Amphenol makes the links that carry signals and power
Connectors
plugs and sockets joining parts of a machine
Cables and fibre
inside AI racks, jets and cars
Sensors and antennas
for cars, factories and phones
Every electronic machine needs links between its parts. Amphenol makes them for AI servers, jets, cars and factories. Equipment makers choose its parts years before a product ships.
The company · 2 · What exactly does it sell?
Communications parts, mostly for AI racks, are over half
Each $1 of sales, twelve months to June 2026
57%23%19%
Data centres and communicationsJets, defence and tough settingsCars, factories and sensors
It is paid per part shipped. Customers design Amphenol's parts into their products, then buy them for the product's whole life.
The company · 3 · How does it make money?
It keeps 27 cents of each sales dollar, up from 21
34 → 39 centskept per $1 after making the parts, since 2024
12 → 11 centsoffice and selling cost per sales dollar
About 4%of sales spent on factories, the one cost rising
Sales nearly doubled from 2024 while office costs fell per dollar, so profit per dollar rose from 20.7 to 27.0 cents. Every business kept more of each sale.
Is it a good business? · 4 · What gives it an edge?
Chosen at the drawing board, hard to swap later
82.1%
of connector sales went to the top 100 makers in 2025
TE and Corning
name Amphenol in their own filings
Swapping a supplier means retesting a product already built, so customers rarely do it once a part is designed in.
Is it a good business? · 5 · What is the catch?
Rivals compete for each socket, and light may replace copper
TE, Molexcompete for the same sockets
Corningcompetes in fibre
NVIDIAships switches with light built onto the chip
Optical links built onto the chip could replace some copper cables and plugs. One forecaster expects copper to stay inside racks to 2031, with only a small share moving.
Is it a good business? · 6 · What does its record show?
Sales tripled in four years as AI took off
Sales, $B
10.9
12.6
12.6
15.2
23.1
33.0
Dec 2021Dec 2022Dec 2023Dec 2024Dec 2025last 12 mo
Sales, third quarter+65%37% without companies it bought
New orders$12.2Bmore than it shipped
Profit after running costs29.5 centsper sales dollar
Nearly all recent growth came from AI products. Part of the record profit per dollar was helped by a one-off tariff refund.
Is it a good business? · 7 · How much room is there to grow?
A connector market of about $99B, growing fast
$99.2Bof connectors bought worldwide in 2025, up 14.7%
About 31%data-centre capacity growth expected in 2026
Amphenol's sales equal about a quarter of that figure, though it also sells cables and sensors the figure leaves out.
Is it a good business? · 8 · Who is in charge?
A chief since 2009, paid on sales and profit growth
Since 2009Adam Norwitt as chief executive
About 1.4%owned by directors and officers
200%bonus payout for 2025, on sales and profit-per-share growth
Insiders sold outside pre-set plans after using their options, their usual habit here. It borrowed about $10.5B to buy CommScope's business.
The story right now · 9 · Why is demand rising so fast?
Every AI rack is full of Amphenol's cables and plugs
About 31%expected growth in data-centre capacity in 2026
30%last quarter's rise in data-centre sales; mid-single digits expected next
AI chips must pass huge amounts of data, so racks are packed with high-speed cables and connectors. Amphenol sells more of them than anyone. Its own forecast says the pace of growth is slowing.
The story right now · 10 · What happens if AI spending pauses?
A pause would cut growth and profit at once
Customers order aheadOrders run above shipments.
Factories expand to meet themCosts rise in advance.
Spending pausesOrders can be cancelled.
Profit falls faster than salesCosts stay; sales drop.
Amphenol is the best in its trade, but the price assumes the AI build carries on without a pause. A book-to-bill below one would be the first warning.
The story right now · 11 · What changed lately?
Sales rose 65% and orders ran ahead of shipments
+65%third-quarter sales
$12.2Bnew orders, more than it shipped
29.5 centsprofit per sales dollar, a record
Solid quarter
The shares hit a record in October, yet got cheaper per dollar of sales over the year because sales grew faster than the price. Next quarter's data-centre rise should slow.
The story right now · 12 · What is everyone arguing about?
Does the price assume the AI boom never pauses?
Many investors think
The AI connector winner, still reasonably priced
We think
An excellent business, priced for the build lasting
We are wrong if orders keep growing above 30%, without bought companies, through 2027. Then the price is fair and the business keeps compounding.
The price and our view · 13 · What is a share worth?
The price sits close to our value
$79.67what we think a share is worth · the last close is of that
$79.67our value
Attractive below $63.74
Neutral
Unattractive above $95.60
$35.72gloomy: growth halves and profit per sale slips
$79.67our estimate: experts' forecasts, then slowing
$112sunny: growth runs a quarter faster
Value per share in three futures. None is a prediction of the share price.
We estimate what a share is worth from the cash the business should earn over many years. Our rule: buy only when the price is at least a fifth below our estimate, as a cushion for being wrong.
The price and our view · 14 · What if it all works?
As a bet on data centres, it is fragile
Priced for successthe kind of bet this is
0.8×if it works
49%kept if it fails
Multiples of the company's value when the bet was last struck, 16 Oct 2026.
There is little extra gain if AI demand holds. If the build stalls, the rest of the business valued like rivals leaves about half of today's value once debt is paid.
The price and our view · 15 · How could we be wrong?
A few big spenders, cancellable orders, record profit
AI demand could pauseBiggest
Nearly all recent growth came from AI products.
Record profit per saleShowing now
Helped by a one-off tariff refund.
Light on the chipWatch
Could replace some copper links.
Who the big buyers areUnknown
How much goes to the few largest.
Over a third of its factory assets sit in China.
The price and our view · 16 · How will we know?
Late January shows whether the AI build is pausing
Sales of $10.3B or more, with orders above salesGood sign
A 2027 outlook above the fourth quarterGood sign
New orders fall below what it shipsWe turn negative
JAN 202728
Fourth-quarter and full-year 2026 results. Companies report every three months; each report is a test of the story.
Our view also changes on price alone: we turn positive if the shares fall a fifth below our estimate, and negative if they rise a fifth above it.
The price and our view · 17 · So, what do we think?
The best maker of AI's links, priced for no pause
$79.67our value
Attractive below $63.74
Neutral
Unattractive above $95.60
NeutralMedium conviction
Amphenol is excellent and keeps more of each sale every year. But nearly all its growth is AI, its orders can be cancelled, and the price assumes the build never pauses. Fair, not cheap.
↑ ↓ to move · C for contents
What it does
More about the company
Who it is
Every electronic system, from a fighter jet to a car to an AI server, needs links that carry signals and power between its parts without failing. Amphenol makes those links: connectors, cable assemblies, antennas, sensors and fibre. It employs about 170,000 people.
Its buyers are equipment makers, which design its parts into a product years before it ships. In AI racks, NVIDIA's own systems run their chips over a copper cable backplane. Rivals TE Connectivity and Corning each name Amphenol among their main competitors.
How it makes money
Amphenol is paid for each part it ships, mostly on short orders. Its biggest segment, Communications Solutions, sells to data centres, phone makers and telecom networks and earned about a third of each dollar of sales as profit in the second quarter. The other two sell to defence, aircraft, factories and carmakers.
No single customer is a tenth of sales, and about a fifth goes through distributors.
Growth and profit
What growth costs
Growth has made Amphenol more profitable. Since 2024 selling and admin costs fell from about 12 to 11 cents per dollar of sales, while the gross margin rose from about 34% to 39% and every segment's margin widened. The cost that is rising is factories: spending on plant has reached about 4% of sales.
Its record
The numbers
Is it growing?Fast
Sales grew 58% over the twelve months to Sep 2026, faster than the 52% in the year to Dec 2025.
Is it profitable?Strong
Operating margin is 28%, up from 20% in the year to Dec 2023.
Does growth pay off?Pays off
Running costs take 11¢ of each sales dollar, about where it was (12¢) in the year to Dec 2021.
Does it earn good returns?Strong
Each dollar put into the business earns 23% a year after tax, up from 16% in the year to Dec 2021.
Is it making cash?Strong
It turned 15% of sales into free cash, $5.12B over the last twelve months.
Is the balance sheet safe?Light debt
It owes $12.6B more than the cash it holds, about 1.9 years of the cash its operations bring in.
Is profit turning into cash?Partly
Free cash flow was 0.9 times net income.
Why it wins
The evidence
A connector is designed into a customer's product at the drawing board and tested with it, so swapping suppliers means testing a product that is already built. Customers have been cutting their supplier lists to the few they trust, and Amphenol's design teams often win approved-vendor status on new programmes.
Outside witnesses agree it is a leader: TE and Corning name it in their own filings, and an industry tracker finds the largest makers gaining share as they buy smaller ones.
Who could beat it
The rivals in full
TE Connectivity and the private Molex compete for the same sockets, and Corning for fibre. The newer threat is optics moving onto the switch chip itself, which NVIDIA has begun to ship and which could replace some copper and plug-in links. One forecaster expects copper to stay inside racks to 2031, with only a small share of links moving.
some pluggable and copper links in AI clusters, by moving the optics onto the switch chip
Molexprivate
high-speed copper connectors and cable assemblies in servers and AI racks
How big it could get
The market and who has signed
The market
The world bought about $99.2B of connectors in 2025, a market that grew 14.7% that year. Amphenol's sales equal about a quarter of that, though it also sells cables and sensors the figure does not count. The fastest part rides AI data centres, whose power capacity TrendForce expects to grow about 31% in 2026 and above 30% in 2027.
Who has signed
Nobody has signed a long contract. At the end of 2025 Amphenol had about $8.9B of orders waiting to ship, nearly all within a year, and in the second quarter it took $10.7B of new orders. Customers may generally cancel orders before they ship.
Who runs it
Leaders, pay and the cash
Who runs it
Adam Norwitt has run Amphenol since 2009 and became chairman in 2026; Craig Lampo has been finance chief since 2015. A new head takes over the harsh-environment segment in January 2027.
Directors and officers own about 1.4% of the shares. Bonuses pay on sales and earnings growth, and long-term pay is options. Insiders sold shares outside pre-set plans after exercising options, which is their usual practice here.
What they did with the money
Amphenol buys companies all the time and paid about $10.5B in cash for CommScope's fibre business in January 2026, borrowing to do it. It buys back shares to offset staff options and pays a dividend, but has not issued shares for deals.
The last report
The latest report
Amphenol's third-quarter sales rose 65% to $10.2B, beating its own forecast
What it did to the thesis Solid quarter
It showed the AI demand is still growing and that CommScope earns more than planned. The pace of growth is slowing, with IT datacom's next rise forecast at a mid-single-digit percent, and the price already pays for the build lasting.
The report
In the third quarter of 2026 sales rose 65% to a record $10.2B, 37% without acquisitions, and new orders were $12.2B, more than it shipped. The data-centre segment grew 51% without acquisitions with a margin of 33.1%, and operating margin was 29.5%.
Management expects fourth-quarter sales of $10.3-10.5B and says IT datacom, now 48% of sales, rose 30% from the second quarter and should rise a mid-single-digit percent more. It also bought a fourth company and raised the CommScope outlook to $5.1B of sales.
What's happening now
The background
In January 2026 the shares fell 12% in a day on worries that growth was slowing. Amphenol then beat its forecast by about 9% twice and the shares hit a record in October. Over the year sales per share grew faster than the price, so the shares got cheaper per dollar of sales.
The third-quarter report on 28 October 2026 beat the $9.3-9.4B forecast with $10.2B. The next check is the fourth-quarter report in late January 2027, against a forecast of $10.3-10.5B.
Where we disagree
Both sides in full
The market
Amphenol is the AI connector winner and its shares still look reasonable after two years of growth.
We believe
The business is excellent but the price already pays for the AI build lasting; a pause would cut growth and margins together.
The argument
Almost every analyst is positive and few investors bet against the shares. We agree the business is excellent. Where we part is the price: it treats AI demand as permanent, while Amphenol's orders can be cancelled and most of its growth comes from a few data-centre spenders.
We could be wrong if orders keep growing above 30% without acquisitions through 2027, running ahead of sales: that would show the build lasting longer than we think.
What it's worth
How we value it
Valued on the analysts' revenue path for three years, each later year closing 30% of the gap to 4%, on official operating profit with stock pay a cost and tax paid in full; the thesis's own evidence is the check on that path.
What the price assumes
Sales growth in the first year, then slowing as ours does, at the same profit per sale.
The price assumed
22.3%
Our estimate
31%
Measured at the close of
6 Nov 2026
The three cases
Bear case$35.72
Growth runs at half the analysts' path, 16%, 9% and 6%, the margin settles a fifth lower at 21.9%, and new capital earns only the 8% rate: the downturn test.
Our value$79.67
Revenue grows 31%, 17% and 12% over three years, the analysts' path (C46), then fades toward 4%, closing 30% of the gap each year; the margin moves to 27.4%, the analysts' year-three figure after stock pay.
Bull case$112
Growth runs a quarter above the analysts' path, 39%, 22% and 16%, the margin reaches 30.1%, and new capital earns 15%.
The bet
Seen as a bet
The bet is on the data-centre segment, which earns most of the profit and grows fastest. Even if it keeps growing quickly for two more years, the whole company would be worth less than today's price at what its rivals cost.
If AI demand stalls, the segment keeps its profit but the shares would lose about half their value. Little upside and real downside: priced for success.
What could go wrong
Every risk
The risks
AI demand could pause
Virtually all of the recent growth came from AI products, bought by a few large data-centre spenders.
Orders can be cancelled
Customers may generally cancel orders before shipment, so the record pile of waiting orders is not a contract.
Optics on the chip
Optics built onto switch chips, which NVIDIA now ships, could replace some copper and plug-in links.
Margins at a record
Profit margins are at their highest, helped by a one-off tariff refund; a slowdown would squeeze them.
China and debt
Over a third of its factory assets sit in China, and it borrowed about $10.5B to buy CommScope's business.
What we still do not know
How much of the data-centre sales goes to the few largest customers, and on what terms.
How the AI growth splits between copper and fibre, and which way that mix is moving.
How big the market is once cables and fibre are counted with connectors: the research firms sell that figure.
What changes our view
Every route to a new view
Every route to a new view
→ Attractive
The share price falls to about $63.74, 80% of what we think the business is worth.
→ Unattractive
The share price rises to about $95.60, 20% above what we think the business is worth.
→ Unattractive
New orders fall below what it ships in a quarter. Or sales growth without acquisitions falls below 10%.
The next test
It supports the case if
Sales of $10.3B or more, book-to-bill of 1.1 or more, organic growth of 20% or more and a first 2027 outlook that implies sales above the fourth-quarter level.
It pressures the case if
Sales below $10.3B, book-to-bill below 1.0, organic growth under 20%, or a 2027 outlook that implies a sequential fall.
We would also be wrong if
The data-centre business shrinks two quarters running.
The operating margin falls below 24% in a quarter.
The verdict
Scores and how we weighed it
The scores
Growth
4 of 4
Business quality
3 of 4
Financial position
3 of 4
Execution record
4 of 4
Per share capture
3 of 4
Potential
3 of 4
Risk profile
3 of 4
Valuation
3 of 4
Evidence quality
3 of 4
How we weighed it
ValueMixed
The price is about nine-tenths of our value, with the bear case far below.
BusinessPositive
The leading maker of links inside AI racks, beating its forecasts year after year.
GrowthMixed
Growth is strong but slowing, and the price assumes the build lasts.
RisksNegative
AI demand could pause, orders can be cancelled, optics may replace copper and margins sit at a record.
RivalsNegative
It costs twice what its listed rivals do, for growth they lack.
The groupMixed
Server and network builders move together on AI spending news.
Latest newsMixed
AI demand still grew, but the pace is slowing.
In plain words
Book-to-bill
New orders received in a period divided by what was shipped. Above one means orders are piling up faster than the company can deliver; below one means the pile of work is shrinking.
In plain words
Our estimate of value
We forecast how much cash the business can earn over the next ten years and beyond, then ask what that stream is worth today to an investor who wants an 8% yearly return. We do it three times: a gloomy future, an expected one and a sunny one. We call a share Attractive only when its price is at least 20% below the expected estimate, and Unattractive when it is 20% above.