has now partly closed, and what it shows is that a little over a third of the company rests on AI programs. The franchise is executing extraordinarily well. The question for the multiple is no longer whether AI is large for Amphenol, but whether the market should value roughly a third of the earnings on AI-hardware terms.

2. The Q4 Datacom Guide: Mid-Single Digits After +30%

The fourth-quarter outlook asks IT datacom to grow only mid-single digits sequentially, and the total Q4 guide midpoint of $10.4B is just 1.7% above Q3. Management said nothing was pulled into Q3, that customers are still procuring at high levels, and that AI's seasonality is hard to judge.

"As we look towards the fourth quarter, we expect sales to increase again in the mid single digits from these very strong third quarter levels."
— R. Adam Norwitt, Chairman and CEO

The history of these market-level guides is the relevant context. In January management guided first-quarter datacom roughly flat organically and delivered 16% organic sequential growth. In April it guided low teens and delivered 22%. In July it guided mid-teens and delivered 30%. Each guide has been beaten by roughly eight to sixteen points.

Assessment: We read the Q4 datacom guide as a floor rather than a forecast, but we would not underwrite the full historical gap. The more important point is structural: the gap between the datacom guide and the datacom result has been the main source of Amphenol's beats this year. If datacom simply grows in line with guidance in a future quarter, the company's ceiling-beat pattern shrinks to what the other half of the business can deliver. That quarter will look like a disappointment even if nothing is wrong, and the stock's fade during the call is consistent with a market already watching for it.

3. CommScope Raised a Second Time: $5.1B of Sales and $0.18 of Accretion

The acquired business was raised from $4.6B of 2026 sales and $0.15 of accretion (the July figure of $0.30 before the split) to $5.1B and $0.18. Management noted it was presented as a $3.6B business at signing; it is now $1.5B larger in its first year of ownership. The 10-Q puts CommScope's sales from the January close through September at $3,683.3M, against $2,100.9M through June, which implies roughly $1.58B in Q3, up about 31% from just over $1.2B in Q2.

"We now expect CommScope to achieve $5.1 billion in sales, and to contribute $0.18 to our 2026 adjusted earnings per share."
— R. Adam Norwitt, Chairman and CEO

The same arithmetic shows how conservative the raised figure is. $5.1B less $3.68B leaves about $1.42B for the fourth quarter, a roughly 10% sequential decline from the derived Q3 level. Some of that is the communications networks decline, where CommScope has its broadband exposure. But CommScope's datacom business, which is expected to double this year and be half of its sales, is the fastest-growing piece of the whole company.

Assessment: This raise is smaller in margin terms than July's. Accretion rose 20% on an 11% revenue raise, against 100% on 12% three months ago, which is what convergence looks like once the easy cost actions are captured. The business is still compounding, and the full-year figure again embeds a step-down that its own run-rate does not support. We continue to treat the CommScope guide as conservative; we no longer treat it as a margin surprise.

4. A 29.7% Margin Earned Through Operating Expenses

Adjusted operating margin of 29.7% was up 220bp year over year and down 10bp sequentially, with the sequential dip fully explained by the smaller tariff recovery. The CFO's bridge is precise on the operating line and silent on the gross line.

"On a sequential basis, the slight decrease in the adjusted operating margin was driven by a meaningfully larger benefit from tariff recoveries in the second quarter. Excluding these benefits, our adjusted operating margin improved by approximately 70 basis points, reflecting strong conversion on higher sales levels as well as continued progress on profitability improvement at recent acquisitions, and particularly, CommScope."
— Craig A. Lampo, CFO

Underneath, gross margin fell from 40.5% to 39.3% and SG&A fell from 10.7% to 9.6% of sales. Asked whether the company would lift the 30% incremental-conversion target it raised a little over a year ago, the CFO said not yet, said he does not see 29.7% as a ceiling, and said acquisitions will compress margins as they come in. The CEO described a "challenging cost environment."

Assessment: The quality of the margin is fine and the source of it moved. A 115bp sequential gain in SG&A leverage is not repeatable at this rate, so the gross line has to stabilize for a 30% operating margin to hold through 2027. A sequential gross-margin decline with no disclosed one-time cause, in a quarter with record volume, is the item in this print we would most like management to have explained. We hold our FY2027 margin assumption at about 30% rather than raising it.

5. Orders of $12.17B: A Third Straight Backlog Build Near $1.9B

Orders rose 99% year over year and 13% sequentially to a record $12.17B. Orders exceeded sales by $1.94B, matching Q2 almost to the dollar, after $1.82B in Q1. Across nine months the excess is $5.7B, more than half a quarter of current revenue.

"Orders in the quarter were a record $12.17 billion up a strong 99% compared to the third quarter of 25 and up 13% sequentially. Resulting in another very strong book to bill ratio of 1.19 to 1. This impressive book to bill was driven by robust bookings in most of our end markets with positive book to bill in all but 1 of our end markets this quarter."
— Craig A. Lampo, CFO

A question on order duration drew an answer worth noting: customers have widened their order windows over the past year, partly because the company asks them to share the risk of capacity investments.

"Part of that is you know, in certain cases when we make investments, we try to make sure that our customers are sharing the risk with us, and sometimes that risk sharing has to do with giving us, you know, orders that have a certain extension."
— R. Adam Norwitt, Chairman and CEO

Assessment: The order book is the strongest evidence against the pull-forward bear: it has now reloaded to a record three quarters running after converting each time. Two qualifications. Breadth narrowed slightly, to all but one market from every market. And some of the longer-dated book is customer-underwritten capacity, which is good for durability but means the ratio is not a pure read of near-term demand. Neither changes the conclusion that 2027 starts with a larger backlog than 2026 did.

6. Concentration at 48%, and Management's Answer to It

One question asked directly whether a datacom share heading above 50% should change how the company uses its balance sheet. Management's answer had three parts: diversify within datacom across customers and layers of the stack, keep investing in the other markets, and keep the organization ready to react to a downturn.

"And if you take out, by the way, AI, and the growth in AI, we are driving organic growth in the high teens percentage for the company. So that is a that is a very, very strong performance and is, again, evidence of the fact that we continue to double down across the electronics industry."
— R. Adam Norwitt, Chairman and CEO

The CEO also described the company's practice of driving "with 1 foot on the gas and 1 on the brake," building flexibility into the cost base even in strong periods. He offered no customer-level figure. On our arithmetic, the Q4 guide (datacom up mid-single digits against the company up 1.7%) takes datacom to roughly half of sales next quarter.

Assessment: The answer is strategically right and leaves the arithmetic unchanged. High-teens organic growth outside AI is genuinely strong and is the best mitigant on offer. But the last customer-concentration statement remains January's comment that there were no 10% customers in 2025. The next natural update is the Q4 call and the 2026 10-K, after a year in which datacom gained roughly eight points of mix. That disclosure is now the most important one in the thesis.

7. Four Acquisitions for $1.6B, Mostly Away From AI

The company closed Empire Electronics (automotive interconnect assemblies, about $320M of 2026 sales) at the start of September and signed AdvancedPCB (defense and industrial printed circuit boards, about $270M), Dexgreen (European fiber-to-the-home connectivity, about $90M) and Humanetics Sensors (force and fiber-optic sensing, about $70M). The combined price is about $1.6B for roughly $750M of annual sales, financed with cash and debt. Only Empire is in the guidance.

"it is not lost on me that we generated this quarter $1.6 billion in free cash for the company and we have devoted that $1.6 billion in free cash to 4 acquisitions that roughly cost about $1.6 billion."
— R. Adam Norwitt, Chairman and CEO

Assessment: At roughly 2.1x sales, against Amphenol's own enterprise value of about 5.6x 2026 sales on our arithmetic, these are accretive purchases even before any margin work. All four point away from datacom, which is the capital-allocation half of management's answer on concentration. The scale is modest: $750M of acquired revenue is about 2% of the company, so the deals improve the mix at the margin and do not change it.

8. Cash Conversion Recovered; a Bigger Buyback and Dividend

Operating cash flow was a record $2.08B and free cash flow a record $1.62B. Working capital absorbed $416M against $677M in Q2, and our days-sales-outstanding estimate eased to about 70 from about 71. Capex rose to $457M, 4.5% of sales, above the 3% to 4% range. The company repurchased 3.0M shares for $245M, paid $308M of dividends, repaid $938M of senior notes, raised the quarterly dividend 20% to $0.15, and authorized a new three-year $4B buyback from January.

"Operating cash flow in the third quarter was $2.1 billion or 102% of net income and free cash flow was $1.6 billion or 80% of net income. An excellent result considering the growth we have experienced. From a working capital standpoint, inventory days, day sales out outstanding and payable days were all within our normal ranges."
— Craig A. Lampo, CFO

"But I would say the overall range of 3% to 4% is still probably what I would think about with you know, maybe slightly higher than 4 and certainly times where we are growing pretty quickly."
— Craig A. Lampo, CFO

Assessment: The cash-conversion concern we added in July eased: 80% against 68%, with working-capital absorption down by more than a third while sales grew 17%. Nine-month conversion is still 77% against 95% a year ago, and capex above 4% is now the main drag rather than receivables. The $4B buyback, about $1.3B a year, is roughly 0.7% of the market value annually. It is sized to offset dilution, not to shrink the share count, which rose 0.7% year over year.

9. Defense: 30% Organic and a Multi-Year Replenishment Cycle

Defense grew 42% in dollars and 30% organically, 12% sequentially, and is guided up mid-single digits in Q4 and more than 35% for the year. Two separate questions asked about durability. Management described a dozen tracked defense applications, nearly all growing double digits, across space, land, unmanned vehicles, communications, avionics, munitions and naval, plus a growing roster of new defense customers and rising budgets in NATO countries, South Asia and the Middle East.

"At the same time, there is no doubt that we are entering into a phase where there will be a significant replenishment requirement, and that is a multiyear, if not multi decade, initiative in multiple countries around the world."
— R. Adam Norwitt, Chairman and CEO

Assessment: Defense has now grown more than 20% organically for five consecutive quarters, and the replenishment argument is a reasonable answer to the obvious objection that conflict-driven demand reverses. Management also conceded the company would not be immune to a sharp fall in military spending. At 8% of sales and a segment margin of 30%, defense is the cleanest non-AI earnings stream Amphenol has.

10. Industrial: 25% Organic, Some of It Data-Center Adjacent

Industrial grew 60% in dollars and 25% organically, the CEO's "strongest quarter we have had in quite some time," and 7% sequentially against a roughly flat expectation. The growth was spread across building automation (including CommScope's building-connectivity business), instrumentation, oil and gas, alternative energy, electrification, factory automation and heavy equipment, across all three regions.

"And the vast majority of those are growing in double digits organically right now."
— R. Adam Norwitt, Chairman and CEO

"I think there is some piece, but it is certainly not the predominance or it is not dominating the growth."
— R. Adam Norwitt, Chairman and CEO

Assessment: The second quote answers whether data-center construction is inside the industrial number, and the honest answer is yes, partly. That matters for the diversification pillar: some of the "non-AI" growth shares a driver with the AI build. Management's view that it is a minority of the growth is plausible given the breadth of sub-markets named, but it is unquantified, and it means the true AI sensitivity of the company is somewhat higher than the 36% figure.

11. Communications Networks: −28% Organic, and Another Leg Down in Q4

Communications networks was 8% of sales and grew 25% in dollars only because of CommScope. Organically it fell 28%, against −6% in Q2, on lower broadband spending and softer demand from other network operators. The 13% sequential decline was in line with guidance, and Q4 is guided lower again.

"On an organic basis, however, sales declined by 28% from prior year. Due in particular to moderations in demand from the broadband market as well as from other communication network operators."
— R. Adam Norwitt, Chairman and CEO

"Looking to the fourth quarter, we expect a further decline in sales from these third quarter levels on continued softness in spending, particularly in broadband."
— R. Adam Norwitt, Chairman and CEO

Assessment: Our July signpost asked this market to absorb its guided decline without a further leg down. Q3 met the guide; the Q4 outlook is the further leg down, so we grade the signpost as missed. The dollar impact is small, roughly $80M a quarter for each 10% decline, and the buying of Dexgreen into European fiber-to-the-home is a small, deliberately contrarian bet. It is still the one market where the direction of travel is clearly negative.

Guidance & Outlook

The fourth-quarter guide is the arithmetic sum of the market commentary, as it was in July. Applying our reading of each market's sequential outlook to its approximate Q3 dollars (datacom up 5%, mobile up 35%, defense up 5%, commercial air flat, industrial down about 5%, automotive down about 2%, communications networks down about 10%) gives roughly $10.4B to $10.5B. Mobile's seasonal 35% step-up adds about $180M and datacom's mid-single-digit gain about $250M; industrial, automotive and communications networks subtract about $190M together. The guide therefore contains no general cushion beyond what sits inside each market's own outlook, and the biggest of those is datacom.

Implied ramp: The full-year range less nine-month sales of $26,607.9M implies Q4 sales of $10.29B to $10.49B, and the full-year EPS range less nine-month adjusted EPS of $2.01 implies $0.81 to $0.83, so the two guides are internally consistent. Full-year 2026 at $37.0B would be 60% above 2025's $23.1B; Q4 at the midpoint annualizes to $41.6B and $3.28 of EPS before any 2027 growth.

Street at: Consensus entering the print was about $9.80B to $9.83B and $0.76 for Q4, so the guide sits 6% and 8% above it. Full-year EPS consensus near $2.69 moves to the $2.82 to $2.84 range by construction. As in July, the guide-versus-Street gap is close to the size of the quarter's own beat, so forward revisions should be at least as large as the Q3 surprise.

Guidance style: Conservative, and now measurably so over four quarters. Revenue has exceeded the guide ceiling by 5.6%, 8.9%, 6.8% and 8.8%, and adjusted EPS by 6.6%, 14.0%, 16.4% and 12.7%. Applying the four-quarter average revenue beat to the Q4 ceiling would put the quarter near $11.3B. We would not model that; the history is mostly datacom beating its own guide, and Section 2 explains why that gap is the part least likely to persist. We model Q4 at about $10.6B and $0.84, a modest beat.

Analyst Q&A Highlights

What Drove the Second CommScope Raise, and Whether Fiber Supply Can Keep Up

The opening question paired the raise with supply risk: the acquired business depends on optical fiber in a market widely described as tight. Management credited execution against demand in data center and industrial, and said the business had diversified its suppliers and secured supply, glass included, for the long term. The answer also framed the raise against the deal's history: a $3.6B business at signing, $4.1B at close, $5.1B now.

Q: "I wanted to start with CommScope. Adam, maybe if you could just give us some insight into the guidance raise this quarter and then bigger picture, I would be really interested in your thinking on the virtues of their business model and the current tight supply chain environment and maybe within that, if you could also just comment on the work you are doing to secure fiber supply and your flexibility in that respect."
— Luke Junk, Baird

A: "Obviously, when we announced the deal, we talked about this company, you know, at signing as a $3.6 billion company. And here we are now $1.5 billion more than that this year… I think the team has done a fabulous job to diversify their supply base, to secure supply of everything that they need, including glass, and to make sure that the company is positioned to support their customers not just here in the third quarter and the fourth quarter of 26, but for the long term."
— R. Adam Norwitt, Chairman and CEO

Assessment: A complete answer on supply, and the second consecutive quarter in which management denied any binding constraint while delivering growth that proves it. The more useful detail was the explanation of why ownership matters: the prior parent's financial condition had constrained capital spending, and the business now invests freely. That supports the view that CommScope's growth is partly self-funded capacity rather than borrowed demand.

Whether a Shift From Backplane to a PCB Midplane Erodes Content

A question named a specific next-generation rack architecture expected around 2028, in which the cabled backplane gives way to a printed-circuit-board midplane, and asked what that does to content. Management declined to discuss any customer or architecture and restated the principle that every customer it meets needs more copper, optical and power interconnect regardless of design.

Q: "So as we shift from Oberon to Kyber theoretically in 2028, we lose the backplane and then move into that PCB midplane structure. How do you think about that from an all content standpoint?"
— Varun Govindaraj, Bernstein

A: "I do not discuss specific customers, and, you know, I am not gonna delve into specific customers specific architectures… They need more interconnect period. And that more interconnect means more copper high speed interconnect. It means more optical interconnect. It means more power interconnect."
— R. Adam Norwitt, Chairman and CEO

Assessment: The second quarter running in which a backplane-to-board question received a principle rather than an answer. The principle is credible, and the pending AdvancedPCB acquisition adds circuit-board capability (for defense and industrial customers), but the margin question we flagged in July (whether a format shift preserves the content dollar at a lower mix) remains unaddressed. With roughly a third of revenue now AI-related, architecture transitions are a 2027 to 2028 modeling item, not a footnote.

Whether the Datacom Upside Was Demand or Execution, and What the Q4 Guide Means

A question asked whether the 30% sequential datacom gain came from customers raising demand or from the company shipping existing demand faster, and how to read the mid-single-digit Q4 guide. Management said the growth was neither pulled forward nor gated by capacity, that customers continue to procure at high levels, and that AI seasonality is difficult to judge.

Q: "Just wondering, the upside and strength that you are seeing over yours, is that coming from customers essentially increasing their demand to you? Or was it more driven by your ability to fulfill existing demand faster than you had expected?"
— Amit Daryanani, Evercore

A: "You know, hard to say what the seasonality of AI will be. We are still guiding to a growth in the fourth quarter from these very high levels… And I think you know, strength in the third quarter it was not that anything was pulled in. It was just we satisfied our customer demand, and we were able to execute upon it."
— R. Adam Norwitt, Chairman and CEO

Assessment: The answer did not choose between the two explanations the question offered, and that is itself informative: management is not claiming accelerating end demand, only that it met the demand that existed. Combined with the cautious Q4 guide, it is the most measured datacom commentary of the year. We read it as discipline rather than warning, but it is the exchange that best explains the stock's fade during the call.

Whether Incremental Margins Can Run Above the 30% Target

With the consolidated margin now nearly equal to the incremental-conversion target the company raised a little over a year ago, a question asked whether that target would rise, and cited input-cost pressures in the industry. The CFO declined to change the target, said the current margin is not a ceiling, and warned that acquisitions will compress margins as they arrive.

Q: "Overall, non GAAP EBIT margins are nearly 30%, and that is actually the incremental conversion margin the company has historically targeted. I am sure you would like to drive profitability higher going forward, but maybe talk a little bit about the ability for the company to take up its year around incremental margins north of the historical 30% target, especially in light of some of the input cost dynamics that are at work in the industry?"
— Mark Delaney, Goldman Sachs

A: "So I do not think we would really change our per se long term target yet. I am not saying that will not happen kind of in the future here… And I certainly do not think this 29.7% is kinda the bottom in terms of where we are today."
— Craig A. Lampo, CFO

Assessment: A sensible refusal, and the "yet" matters. The input-cost part of the question went unanswered, and it is the part that connects to the gross-margin decline in this quarter. We take the answer as support for holding a roughly 30% operating margin in 2027, not for modeling expansion beyond it.

Longer-Dated Backlog and Exposure to Smaller AI Cloud Builders

One question concerned the order book: whether its lengthening duration exposes the company to smaller AI-focused cloud providers that might struggle to fund their builds, and whether Q3 drew on Q4 through large project ramps. Management said its exposure to those providers is not large, that most demand comes from the most highly capitalized companies in the world, and that it sees no project-based lumpiness.

Q: "How concerned should we be on this further dated backlog on the potential that perhaps NeoCloud build outs might suffer from funding concerns? And would you say that there was any lumpiness or timing in demand in Q3 at the expense of Q4 given some very material project ramps."
— Wamsi Mohan, Bank of America

A: "I mean, the vast majority of the demand that I think we are seeing, that our customers are seeing, are from some of the world's most highly capitalized companies who are doubling down on creating new functionalities, new capabilities, and new capacity of compute for their customers… We are not really seeing project based lumpiness."
— R. Adam Norwitt, Chairman and CEO

Assessment: A direct answer on counterparty quality, and the one most relevant to the financing-risk version of the AI bear. It does not quantify the smaller providers' share, so "not an enormous exposure" is a qualitative assurance. Read alongside the customer risk-sharing comment on order duration, it suggests the longer book is anchored by well-funded buyers, which is the reassuring version.

Whether a Datacom Share Above 50% Should Change Capital Allocation

The most direct question on concentration noted that the company used to describe itself as having no market above 20% of sales, and asked whether a datacom share heading above half should change how the balance sheet is used. Management said the growth happened for good reasons, described diversification within datacom across customers and the supply chain, and pointed to continued investment in every other market.

Q: "But with IT Datacom poised to be, like, 50% plus of the company going, into next year, does it force you to reexamine balance sheet usage? Do you need to scale up in other places to kind of ward off potential as cycles ultimately happen in these businesses?"
— Joe Giordano, TD Cowen

A: "So that diversification within IT data is something that we are heavily, heavily focused on, and I am very proud of what we have achieved there so far. But at the same time, we will continue to double down on the other markets in Amphenol."
— R. Adam Norwitt, Chairman and CEO

Assessment: The question was the right one and the answer was candid in tone, light on numbers. Diversification within datacom is real at the product level, but the end buyers are a short list of very large spenders, and no customer-level disclosure was offered. The $1.6B of non-datacom acquisitions is the concrete part of the response. This is the exchange we will grade against the next customer-concentration disclosure.

The Path for CommScope's Margin to the Corporate Average

A question asked what is driving CommScope's earnings upside relative to its sales and how long convergence to the corporate margin will take. The CEO rejected the word "synergies" and credited the acquired team's ownership mindset and access to capital. The CFO gave the number: CommScope is in the low 20s, already at the corporate margin of a couple of years ago, with more confidence now than at the start of the year that it reaches the current average over time.

Q: "And as we take a step back, how should we think about the path for CommScope margins to reach the corporate average? What are the main moving pieces now? How are you thinking about the time line there just given the execution we have seen today?"
— Joseph Cardoso, JPMorgan

A: "I would say, actually, you know, right now, they are already kind of in the low 20s on an operating margin basis. So they are already at what the corporate average was as of a couple of years ago or less than a couple years ago."
— Craig A. Lampo, CFO

Assessment: "Low 20s" against "over 20%" in July is progress, but modest progress, and no timeline was given. A gap of seven to eight points to a 29.7% corporate margin on a $5B-plus business is the largest single source of margin upside left in the model, and the slower pace of this raise suggests it will take years rather than quarters.

How Large Power Is Within the AI Business

A question asked management to size power interconnect within AI and the pace of the move to 800-volt designs. Management described power as a long-standing capability and a high-growth opportunity, called it a very meaningful part of AI sales, and said it is smaller than high-speed copper and optics, without giving a figure.

Q: "But curious how sizable power is now relative to your overall AI business and how fast are you seeing that transition to higher voltage designs?"
— Shreyas Patel, Wolfe Research

A: "In terms of the scale of our power business, and we have a very significant offering in power I would tell you that, you know, across our AI business, you know, without giving, know, specific decimal level numbers, I mean, it is a very meaningful portion of what we sell."
— R. Adam Norwitt, Chairman and CEO

Assessment: A ranking rather than a size: power is meaningful and third of three. Having sized AI within datacom for the first time, management chose not to go further on the mix inside AI. The ranking still helps, because it tells a modeler that the architecture risk on high-speed copper applies to the largest piece of the AI business, not to a niche.

What They're NOT Saying

1. No customer-concentration update, even when asked: The most direct concentration question of the call drew a strategic answer and no customer-level figure. The 10-Q filed two days later contains no customer-concentration statement either; we searched it for any 10%-customer language. With datacom at 48% of sales and AI at about 36%, the January comment that no customer exceeded 10% of 2025 sales is the only data point available, and it describes a smaller AI business.
2. Gross margin was never mentioned: Gross margin fell 125bp sequentially to 39.3% in a record-volume quarter, and the words "gross margin" do not appear in the call. The 10-Q discusses operating expenses as a share of sales, not the gross line. An analyst question that cited industry input-cost pressure got an answer about the conversion target instead. If the decline is mix, as we believe, it was an easy point to make, and management chose not to make it.
3. The implied CommScope step-down went unexplained: The raised $5.1B full-year figure, set against the $3,683.3M the 10-Q reports through September, implies about $1.42B for Q4, roughly 10% below our derived Q3 figure of about $1.58B. Nobody asked, and management did not volunteer whether this is broadband weakness, conservatism, or something specific to the fourth quarter.
4. Backlog is still not quantified: Orders have exceeded sales by $5.7B in nine months. The 10-Q again uses the one-year practical expedient and does not disclose remaining performance obligations, so the size and duration of the book, which management says has lengthened partly through customer risk-sharing, can only be inferred from book-to-bill.
5. Foreign tax accruals keep arriving and are not discussed: The 10-Q confirms the $230M China assessment was paid in full in Q2, which closes the assessment itself; the separate $160M reassessment of prior-year rates is an accrued obligation whose payment the filings do not describe. Q2 also added a $39.0M reserve for "potential settlement of various foreign tax matters" and Q3 a $60.0M charge for "settlements of foreign tax matters," neither attributed to a jurisdiction and neither mentioned on the call. Nine-month discrete tax items excluded from adjusted EPS now total $389.0M. Our July note said Q2 took no new accrual; it took the $39.0M reserve, and we correct that here.
6. The mix inside AI stayed private: Management sized AI within datacom for the first time, then declined to size power ("without giving, know, specific decimal level numbers") and gave no split between high-speed copper and optics. It is a reasonable line to hold commercially, and it leaves the architecture-transition risk unmodelable from outside.
7. The market with a negative book-to-bill was not named: Book-to-bill was positive "in all but 1" end market, after two quarters in which every market was positive. Management did not say which, and no analyst asked.
8. Communications networks drew no questions: The only organically declining market fell 28% organically, is guided lower again, and is the destination of one of the four new acquisitions, yet none of the call's questions addressed it. That is two consecutive calls in which the weakest market went unexamined.

Market Reaction

• Pre-print setup: The stock closed at $78.21 on October 27, up 15.7% year to date against the S&P 500's 12.8%, up 13.9% over twelve months, and down 7.6% over the trailing 30 days from $84.62. It sat 11.7% below its 52-week closing high of $88.62, set on October 6. The setup included an October 22 session in which the shares fell 7.7% to $73.07 on a day the S&P 500 fell 1.2% and AI-hardware names fell harder (Corning −8.1%, Vertiv −7.8%, Celestica −11.7%), followed by a 7.0% recovery over the three sessions into the print. Options were pricing a move of roughly ±11%.
• Reaction session (October 28; results before the open, call at 1:00 PM ET): Indicated up roughly 7% before the open, the stock opened at $82.88 (+6.0%) and printed the session high of $83.73 (+7.1%) in the first five minutes. It was back at $79.60 within ten minutes, rebuilt to $82.53 by 12:50 PM, then slid during the call to the session low of $79.30 at 1:35 PM. It closed at $80.96, up 3.5%, on 20.4M shares, 1.7 times the 30-day average. The S&P 500 rose 0.3%.
• Peer reaction: On the same session Vertiv rose 16.5%, Celestica 8.4% and Corning 3.9%, while TE Connectivity fell 1.8%. Amphenol finished in the middle of the AI-hardware group.
• Next-day sessions: The shares added 1.4% to $82.08 on October 29 against a 0.3% decline in the index, then fell 3.4% to $79.30 on October 30, when the S&P 500 fell 1.8% and AI-hardware peers fell 4% to 6% (Celestica −6.1%, Vertiv −5.5%, TE Connectivity −4.4%, Corning −4.3%). The three-session move from the pre-print close was +1.4%.
• Since our last note: $79.30 against the split-adjusted $80.35 at which we wrote in July, a 1.3% decline across a quarter in which our FY2026 EPS estimate rose about 9% and our FY2027 estimate about 7%.

Two Sell-Offs Inside One Up Day

The reaction session contained two distinct episodes of selling. The first came at the open: a 6% gap was cut to under 2% within ten minutes. The release, which was public before the open, already carried the Q4 total guide of $10.3B to $10.5B, just 1.7% above the quarter just reported, and a market that had watched Amphenol grow 15% to 17% sequentially in each of the two prior quarters could read that as deceleration before hearing a word of commentary. The second came during the call. The stock had recovered to $82.53 by 12:50 PM, and it fell to $79.30 over the following 45 minutes, with the slide's heaviest five-minute volume at 1:20 PM. The timing fits the point in the prepared remarks where management walked through each market's Q4 outlook and gave the mid-single-digit datacom guide, though we cannot prove the link from price data alone. The shape repeated July, when a 5.9% gap closed at 4.5%.

The Stock Now Trades as AI-Hardware Beta

The biggest moves of the month were sector days, not company days. The 7.7% drop on October 22 and the 3.4% drop on October 30 both came with the S&P 500 down and AI-hardware peers down by similar or larger amounts. The realized earnings-day move of 3.5% was about a third of what options had priced. In other words, a quarter that beat on every line moved the stock less than the sector's mood did the week around it. That is the market's own verdict on the concentration point in this report: at 48% datacom, the shares are being priced as part of the AI-infrastructure complex, and the stock-specific information in a print matters less than the group's direction.

The Estimates Moved, the Price Did Not

For a third consecutive quarter the forward multiple compressed without the price falling materially. Against our July note, and on the same earnings years, the shares have gone from 30.6 times to 28.0 times FY2026 EPS and from 24.2 times to 22.3 times FY2027. The cause is entirely the numerator: the 2026 guide midpoint of $2.83 sits 7.8% above the $2.63 we carried in July, and our FY2027 estimate rises from $3.33 to $3.55. We read the flat price as the sector discount widening rather than as the market doubting the numbers, which is the version of the setup we said in July we would want to buy.

Street Perspective

Debate: Is the Mid-Single-Digit Q4 Datacom Guide a Floor or the First Sign of Deceleration?

Bull view: The bull case on the Street holds that every datacom guide this year has been beaten by eight points or more, that orders of $12.17B at a 1.19 book-to-bill on a $10B base are not the profile of a slowing market, and that management explicitly said nothing was pulled forward. On this view the Q4 guide is simply the company's habit of guiding to visibility.

Bear view: The bear camp contends that sequential growth falling from 30% to mid-single digits is deceleration by definition, that the total company guide implies under 2% sequential growth, and that 2027 comparisons against a datacom business that doubled in 2026 will make year-over-year growth rates fall sharply whatever the absolute level does.

Our take: The guide is a floor, and the order book says 2027 starts from a higher base than 2026 did. The bears are right that the growth rate will fall and wrong to treat that as the same thing as demand falling. The practical risk is that a quarter in which datacom merely meets its guide will look like a miss, because the datacom gap has been most of the beat. We would buy a guide-driven sell-off as long as book-to-bill stays above 1.0.

Debate: AI-Hardware Stock or Diversified Compounder?

Bull view: Some desks argue that the company grows organically in the high teens even without AI, that five of seven markets are growing above 20% organically, that defense and industrial run on multi-year cycles, and that the acquisition program keeps adding non-AI revenue at accretive prices. That is a premium-compounder profile and deserves a compounder multiple.

Bear view: The skeptical view is that AI is now about 36% of sales, datacom about half, and some industrial growth is tied to data-center construction, so the true AI sensitivity is higher still. The stock trades with the AI-hardware group on sector days, there is no customer-level disclosure, and the multiple should reflect a cyclical capex supplier.

Our take: Both describe the same company at different points in the cycle. Today the earnings have diversified-compounder quality and the share price behaves like AI beta. Our 29x base multiple is deliberately a blend of the two. The bears' strongest point is disclosure, not fundamentals: until the company says how large its largest customers are, the market will keep pricing the concentration at its worst case.

Debate: Is There Still Upside After a Flat Quarter on Rising Estimates?

Bull view: A growing consensus view holds that estimates rose roughly 7% to 8% while the stock fell, that post-print targets cluster around the low-to-mid $110s, and that 22 times next year's earnings for a business growing EPS more than 25% is not demanding.

Bear view: The bear camp contends that 28 times the current year on peak datacom mix is a full multiple, that the stock gets cheaper only because earnings are compounding at a rate that cannot last, and that sell-side targets implying 32 to 33 times next year's earnings assume the cycle never turns.

Our take: The arithmetic favors the bulls for a third straight quarter, and on FY2027 the stock is cheaper than at either of our two previous readings (26.7x in April, 24.2x in July). It is still not cheap, and we would not underwrite the Street's multiple. At 29 times our estimate the shares have about 30% upside, which is enough for an Outperform without needing the multiple to expand.

Model Update Needed

Valuation impact: Rolling the framework forward on the revised numbers, against the October 30 close of $79.30, and keeping July's scenario multiples unchanged:

The base case implies about 30% upside with a +64% / −16% skew, wider and more favorable than July's +54% / −20% because estimates rose and the price fell. Our base sits below the post-print Street target cluster in the low-to-mid $110s, which implies 32 to 33 times our FY2027 estimate. As in July, we would rather underwrite 29 times and be surprised. The bear case is the one to respect: it requires only that AI demand pause for a few quarters, and at roughly half of sales that is no longer a tail scenario for this company.

Thesis Scorecard Post-Earnings

Q2 2026 Signposts Revisited

Standing Thesis

Overall: Thesis strengthened modestly. All four bull pillars were confirmed, valuation eased for a third consecutive quarter, cash conversion improved, and the China assessment has been paid. Five of the eight July signposts were met outright, two partly, and one, communications networks, was missed. The concentration bear moved against us for a second straight quarter: datacom rose five points of mix, AI was sized at about a third of the company, and the stock's behavior on sector days shows the market is pricing that exposure.

Action: Maintain Outperform at conviction 8. In July we said we would not chase above the low $170s and would add on a datacom-cadence scare that left the order book intact; split-adjusted, that ceiling is the mid-$80s. The shares are at $79.30 after a sector-wide sell-off and a cautious Q4 datacom guide, with the order book at a record. That is close to the setup we described, and we would add here.

Bottom Line: A Third Quarter of Rising Estimates and a Flat Price

Rating decision: we maintain Outperform on Amphenol with conviction unchanged at 8. The first $10 billion quarter in the company's history beat its own ceiling by $830M and $0.09, datacom grew 30% sequentially, orders set a record for a fourth straight quarter, cash conversion recovered, and CommScope was raised for the second time. Five of our eight July signposts were met, and the one clear miss is the only market declining organically, at 8% of sales.

The arc of this coverage has been an argument about the distance between a great business and a fully priced stock. We initiated at Hold on price, upgraded when the AI durability was proven, held through the January flush, and in July noted that the stock had become cheaper while rising. This quarter goes one step further. The price is where it was in July and the earnings are about 8% higher, so the forward multiple compressed a third time. On FY2027 the shares now trade at 22.3 times our estimate, against 26.7 times in April and 24.2 times in July.

What keeps this from a maximum-conviction call is the same thing that kept it there in July, only larger. IT datacom is 48% of sales and heading to about half, AI is roughly 36% of the company and some industrial growth is data-center adjacent, the stock trades with the AI-hardware group, and the only customer-concentration statement describes 2025. The non-AI half of the business is growing in the high teens organically with 30% margins in defense and harsh-environment markets, and that is a real mitigant. It is not an answer to what a pause in AI spending would do to a company with this mix, and the Q4 datacom guide is the first hint of what a normal quarter will look like.

What would move us to Hold (July's triggers, unchanged): a genuine sequential decline in organic IT datacom with book-to-bill falling below 1.0; a second and third quarter of deteriorating free-cash-flow conversion without a working-capital explanation that resolves; a CommScope margin path that stalls below the corporate average; or a re-rating back toward 32 times forward without further estimate revisions. None fired this quarter: datacom grew, book-to-bill was 1.19, conversion improved, CommScope progressed from above 20% to the low 20s, and the multiple compressed to 22 times. We add one trigger openly, before the disclosure it depends on: if the next customer-concentration disclosure shows any customer above 15% of sales, we move to Hold unless book-to-bill and non-AI organic growth both hold; a customer between 10% and 15% would cost at least one point of conviction.

What would move us to Underperform (unchanged): a hyperscaler capex contraction driving outright organic datacom declines and order cancellations; a margin reversal that establishes 29.8% as a cycle peak rather than a step; or a re-escalation of the China tax matter into a recurring liability. None fired. The China assessment is paid, and we are watching the separate run of foreign-tax accruals for any sign that it is becoming the recurring item this trigger describes.

Signposts for Q4 2026 earnings (late January 2027):

1. Q4 lands at or above the $10.3B to $10.5B and $0.81 to $0.83 guide, a fifth straight quarter above the ceiling on both lines.
2. IT datacom grows at least mid-single digits sequentially, and the first-quarter 2027 outlook for datacom is not a decline.
3. Book-to-bill at or above 1.0 for a fifth consecutive quarter.
4. Gross margin stabilizes at or above 39% with adjusted operating margin at or above 29.5%.
5. CommScope finishes at or above $5.1B, which requires only about $1.42B in Q4 against about $1.58B in Q3.
6. Free-cash-flow conversion of at least 85% in Q4, taking the full year to about 80%.
7. A customer-concentration statement for 2026, on the call or in the 10-K.
8. The communications networks decline narrows in the first-quarter 2027 outlook.
9. AdvancedPCB and Humanetics Sensors close on schedule and enter the guidance.