← All thesesSANMAttractiveMedium convictionWednesday 18 November 2026עברית

Sanmina

This is the archived full research. Read the story

A contract manufacturer that builds other companies' electronics, now above all the rack-sized AI computers AMD designs, assembled and tested in Sanmina's own factories.

Our viewAttractive Medium conviction
The scores
The profile · 1-4
Growth3/4 ↑

A material part was bought, and what is left narrows to one end market while the diversified half grew 0.2%.

Business quality2/4 ↑

A third-party valuation found $140M of goodwill and know-how in $1.62B, and Celestica's own filing calls this work aggressively priced.

Financial position3/4 ↓

Net debt is trivial and nothing matures early, but management intends to lever to 1.0-2.0 times and cash is not covering the build.

Execution record3/4 ↑

Eight straight beats and a fourth quarter 12% above its own guide, but the deal's returns are unproven.

Per share capture3/4 →

Shares shrank and the deal was paid in cash, but up to $450M of earn-out leaves precisely when the acquired business works.

Potential2/4 ↑

The upside is somebody else's spending decision on a programme that has shipped nothing, and no order discloses a volume or a term.

Risk profile3/4 ↓

Two customers now pass a tenth of sales and one relationship carries the growth case, but the diversified business survives without it.

Valuation4/4 ↑

About 13.4 times operating profit, 60% of a three-stage value near $255 at an 8% return: cheap, past the 20% margin.

Evidence quality3/4 →

Central facts are filed and corroborated outside the company, but the margin question and the fiscal 2027 step-up rest on management claims only.

The bet · 2-3 years
Priced for success0.6×

If it works 0.6× · if it fails you keep about 44%.

The Verdict

We rate Sanmina Attractive. It builds other companies' electronics, a thin-margin trade, and is about to assemble AMD's AI server racks in factories it bought from AMD; AI work is most of its sales, so the business is mixed. The price is about three-fifths of our value, and the latest report cleared our test, with a forecast above the promised target. The doubt is concentration: one rack programme for one customer carries the growth, part of the margin is engineering fees paid before any rack ships, and our gloomy case sits level with the price. The racks must still ship.

The long caseIf AMD's rack wins buyers from NVIDIA's, someone in America must build and test thousands of them a year, and Sanmina owns the factories AMD used.

How we weighed it

The price against our value

In its favour

Price at 60% of our value

The price is about three-fifths of our value and level with the bear case.

The business

Against it

Mixed

A thin-margin assembly trade, now dominated by AI work, with its best margin leaning on fees paid before shipments.

Growth against what the price assumes

In its favour

The guide for next year sits above the promised target, and the price needs only a modest margin on it.

What could go wrong

Against it

Ten customers make up most of the sales, the chip maker keeps control of volume, tax is rising and the earn-out owed to it has grown.

Against competitors

Cuts both ways

Rival assemblers chase the same racks; holding the factories the chip maker sold gives it a head start.

How its group is behaving

Cuts both ways

Server and network builders move together on AI spending news; the rack programme is its own fact.

The latest news and report

In its favour

The latest report cleared our test with a forecast above target and a margin well above the line; the racks must still ship.

The evidence under this - 8 claims
C01facthigh confidence
Sanmina builds other companies' electronic products in its own factories, reporting one segment, Integrated Manufacturing Solutions, which was about 90% of revenue for the nine months to 2026-06-27, alongside a components and services business.
current state · 2026-06-27written 2026-09-06
C02facthigh confidence
On 2025-10-27 Sanmina completed the purchase of ZT Systems' data centre infrastructure manufacturing business from AMD for $1.62B - $1.356B of cash net of $295M of cash acquired, plus 1,151,052 shares worth $155M - with up to $450M more contingent on gross profit and revenue metrics over three years.
historical · Q1 2026written 2026-09-06
C06facthigh confidence
For the nine months to 2026-06-27, the industrial, medical, defence and aerospace, and automotive end markets grew 0.2% to $3,783.9M, while communications networks and cloud and AI infrastructure grew from $2,256.1M to $6,883.1M.
historical · nine months ended 2026-06-27written 2026-09-06
C08facthigh confidence
Q3 FY2026 adjusted operating margin was 8.0% against 5.7% a year earlier, and adjusted gross margin 10.7% against 9.1%. Both are the highest Sanmina has reported.
historical · Q3 2026written 2026-09-06
C09management claimmedium confidence
Management names non-recurring engineering services - the pre-production work on a programme not yet in volume - as a driver of both the Q3 FY2026 margin beat and the Q4 margin guidance, and does not disclose how much of the margin it is.
against this
  • The same quarter's improvement is also attributed to favourable mix from adding ZT Systems, which is recurring, so engineering fees are one of at least two causes and possibly the smaller.
  • Sanmina's segment gross margins moved in the same direction independently: IMS adjusted gross margin rose 270 basis points year on year, which the addition of a whole business explains without any engineering fee.
  • Management has guided Q4 adjusted operating margin at 7.5%-8.0%, barely below the record, which would be odd if the record depended heavily on a fee that stops.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release does not split engineering fees from volume margin; shelf life moves to the first-quarter report
current state · 2026-07-27written 2026-09-06
C45estimatemedium confidence
At $8.7B of enterprise value (the 9 November close of $149.64 on 55.08M diluted shares, plus $2.09B of debt, less $1.68B of cash), paying 8 times operating profit, an ordinary price for contract manufacturing, needs $1.08B of operating profit, which on the $18.5B middle of the fiscal 2027 guide is a 5.8% margin. The best full year Sanmina has filed is 5.1%, and the guide is 7.0% to 7.5% before stock pay.
against this
  • 8 times operating profit is a judgement, not a filed fact. Sanmina's own history is nearer six times cash profit, which would demand more; a market willing to pay for AI exposure may accept twelve, which would demand a 5.9% margin and is within reach.
  • The 5.1% record is a GAAP figure and the 8.8% requirement is struck on the adjusted line management guides to, which runs roughly 130 basis points higher - so the true gap is smaller than the two numbers suggest.
  • Fiscal 2026 is already guided at a 6.85%-7.25% adjusted operating margin, so most of the distance to 8.8% has been covered in one year.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedre-struck on the fiscal 2027 guide and the 6 November close; the price had been $11.2B of enterprise value on the $16B target
  3. 2026-11-10editedRe-struck at the 9 November close of $149.64 (was $134.68 at 6 November): $8.7B of enterprise value, $1.08B of operating profit needed, a 5.8% margin on the guide's middle.
current state · 2026-11-09written 2026-09-06
C57management claimhigh confidence
On 9 November management guided fiscal 2027 revenue of $17.5B to $19.5B, non-GAAP operating margin of 7.0% to 7.5% and non-GAAP diluted earnings of $15.50 to $17.50, and first-quarter fiscal 2027 revenue of $3.8B to $4.1B at a 7.0% to 7.5% margin; the release does not split the accelerated-compute programme out.
against this
  • The guide does not say how much of $17.5B to $19.5B is the new rack programme or when in the year it lands; the first quarter's $3.8B to $4.1B is flat on the fourth quarter, so most of the step falls in the last three quarters.
  • The 7.0% to 7.5% margin is the adjusted line, which adds back stock pay and deal costs; the filed margin was 4.7% for fiscal 2026.
  1. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
  2. 2026-11-09editedthe business re-valued in three stages at an 8% required return, with analysts' margins put on the filed basis, after the day's review
forwardwritten 2026-11-09
C55estimatelow confidence
The Q4 fiscal 2026 guide, with no revenue from the new accelerated-compute programme in it, runs at about $12.8B a year, so the $16B-plus management states for fiscal 2027 needs about $3.2B a year from that programme and anything else that ramps.
derived$246.4M a week at the week-adjusted Q4 guidance midpoint (C12) times 52 weeks, against the $16B management states (C14); the programme contributes nothing to that guide (C10)
against this
  • The ZT Systems half of the run rate was falling 24% a week into Q4 (C12), so the existing programmes may not hold the $12.8B, which would leave more of the $16B to the new one.
  • Management names no customer, volume or term for the new programme's orders (C46), so the split of the step between it and other programmes is an inference.
forwardwritten 2026-10-11
01

The company

Sanmina builds electronics to other companies' designs in its own factories. For AI that means a whole rack-sized computer: it buys the parts, assembles the circuit boards, fits them into a rack with power and liquid cooling, runs it at full load and ships it to a data centre. AMD's Helios rack, which it is preparing for production, holds 72 AI chips.

It also makes circuit boards and other parts that go inside, which most assemblers buy in.

The evidence under this - 3 claims
C01facthigh confidence
Sanmina builds other companies' electronic products in its own factories, reporting one segment, Integrated Manufacturing Solutions, which was about 90% of revenue for the nine months to 2026-06-27, alongside a components and services business.
current state · 2026-06-27written 2026-09-06
C42factmedium confidence
An AMD Helios rack holds 72 accelerators in 18 Open Rack Wide four-GPU compute trays, on open standards, and is the product Sanmina is being paid to prepare for production.
current state · 2026-09-06written 2026-09-06
C38facthigh confidence
In Q3 FY2026 the manufacturing segment's adjusted gross margin was 10.2%, up 270 basis points on the year, and the components segment's 12.8%, down 190 basis points on the year but up 120 on the quarter.
historical · Q3 2026written 2026-09-06

How it makes money

It earns a thin cut on everything it builds: even June's record was 10.7 cents of adjusted gross profit per sales dollar. Before volume starts, customers also pay engineering fees to prepare their products for production. Because Sanmina buys the parts, much of its money sits in stock and unpaid customer bills: most of what it paid AMD bought exactly that.

The evidence under this - 3 claims
C08facthigh confidence
Q3 FY2026 adjusted operating margin was 8.0% against 5.7% a year earlier, and adjusted gross margin 10.7% against 9.1%. Both are the highest Sanmina has reported.
historical · Q3 2026written 2026-09-06
C09management claimmedium confidence
Management names non-recurring engineering services - the pre-production work on a programme not yet in volume - as a driver of both the Q3 FY2026 margin beat and the Q4 margin guidance, and does not disclose how much of the margin it is.
against this
  • The same quarter's improvement is also attributed to favourable mix from adding ZT Systems, which is recurring, so engineering fees are one of at least two causes and possibly the smaller.
  • Sanmina's segment gross margins moved in the same direction independently: IMS adjusted gross margin rose 270 basis points year on year, which the addition of a whole business explains without any engineering fee.
  • Management has guided Q4 adjusted operating margin at 7.5%-8.0%, barely below the record, which would be odd if the record depended heavily on a fee that stops.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release does not split engineering fees from volume margin; shelf life moves to the first-quarter report
current state · 2026-07-27written 2026-09-06
C04facthigh confidence
Of the $1.62B Sanmina paid, only $91M was recorded as goodwill and $49M as identifiable intangible assets - 8.6% of the consideration. The rest was working capital and equipment: $1,285M of receivables, $1,279M of inventory and $240M of plant.
current state · 2026-06-27written 2026-09-06
Where the money comes from, stream by stream
StreamShare of sales (nine months to 2026-06-27)How it earns it
Communications networks and cloud and AI infrastructure64.5%Building servers, racks, network and storage gear for equipment makers and cloud companies, now including ZT Systems' data-centre racks.
Industrial, medical, defence and aerospace, and automotive35.5%Building and testing other companies' industrial, medical, defence and vehicle electronics to order, on long-running programmes.

Who runs it

Jure Sola co-founded Sanmina in 1980 and has run it for most of the time since; Jonathan Faust has run the finances since 2023. Insiders own about 3%, nearly all of it Sola's. Their bonus rides on sales and an adjusted profit margin, their shares on adjusted earnings per share over three years. Nothing pays on the return the money earns, and the adjusted figures leave out the cost of share pay and of the ZT earn-out.

The evidence under this - 4 claims
C49facthigh confidence
Jure Sola co-founded Sanmina in 1980, was chief executive from April 1991 to October 2017, executive chairman from October 2017 to August 2020, and has been chairman and chief executive again since August 2020; Jonathan Faust has been chief financial officer since December 2023, after serving as HP's global controller and as chief financial officer of Aruba.
current state · 2026-01-23written 2026-10-04
C50facthigh confidence
As of 2025-12-31 directors and executive officers as a group (11 persons) owned 3.10% of the shares, nearly all of it Jure Sola's 2.82%; BlackRock at 15.13%, Vanguard at 13.83%, FMR at 10.69% and Dimensional at 6.25% were the holders above 5%. There is one class of common stock.
against this

The chief executive sold 118,368 shares in May 2026, after this table was struck (C32), so his stake is now smaller than the proxy states.

current state · 2025-12-31written 2026-10-04
C51facthigh confidence
For fiscal 2025 the annual bonus paid on revenue and non-GAAP operating margin with operating cash flow as a modifier, at a 140% corporate factor; long-term awards are PSUs on three-year cumulative non-GAAP earnings per share, paying 70% to 130%, and were all of the chief executive's fiscal 2025 grant. The fiscal 2023 PSUs vested at 108% on $18.51 of cumulative non-GAAP EPS, and the committee raised the chief financial officer's bonus by discretion to $1,300,000 against $910,000 by formula, for closing ZT Systems.
historical · FY2025written 2026-10-04
C52inferencemedium confidence
Neither half of the pay plan bears the two costs this thesis watches: both the bonus margin and the PSU earnings goal are non-GAAP, which exclude stock pay and the earn-out remeasurement, and nothing pays on return on invested capital, the measure the deal has lowered.
derivedC51's metrics, against C16 (stock pay excluded from adjusted figures), C24 (the earn-out remeasurement excluded) and C35 (trailing return on capital below its median)
against this
  • The operating cash flow modifier does bear the working capital the growth consumes (C20), so one part of the bonus answers to cash.
  • A three-year cumulative EPS goal still rewards the buybacks only if the earnings per share rise, so dilution is not wholly outside the plan.
current state · 2026-01-23written 2026-10-04
02

Why it wins, and who could beat it

Why customers come back

Two kinds of lock-in, of different strength. In medical, defence and aerospace a factory line is certified to the product, and moving it means certifying again. In AI racks the cost is time: a customer that put a product through Sanmina's pre-production work would have to repeat it elsewhere mid-launch.

AMD's choice is the evidence. Having bought ZT Systems for $4.4B, it kept the designers, sold the factories to Sanmina and named Sanmina its preferred American partner for launching new AI racks.

The evidence under this - 2 claims
C03facthigh confidence
AMD paid $4.4B for the whole of ZT Systems in March 2025, then sold only the manufacturing business to Sanmina in October 2025, keeping the design and research functions.
historical · FY2025-FY2026written 2026-09-06
C36facthigh confidence
AMD entered a manufacturing services agreement with ZT Systems at closing with an initial term of five years, and named Sanmina its United States new product introduction manufacturing partner of choice for AMD AI rack and cluster-scale systems. No volume or minimum is disclosed anywhere.
current state · 2026-06-27written 2026-09-06

The threat

Bigger rivals do the same work. Celestica's hardware business grew 81% to $5.0B in 2025, and Jabil's data-centre segment was 41% of its $29.8B of sales. And AMD kept ZT Systems' designers, so the higher-value half of the work stays with AMD, which decides where the volume goes.

The evidence under this - 3 claims
C26facthigh confidence
Celestica, doing the same work, reported Hardware Platform Solutions revenue of $5.0B in 2025, up 81% and 41% of its total, while its diversified segment grew 1% - and its own risk factors describe a highly competitive industry with aggressive pricing dynamics.
historical · FY2025written 2026-09-06
C27facthigh confidence
Jabil's Intelligent Infrastructure segment, which covers cloud and data centre infrastructure, was 41% of its $29.8B of fiscal 2025 revenue against 32% a year earlier, on gross margin for the whole group of 8.9%.
historical · FY2025written 2026-09-06
C03facthigh confidence
AMD paid $4.4B for the whole of ZT Systems in March 2025, then sold only the manufacturing business to Sanmina in October 2025, keeping the design and research functions.
historical · FY2025-FY2026written 2026-09-06

The rivals: CLS, JBL, AMD.

Who else does this, rival by rival
TickerRivalWhere it competesHow far alongPriced against it
CLSCelestica, through its Hardware Platform Solutions businessPriced against it: Same productToday's manufacturing (IMS and CPS, ZT's current programmes included)On saleSame product
JBLJabil, through its Intelligent Infrastructure segmentPriced against it: Same productToday's manufacturing (IMS and CPS, ZT's current programmes included)On saleSame product
AMDAMD's own retained ZT Systems design organisationPriced against it: Different productAMD Helios rack-scale systemsOn saleDifferent product
03

How big it could get

How big this could get

The market is rack-scale AI computers built by outside manufacturers, and nobody counts them. The spending behind them is visible: Amazon spent $173.0B on property and equipment in the year to June 2026, against $107.7B a year earlier. Celestica's $5.0B of hardware sales and Jabil's data-centre segment show what outside builders already earn from it.

The evidence under this - 3 claims
C41factmedium confidence
Amazon spent $173.0B on property and equipment in the twelve months to 2026-06-30 against $107.7B a year earlier, and Microsoft added $58.0B to computer equipment and software at cost in the nine months to 2026-03-31.
historical · twelve months ended 2026-06-30written 2026-09-06
C26facthigh confidence
Celestica, doing the same work, reported Hardware Platform Solutions revenue of $5.0B in 2025, up 81% and 41% of its total, while its diversified segment grew 1% - and its own risk factors describe a highly competitive industry with aggressive pricing dynamics.
historical · FY2025written 2026-09-06
C27facthigh confidence
Jabil's Intelligent Infrastructure segment, which covers cloud and data centre infrastructure, was 41% of its $29.8B of fiscal 2025 revenue against 32% a year earlier, on gross margin for the whole group of 8.9%.
historical · FY2025written 2026-09-06

Who has signed

AMD signed a five-year manufacturing agreement with the ZT business when Sanmina bought it, and management says it holds orders from large cloud companies and equipment makers for the next AI programme. No volume, minimum or customer name has been disclosed for any of it.

The evidence under this - 2 claims
C36facthigh confidence
AMD entered a manufacturing services agreement with ZT Systems at closing with an initial term of five years, and named Sanmina its United States new product introduction manufacturing partner of choice for AMD AI rack and cluster-scale systems. No volume or minimum is disclosed anywhere.
current state · 2026-06-27written 2026-09-06
C46management claimlow confidence
Management says it secured orders for the next-generation accelerated compute business with both hyperscale and original equipment customers, and won further orders in Q3, without naming a customer, a volume or a term.
against this
  • No order is quantified anywhere in the filings, and management declined an analyst's direct question on what share of AMD volume to assume, calling it tough to gauge.
  • Customers are described as still finalising production schedules and operating models, which is not the language of a fixed order book.
  • The 10-Q discloses no backlog figure and no remaining performance obligation, because the company elects not to disclose obligations with durations of a year or less.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedno order, customer or volume was added in the release; shelf life moves to the first-quarter report
current state · 2026-07-27written 2026-09-06
Where the growth is coming from, business by business

Where the growth is coming from

BusinessShare of revenueGrowth: last quarterLast half-yearLast year
IMS85%+78.8%+100.6%+72.3%
CPS Third Party Revenue15%+22.3%+14.6%+9.3%
Whole company100%+67.3%+83.0%+59.4%

The filed split through 2026-06-27. These rates are not carried forward: the reading below says why.

How we read it: The score reads neither figure as organic: IMS's rate is ZT Systems, bought October 2025 and supplying about $4B of the nine months' $4.6B rise, while the diversified end markets grew 0.2%.

04

The results

The numbers

Is it growing?Fast

Sales grew 79% in the year to Oct 2026, faster than the 7.4% in the year before.

Oct 2022: $7.92BSep 2023: $8.94BSep 2024: $7.57BSep 2025: $8.13BOct 2026: $14.5B$7.92B$14.5B
Is it profitable?Thin

Operating margin is 4.7%, about where it was (5.1%) in the year to Sep 2023.

Oct 2022: 4.4%Sep 2023: 5.1%Sep 2024: 4.4%Sep 2025: 4.4%Oct 2026: 4.7%4.4%4.7%
Does growth pay off?Flat

Running costs take 5¢ of each sales dollar, up from 3¢ in the year to Oct 2022.

Oct 2022: 3¢Sep 2023: 3¢Sep 2024: 4¢Sep 2025: 4¢Oct 2026: 5¢3¢5¢
Does it earn good returns?Decent

Each dollar put into the business earns 12% a year after tax, about where it was (13%) in the year to Oct 2022.

Oct 2022: 13%Sep 2023: 15%Sep 2024: 9.4%Sep 2025: 10%Oct 2026: 12%13%12%
Is it making cash?Thin

It turned 3.7% of sales into free cash, $540M over the year.

Oct 2022: $192MSep 2023: $43.8MSep 2024: $229MSep 2025: $473MOct 2026: $540M$192M$540M
Is the balance sheet safe?Light debt

It owes $412M more than the cash it holds, about 0.4 years of the cash its operations bring in.

Oct 2022: $183MSep 2023: $329MSep 2024: $309MSep 2025: $626MOct 2026: −$412M$183M−$412M
Are shares being diluted?Shrinking

The share count fell 13% in 4.0 years, and pay in shares costs 0.7% of sales.

Oct 2022: 63.1MSep 2023: 59.8MSep 2024: 57.0MSep 2025: 55.2MOct 2026: 55.2M63.1M55.2M
All the numbers

Filed figures from the company's 10-K and 10-Q, read 2026-11-17. Fiscal years end in October.

Oct 2022Sep 2023Sep 2024Sep 2025Oct 2026
Size and growth
Sales$7.92B$8.94B$7.57B$8.13B$14.5B
Sales growth+18%+13%−15%+7.4%+79%
Profit
Gross margin7.9%8.3%8.5%8.8%9.6%
Operating margin4.4%5.1%4.4%4.4%4.7%
Net margin3.0%3.5%2.9%3.0%2.6%
What growth costs, per sales dollar
Research and development0.3%0.3%0.4%0.4%0.2%
Selling, general and admin3.1%2.9%3.5%3.6%3.1%
Pay in shares0.5%0.6%0.8%0.8%0.7%
Capital spending1.8%2.1%1.5%1.8%2.8%
Cash
Cash from operations$331M$235M$340M$621M$944M
Free cash flow$192M$43.8M$229M$473M$540M
Balance sheet
Cash and short-term investments$530M$668M$626M$926M$1.68B
Debt$347M$338M$317M$300M$2.09B
Net cash$183M$329M$309M$626M−$412M
Returns and shares
Return on capital13%15%9.4%10%12%
Shares (diluted)63.1M59.8M57.0M55.2M55.2M

What growth costs

Growing costs Sanmina little in overhead: office costs and share pay barely rise with sales. What does rise is materials and, since ZT Systems, spending on factories. So growth adds profit roughly in step with sales rather than faster, and the margin has not widened since the deal.

The evidence under this - 2 claims
C54facthigh confidence
Per sales dollar, selling, general and administrative cost was 3.5 cents in fiscal 2021, 3.6 in fiscal 2025 and 3.2 over the four quarters to 2026-06-27; research and development 0.3, 0.4 and 0.3; stock pay 0.5, 0.8 and 0.7; capital spending 1.1, 1.8 and 2.4. GAAP operating margin was 3.8%, 4.4% and 4.2%, and gross margin 7.8%, 8.8% and 9.0%.
against this

The trailing year carries three quarters of ZT Systems on a cost base not yet integrated and $72M of earn-out remeasurement in GAAP operating profit (C24), so the latest column understates what the combined overhead will be.

historical · FY2021 to the four quarters to 2026-06-27written 2026-10-04
C20facthigh confidence
Operating cash flow for the nine months to 2026-06-27 was $702.0M against $421.6M, and capital spending $244.2M against $80.2M, leaving free cash flow of $457.8M. Working capital was a $43.7M source over the period.
historical · nine months ended 2026-06-27written 2026-09-06

The last report (9 Nov)

Quarter ended October 2026, filed figures.

84%
Sales growth
against the same quarter a year earlier
11.4%
Growth from the quarter before
against -13.7% the quarter before that
6¢
Operating profit
per dollar of sales

The fourth quarter of fiscal 2026 brought in $3.86B, up 84%, about 12% a week above the company's own forecast once the extra week is removed. For the year, sales were $14.5B, up 79%, and free cash flow was $541M, above last year's $478M.

Next year's forecast is $17.5B to $19.5B at an adjusted margin of 7.0% to 7.5%. The first quarter's $3.8B to $4.1B is level with the fourth, so the step comes late in the year. On the call management said the AI sales are still weighted to the second half, with a second customer's platform in the forecast and a third outside it.

What it did to the thesisIt cleared the test we set: a forecast above the promised target and a margin well above the line. The call kept the picture of a late-year step and gave no sign of a slip. With the shares well below their September price, the view moves up to Attractive.

The quarter in detail

Sales growth from the previous quarter, filed figures.

Earnings per share $4.43 against $3.27 expected.

The evidence under this - 6 claims
C56facthigh confidence
Q4 FY2026 revenue was $3.858B, up 84% on $2.096B, and fiscal 2026 revenue $14.525B, up 79%. GAAP operating margin was 6.0% in the quarter and 4.7% for the year; non-GAAP operating margin 9.0% and 7.4%. Fiscal 2026 operating cash flow was $944M and free cash flow $541M; year-end cash $1,676M against $2,088M of debt.
  1. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
historical · Q4 FY2026written 2026-11-09
C57management claimhigh confidence
On 9 November management guided fiscal 2027 revenue of $17.5B to $19.5B, non-GAAP operating margin of 7.0% to 7.5% and non-GAAP diluted earnings of $15.50 to $17.50, and first-quarter fiscal 2027 revenue of $3.8B to $4.1B at a 7.0% to 7.5% margin; the release does not split the accelerated-compute programme out.
against this
  • The guide does not say how much of $17.5B to $19.5B is the new rack programme or when in the year it lands; the first quarter's $3.8B to $4.1B is flat on the fourth quarter, so most of the step falls in the last three quarters.
  • The 7.0% to 7.5% margin is the adjusted line, which adds back stock pay and deal costs; the filed margin was 4.7% for fiscal 2026.
  1. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
  2. 2026-11-09editedthe business re-valued in three stages at an 8% required return, with analysts' margins put on the filed basis, after the day's review
forwardwritten 2026-11-09
C12facthigh confidence
Q4 FY2026 revenue was $3.858B, or $275.6M a week over the fourteen weeks, 3.4% above Q3's $266.5M a week and 11.9% above the $246.4M a week its own guidance midpoint implied, where the guide had read as a 7.5% weekly decline.
against this
  • Management attributes the ZT Systems decline entirely to a few legacy programmes whose timing moves between quarters, and those programmes were always expected to run down to zero.
  • The next-generation programme is excluded from the quarter by revenue recognition, so the guide is not a measure of demand; the order book behind it is described as secured.
  • Guidance has been beaten in each of the last eight quarters, and Q1 FY2026 came in above the top of both guided ranges, so a midpoint is a conservative baseline.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe Q4 print replaced the guidance-based estimate with the reported figure
historical · Q4 FY2026written 2026-09-06
C21facthigh confidence
Fiscal 2026 free cash flow was $541M on revenue of $14.53B, above fiscal 2025's $478M on $8.13B: operating cash flow $944M against $621M, less about $403M of net capital spending.
against this
  • Capital spending is the reason, and it is deliberate growth investment: $244.2M in nine months against $80.2M, with $135M more guided for Q4, on power, liquid cooling, test cells, metal fabrication and printed circuit board capacity.
  • Management explicitly warns working capital will grow ahead of revenue, so the cash shortfall is the cost of the growth rather than a deterioration in the business.
  1. 2026-11-09editedthe full-year result replaced the nine-month estimate that had it level with fiscal 2025
historical · FY2025written 2026-09-06
C08facthigh confidence
Q3 FY2026 adjusted operating margin was 8.0% against 5.7% a year earlier, and adjusted gross margin 10.7% against 9.1%. Both are the highest Sanmina has reported.
historical · Q3 2026written 2026-09-06
C60management claimmedium confidence
On the 9 November call management said fiscal 2027 cloud and AI systems revenue of $7.5B to $9B (against $5.3B in fiscal 2026, $4.9B of it ZT Systems) is still expected to be back-end loaded to the second half, with the Helios business the bulk and other platforms ramping through the year; a second platform beyond AMD is in the guide, a third is not, and net accounting on the consignment model has been finalised with a couple of customers, the rest in progress.
against this
  • The call does not say whether the programme counts as revenue in the first quarter, the test the thesis pre-registered; 'back-end loaded' with a first-quarter guide level with the fourth quarter fits either reading.
  • Management calls the full-year range wider than usual because the programmes are new and cites supply-chain shortages, so the guide carries execution risk it names itself.
current state · 2026-11-09written 2026-11-10
What management did with the money

It borrowed to pay for ZT, taking debt to $2.17B, though $1.84B of cash covers most of it. It bought back 1.59 million shares for $239.2M, more than the shares issued to AMD. Free cash flow is on course to be no higher than fiscal 2025's $478M, on nearly twice the sales.

The evidence under this - 4 claims
C18facthigh confidence
Total debt was $2,087.5M at 2026-10-03 against $300.5M a year earlier, and cash $1,675.6M, leaving net debt of $411.9M.
  1. 2026-11-09editedyear-end balance sheet from the Q4 release
current state · 2026-10-03written 2026-09-06
C22facthigh confidence
Diluted shares were 55.13M in Q3 FY2026 against 54.49M a year earlier. Over the nine months the company repurchased 1.593M shares for $239.2M and issued 1.151M for the acquisition, leaving shares outstanding slightly lower.
current state · 2026-06-27written 2026-09-06
C20facthigh confidence
Operating cash flow for the nine months to 2026-06-27 was $702.0M against $421.6M, and capital spending $244.2M against $80.2M, leaving free cash flow of $457.8M. Working capital was a $43.7M source over the period.
historical · nine months ended 2026-06-27written 2026-09-06
C21facthigh confidence
Fiscal 2026 free cash flow was $541M on revenue of $14.53B, above fiscal 2025's $478M on $8.13B: operating cash flow $944M against $621M, less about $403M of net capital spending.
against this
  • Capital spending is the reason, and it is deliberate growth investment: $244.2M in nine months against $80.2M, with $135M more guided for Q4, on power, liquid cooling, test cells, metal fabrication and printed circuit board capacity.
  • Management explicitly warns working capital will grow ahead of revenue, so the cash shortfall is the cost of the growth rather than a deterioration in the business.
  1. 2026-11-09editedthe full-year result replaced the nine-month estimate that had it level with fiscal 2025
historical · FY2025written 2026-09-06
05

The price, and where we disagree

In short. Cheap. The price is about 60% of what we think the business is worth, at an 8% yearly return.

Share price
$198.47
close,
Price for each $1 of yearly profit
34.8
its 6-year normal 12.4
Company value for each $1 of cash profit
16.8
its 6-year normal 6.0

The forecast Sanmina gave on 9 November is the test the thesis was waiting for. Getting to its middle needs weekly sales 29% above the fourth quarter's, from an AI programme not yet counted as revenue. On the call management said the AI rack sales are still weighted to the second half.

Meanwhile insiders have sold: the chief executive sold 118,368 shares in May, and 19 of 20 open-market insider trades in the past year were sales.

The evidence under this - 4 claims
C57management claimhigh confidence
On 9 November management guided fiscal 2027 revenue of $17.5B to $19.5B, non-GAAP operating margin of 7.0% to 7.5% and non-GAAP diluted earnings of $15.50 to $17.50, and first-quarter fiscal 2027 revenue of $3.8B to $4.1B at a 7.0% to 7.5% margin; the release does not split the accelerated-compute programme out.
against this
  • The guide does not say how much of $17.5B to $19.5B is the new rack programme or when in the year it lands; the first quarter's $3.8B to $4.1B is flat on the fourth quarter, so most of the step falls in the last three quarters.
  • The 7.0% to 7.5% margin is the adjusted line, which adds back stock pay and deal costs; the filed margin was 4.7% for fiscal 2026.
  1. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
  2. 2026-11-09editedthe business re-valued in three stages at an 8% required return, with analysts' margins put on the filed basis, after the day's review
forwardwritten 2026-11-09
C13estimatehigh confidence
The $18.5B middle of the fiscal 2027 guide needs weekly revenue of $355.8M over 52 weeks, 29% above the fourth quarter's $275.6M; the first-quarter guide's middle of $3.95B over 13 weeks is $303.8M, 10% above, so the step is weighted to the last three quarters.
against this
  • The accelerated compute programme excluded from Q4 is expected to begin in Q1 FY2027 and ramp through the year, so the exit rate is not meant to be the starting point.
  • Core Sanmina grew 12.6% in fiscal 2026 against an initial plan of high single digits, so the non-AI half is running ahead of its own plan.
  • Management has said it can deliver more than it thought a year ago, and moved the $16B target forward from fiscal 2028 to fiscal 2027 rather than back.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedre-struck on the 9 November guide and the reported fourth quarter
forwardwritten 2026-09-06
C10management claimhigh confidence
The next-generation accelerated compute programme contributes no revenue to the Q4 FY2026 outlook, on revenue-recognition timing, and management expects it to begin contributing in Q1 FY2027 and ramp over that year.
against this
  • Management describes the programme as on track and says customer validation is going well, so the exclusion is presented as accounting timing rather than a delay.
  • Pre-production work is already being paid for as engineering services, so the relationship is generating revenue even though the product is not.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release is silent on when the accelerated-compute programme first contributes; the call was not yet on the vendor, so the claim stays a management claim and its shelf life moves to the first-quarter report
forwardwritten 2026-09-06
C32factmedium confidence
In the year to 2026-09-06 insiders filed 19 open-market sales and one open-market purchase of 700 shares. The chief executive sold 118,368 shares at $228.84 on 2026-05-06 and the chief financial officer 10,076 at about $265.80 three weeks later.
against this
  • The chief executive retained 1,227,573 shares after that sale, so he sold under a tenth of his holding and remains by far the largest individual holder.
  • The selling clustered in April and May 2026 at $207-$266, when the shares were near their high, which is consistent with routine diversification rather than a view.
historical · year to 2026-09-06written 2026-09-06
The betPriced for success
If it works
0.6×
from today's value, at the stated multiple
If it fails
keep 44%
of today's market value: the rest of the company, less its debt
How we value it
Method
Cash the business will produce, in three stages
Required return
8%
Long-run growth
4%
Moat
No moat
BearBaseBull
Sales growth, years 1 to 313.7%, 9.9%27.3%, 19.7%34.1%, 24.6%
Normal operating margin4.8%6%6.6%
Return on new capital after year ten8%8%11%
Value a share$150$255$416
Price
$153
Priced on
17 Nov
Price as a share of the base value
60%
Starting growth the price pays for
1.7%
Yearly return the price offers on the base case
11.4%
Risk check: valued at 10%, the rate for a cyclical or indebted business
$186

Where each figure comes from

FigureSourceWhere exactly
Sales, the last full yearFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026; the full year.
Operating margin todayFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026; the full year.
Tax rateFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026; the full year.
Sales per dollar of new capitalFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026: the year's sales over its operating capital.
SharesFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026; the diluted share count.
Cash and investmentsFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026; the balance sheet on that date.
DebtFiled reportAnnual report (10-K) for the period to 3 Oct 2026, filed 17 Nov 2026; the balance sheet on that date.
Sales growth, years 1 to 3Analysts' forecastsFMP's consensus for the years to 3 Oct 2027 and 3 Oct 2028: the average of 3 analysts' revenue estimates, 27.3%, 19.7%. FMP publishes how many analysts, not who they are.
Normal operating marginAnalysts' forecastsFMP's consensus operating margin for the year to 3 Oct 2028, the average of 3 analysts' estimates, 6.7%, less stock pay of 0.7% of sales. FMP publishes how many analysts, not who they are.
Return on new capital after year tenOur method's ruleSet by our reading of the moat: the 8% rate with none, 12% with an average one, 15% with a strong one; the bear case earns the rate.
The bear and bull casesOur method's ruleOur method's spread: the bear case takes half the growth and a fifth off the margin, the bull a quarter more growth and a tenth more margin.
PriceMarket priceFMP's closing price on 17 Nov 2026.

What each case rests on

Bear caseGrowth runs at half the base path, 14% and 10%, the margin settles a fifth lower at 4.8%, and new capital earns only the 8% rate: the downturn test.

Base caseRevenue grows 27% and 20% over the next two years: the middle of the $17.5B to $19.5B guide (C57) on fiscal 2026's $14.53B, which is where the analysts now sit, then their fiscal 2028 revenue (C58); it then fades toward 4%, closing 30% of the gap each year. The margin moves to 6.0%, the analysts' year-three margin after stock pay, below the 6.4% best it has filed.

Bull caseGrowth runs a quarter above the base path, 34% and 25%, the margin reaches 6.6%, and new capital earns 11%.

What moves the base value

LowerHigher
Growth in years 1 to 3, a quarter lower or higher$221$294
The normal margin, 15% lower or higher$210$301
Capital per dollar of growth, a quarter more or less$235$267
Return on new capital after year ten, at the rate or one step up$255$320
The rate, a point higher or lower$197$354
Long-run growth a point lower$248-

Value a share as growth and margin change

Normal margin ↓ · growth →50%75%100%125%150%
4.8%$149$171$195$222$254
5.4%$171$196$225$258$296
6%$192$221$255$294$338
6.6%$213$247$286$330$380
7.2%$234$272$316$366$423

Growth is shown as a share of the base case's. Green: the price is at or under 80% of that value. Red: at or over 120%.

How we read the multiple

Taking the whole company, debt included, buyers pay about 12.7 times the operating profit Sanmina filed for the year.

To be worth 8 times its operating profit, an ordinary price for a contract manufacturer, Sanmina must:

  1. Earn about $1.08B a year from operations, since the whole company, debt included, is valued at about $8.7B.
  2. On the $18.5B middle of its forecast, that is 5.8 cents of profit on each dollar.
  3. Its best full year ever filed was 5.1 cents; it forecasts 7.0 to 7.5 before stock pay.

Against its own history. Still dearer than its own six-year normal on yearly profit per share, though well below its early-September price.

The evidence under this - 3 claims
C45estimatemedium confidence
At $8.7B of enterprise value (the 9 November close of $149.64 on 55.08M diluted shares, plus $2.09B of debt, less $1.68B of cash), paying 8 times operating profit, an ordinary price for contract manufacturing, needs $1.08B of operating profit, which on the $18.5B middle of the fiscal 2027 guide is a 5.8% margin. The best full year Sanmina has filed is 5.1%, and the guide is 7.0% to 7.5% before stock pay.
against this
  • 8 times operating profit is a judgement, not a filed fact. Sanmina's own history is nearer six times cash profit, which would demand more; a market willing to pay for AI exposure may accept twelve, which would demand a 5.9% margin and is within reach.
  • The 5.1% record is a GAAP figure and the 8.8% requirement is struck on the adjusted line management guides to, which runs roughly 130 basis points higher - so the true gap is smaller than the two numbers suggest.
  • Fiscal 2026 is already guided at a 6.85%-7.25% adjusted operating margin, so most of the distance to 8.8% has been covered in one year.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedre-struck on the fiscal 2027 guide and the 6 November close; the price had been $11.2B of enterprise value on the $16B target
  3. 2026-11-10editedRe-struck at the 9 November close of $149.64 (was $134.68 at 6 November): $8.7B of enterprise value, $1.08B of operating profit needed, a 5.8% margin on the guide's middle.
current state · 2026-11-09written 2026-09-06
C28estimatemedium confidence
At the 9 November close of $149.64 buyers pay about 21.8 times fiscal 2026 GAAP diluted earnings of $6.85 a share, against a six-year median near 12.5 on the 6 September pack; the cash-profit comparison is not re-struck.
against this
  • The trailing figures are depressed by acquisition charges that will not repeat: $137.0M of acquisition and integration costs, $49M of inventory step-up and $72M of earn-out remeasurement in nine months.
  • On management's own fiscal 2026 adjusted guidance the price is about $16.50 for each dollar of profit, and on the vendor panel's fiscal 2027 estimate about $14.20 - a re-rating still, but a far smaller one.
  • The business is not the business the median was struck on: revenue is up 74% and adjusted operating margin is running roughly 130 basis points higher.
  1. 2026-11-09editedre-struck on the fiscal 2026 filed earnings and the 6 November close
  2. 2026-11-10editedRe-struck at the 9 November close of $149.64: 21.8 times fiscal 2026 GAAP diluted earnings of $6.85 (was 19.7 at $134.68).
current state · 2026-11-09written 2026-09-06
C09management claimmedium confidence
Management names non-recurring engineering services - the pre-production work on a programme not yet in volume - as a driver of both the Q3 FY2026 margin beat and the Q4 margin guidance, and does not disclose how much of the margin it is.
against this
  • The same quarter's improvement is also attributed to favourable mix from adding ZT Systems, which is recurring, so engineering fees are one of at least two causes and possibly the smaller.
  • Sanmina's segment gross margins moved in the same direction independently: IMS adjusted gross margin rose 270 basis points year on year, which the addition of a whole business explains without any engineering fee.
  • Management has guided Q4 adjusted operating margin at 7.5%-8.0%, barely below the record, which would be odd if the record depended heavily on a fee that stops.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release does not split engineering fees from volume margin; shelf life moves to the first-quarter report
current state · 2026-07-27written 2026-09-06
Priced against its rivals

There is no group to price it against. No cohort was established, so nothing says whether this price is Sanmina's own finding or something the whole contract-manufacturing trade is carrying. The price is judged against its own history instead.

The bet: what it pays if it works

The bet is on AMD's Helios racks alone: complete AI server racks Sanmina is preparing to build in America, which have not shipped and carry no disclosed volume.

Today's business, valued at an ordinary contract manufacturer's price, covers under half of today's price, so buyers already pay a lot for the racks. Even if they bring the step in sales management promises for next year, the whole company would be worth less than today: an expensive bet.

If the racks never ship in volume, shareholders keep today's business.

AMD Helios rack-scale systems operating profit in two to three yearsBillion dollars a year of rack revenue neededAt the stated multiple
$51.00Mabout 1 at 5% kept0.5×
$163.00M (the big win)about 3 at 5% kept0.6×
$255.00Mabout 5 at 5% kept0.6×

To pay 2× (a Fair bet) it would need about $2.15B of AMD Helios rack-scale systems operating profit (about 42 billion dollars a year of rack revenue at 5% kept); to pay 5× (a Cheap bet), about $6.36B of AMD Helios rack-scale systems operating profit (about 125 billion dollars a year of rack revenue at 5% kept), at the same multiple.

Each row is AMD Helios rack-scale systems alone, plus the rest of the company (Today's manufacturing (IMS and CPS, ZT's current programmes included)) counted at about $5.22B.

The steps it has to take

  • Done AMD names Sanmina its US new-product-introduction partner; Helios pre-production under way
  • Done Orders won from hyperscale and original equipment customers
  • Not started Helios revenue recognised due 2027-01
  • Not started Fiscal 2027 revenue at or above $16B due 2027-11
The detail behind the label
Rating
Expensive
the big win's payoff against the 2× and 5× lines
Cheap bet at about
≈ $13.39
93% below the price (close, )
The option
0.2×
what AMD Helios rack-scale systems is worth if it works, against what the price pays for it
The evidence under this - 6 claims
C55estimatelow confidence
The Q4 fiscal 2026 guide, with no revenue from the new accelerated-compute programme in it, runs at about $12.8B a year, so the $16B-plus management states for fiscal 2027 needs about $3.2B a year from that programme and anything else that ramps.
derived$246.4M a week at the week-adjusted Q4 guidance midpoint (C12) times 52 weeks, against the $16B management states (C14); the programme contributes nothing to that guide (C10)
against this
  • The ZT Systems half of the run rate was falling 24% a week into Q4 (C12), so the existing programmes may not hold the $12.8B, which would leave more of the $16B to the new one.
  • Management names no customer, volume or term for the new programme's orders (C46), so the split of the step between it and other programmes is an inference.
forwardwritten 2026-10-11
C14management claimmedium confidence
On 9 November management guided fiscal 2027 revenue of $17.5B to $19.5B, above the '$16 billion-plus' it had stated since bringing the target forward from fiscal 2028; the middle is 27% above fiscal 2026.
against this
  • It is not a formal guide. Management said twice on the same call that the formal fiscal 2027 guidance comes in 90 days, and declined to give a share estimate for AMD volume.
  • The target was raised before any of the volume it depends on had shipped, and the programme delivering it has since been pushed out of the fourth quarter.
  • AMD's own commentary describes a measured ramp to make sure manufacturing partners are ready, which is a slower framing than the target implies.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe guide arrived above the stated target
forwardwritten 2026-09-06
C45estimatemedium confidence
At $8.7B of enterprise value (the 9 November close of $149.64 on 55.08M diluted shares, plus $2.09B of debt, less $1.68B of cash), paying 8 times operating profit, an ordinary price for contract manufacturing, needs $1.08B of operating profit, which on the $18.5B middle of the fiscal 2027 guide is a 5.8% margin. The best full year Sanmina has filed is 5.1%, and the guide is 7.0% to 7.5% before stock pay.
against this
  • 8 times operating profit is a judgement, not a filed fact. Sanmina's own history is nearer six times cash profit, which would demand more; a market willing to pay for AI exposure may accept twelve, which would demand a 5.9% margin and is within reach.
  • The 5.1% record is a GAAP figure and the 8.8% requirement is struck on the adjusted line management guides to, which runs roughly 130 basis points higher - so the true gap is smaller than the two numbers suggest.
  • Fiscal 2026 is already guided at a 6.85%-7.25% adjusted operating margin, so most of the distance to 8.8% has been covered in one year.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedre-struck on the fiscal 2027 guide and the 6 November close; the price had been $11.2B of enterprise value on the $16B target
  3. 2026-11-10editedRe-struck at the 9 November close of $149.64 (was $134.68 at 6 November): $8.7B of enterprise value, $1.08B of operating profit needed, a 5.8% margin on the guide's middle.
current state · 2026-11-09written 2026-09-06
C12facthigh confidence
Q4 FY2026 revenue was $3.858B, or $275.6M a week over the fourteen weeks, 3.4% above Q3's $266.5M a week and 11.9% above the $246.4M a week its own guidance midpoint implied, where the guide had read as a 7.5% weekly decline.
against this
  • Management attributes the ZT Systems decline entirely to a few legacy programmes whose timing moves between quarters, and those programmes were always expected to run down to zero.
  • The next-generation programme is excluded from the quarter by revenue recognition, so the guide is not a measure of demand; the order book behind it is described as secured.
  • Guidance has been beaten in each of the last eight quarters, and Q1 FY2026 came in above the top of both guided ranges, so a midpoint is a conservative baseline.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe Q4 print replaced the guidance-based estimate with the reported figure
historical · Q4 FY2026written 2026-09-06
C36facthigh confidence
AMD entered a manufacturing services agreement with ZT Systems at closing with an initial term of five years, and named Sanmina its United States new product introduction manufacturing partner of choice for AMD AI rack and cluster-scale systems. No volume or minimum is disclosed anywhere.
current state · 2026-06-27written 2026-09-06
C10management claimhigh confidence
The next-generation accelerated compute programme contributes no revenue to the Q4 FY2026 outlook, on revenue-recognition timing, and management expects it to begin contributing in Q1 FY2027 and ramp over that year.
against this
  • Management describes the programme as on track and says customer validation is going well, so the exclusion is presented as accounting timing rather than a delay.
  • Pre-production work is already being paid for as engineering services, so the relationship is generating revenue even though the product is not.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release is silent on when the accelerated-compute programme first contributes; the call was not yet on the vendor, so the claim stays a management claim and its shelf life moves to the first-quarter report
forwardwritten 2026-09-06

Where we disagree with the market

The market believes

An assembler deserves an assembler's price: thin margins, one large customer, and AI sales that have not yet shipped.

We believe

The market is right about the thin margin. At November's price it already pays for the forecast falling short, which the forecast itself does not.

The first quarter's forecast is level with the fourth quarter, so the whole step to $18.5B rests on the AI programme, which has shipped nothing.

We are wrong if first-quarter sales miss $3.8B, or if the racks are pushed to the second half or later.

What analysts say, and the evidence
The evidence under this - 5 claims
C11inferencehigh confidence
Q4 FY2026 is a 14-week quarter in a 53-week fiscal year, so the guidance covers one more week of trading than the quarter it is compared against.
derivedThe 10-K states Sanmina operates on a 52- or 53-week year ending the Saturday nearest 30 September, and that fiscal 2025, 2024 and 2023 were each 52 weeks. The Q3 release names the fourth quarter as ending 2026-10-03, which is 371 days - 53 weeks - after fiscal 2025 ended on 2025-09-27, leaving 98 days or 14 weeks in the quarter. Management states on the call that the outlook 'includes an extra week'.
against this
  • The extra week is disclosed rather than hidden: the chief financial officer names it in the first sentence of the fourth-quarter outlook, so any analyst modelling the quarter already has it.
  • A fourteenth week does not add a proportionate thirteenth of revenue. The quarter ends 3 October and manufacturing weeks around a fiscal year end carry holidays and customer shutdowns, so treating the guide as fourteen equal weeks overstates the adjustment by an unknown amount.
  • Sanmina has run 53-week years before under the same convention, so the comparison distortion is a recurring feature of its calendar rather than something new about this guide.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
forwardwritten 2026-09-06
C12facthigh confidence
Q4 FY2026 revenue was $3.858B, or $275.6M a week over the fourteen weeks, 3.4% above Q3's $266.5M a week and 11.9% above the $246.4M a week its own guidance midpoint implied, where the guide had read as a 7.5% weekly decline.
against this
  • Management attributes the ZT Systems decline entirely to a few legacy programmes whose timing moves between quarters, and those programmes were always expected to run down to zero.
  • The next-generation programme is excluded from the quarter by revenue recognition, so the guide is not a measure of demand; the order book behind it is described as secured.
  • Guidance has been beaten in each of the last eight quarters, and Q1 FY2026 came in above the top of both guided ranges, so a midpoint is a conservative baseline.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe Q4 print replaced the guidance-based estimate with the reported figure
historical · Q4 FY2026written 2026-09-06
C10management claimhigh confidence
The next-generation accelerated compute programme contributes no revenue to the Q4 FY2026 outlook, on revenue-recognition timing, and management expects it to begin contributing in Q1 FY2027 and ramp over that year.
against this
  • Management describes the programme as on track and says customer validation is going well, so the exclusion is presented as accounting timing rather than a delay.
  • Pre-production work is already being paid for as engineering services, so the relationship is generating revenue even though the product is not.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release is silent on when the accelerated-compute programme first contributes; the call was not yet on the vendor, so the claim stays a management claim and its shelf life moves to the first-quarter report
forwardwritten 2026-09-06
C14management claimmedium confidence
On 9 November management guided fiscal 2027 revenue of $17.5B to $19.5B, above the '$16 billion-plus' it had stated since bringing the target forward from fiscal 2028; the middle is 27% above fiscal 2026.
against this
  • It is not a formal guide. Management said twice on the same call that the formal fiscal 2027 guidance comes in 90 days, and declined to give a share estimate for AMD volume.
  • The target was raised before any of the volume it depends on had shipped, and the programme delivering it has since been pushed out of the fourth quarter.
  • AMD's own commentary describes a measured ramp to make sure manufacturing partners are ready, which is a slower framing than the target implies.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe guide arrived above the stated target
forwardwritten 2026-09-06
C57management claimhigh confidence
On 9 November management guided fiscal 2027 revenue of $17.5B to $19.5B, non-GAAP operating margin of 7.0% to 7.5% and non-GAAP diluted earnings of $15.50 to $17.50, and first-quarter fiscal 2027 revenue of $3.8B to $4.1B at a 7.0% to 7.5% margin; the release does not split the accelerated-compute programme out.
against this
  • The guide does not say how much of $17.5B to $19.5B is the new rack programme or when in the year it lands; the first quarter's $3.8B to $4.1B is flat on the fourth quarter, so most of the step falls in the last three quarters.
  • The 7.0% to 7.5% margin is the adjusted line, which adds back stock pay and deal costs; the filed margin was 4.7% for fiscal 2026.
  1. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
  2. 2026-11-09editedthe business re-valued in three stages at an 8% required return, with analysts' margins put on the filed basis, after the day's review
forwardwritten 2026-11-09
06

What could go wrong, and how we would know

The risks

  • One customer carries it. The ten largest customers were 63% of the year's sales, against about half before, and one relationship carries most of the growth.
  • AMD holds the volume. AMD kept the design team and has disclosed no minimum, so it could hand Helios volume to a second manufacturer.
  • A heavier tax bill. The tax rate rose to 28.5% from 19.5% as production moved to the United States.
  • An earn-out valued twice. AMD expects to receive $348M from the earn-out; Sanmina carries $183M for the same agreement, capped at $450M.
The evidence under this - 5 claims
C07facthigh confidence
The ten largest customers were 63% of net sales in fiscal 2026 against about half in earlier years; two customers were 16.7% and 10.1% of sales, where in fiscal 2025 and 2024 a single customer reached 10.1%.
  1. 2026-11-18editedQ4 and fiscal 2026 annual report (10-K, filed 17 November): the full year's ten largest customers were 63% of sales, down from 67% over the nine months to June, with two customers over a tenth.
current state · 2026-10-03written 2026-09-06
C03facthigh confidence
AMD paid $4.4B for the whole of ZT Systems in March 2025, then sold only the manufacturing business to Sanmina in October 2025, keeping the design and research functions.
historical · FY2025-FY2026written 2026-09-06
C36facthigh confidence
AMD entered a manufacturing services agreement with ZT Systems at closing with an initial term of five years, and named Sanmina its United States new product introduction manufacturing partner of choice for AMD AI rack and cluster-scale systems. No volume or minimum is disclosed anywhere.
current state · 2026-06-27written 2026-09-06
C17factmedium confidence
The effective tax rate for the nine months to 2026-06-27 was 28.5% against 19.5% a year earlier, as production shifted to the United States: the US share of Americas sales rose from 30% to 64% over the same nine months.
historical · nine months ended 2026-06-27written 2026-09-06
C23facthigh confidence
The two parties value the same earn-out very differently: at 2026-09-26 AMD carries a $348M receivable it says is probable of being received, while at 2026-10-03 Sanmina carries a $261M liability at fair value, on the same agreement capped at $450M.
  1. 2026-11-18editedQ4 and fiscal 2026 annual report (10-K, filed 17 November): Sanmina's liability rose from $183M to $261M after a $150M remeasurement over the year; AMD's September 10-Q still carries $348M.
current state · 2026-10-03written 2026-09-06
What we still do not know
  • How much of the record margin is engineering fees that will not repeat once racks ship in volume. Management has not said.
  • How many Helios racks AMD will order, and when. No volume or minimum is disclosed.
The evidence under this - 3 claims
C09management claimmedium confidence
Management names non-recurring engineering services - the pre-production work on a programme not yet in volume - as a driver of both the Q3 FY2026 margin beat and the Q4 margin guidance, and does not disclose how much of the margin it is.
against this
  • The same quarter's improvement is also attributed to favourable mix from adding ZT Systems, which is recurring, so engineering fees are one of at least two causes and possibly the smaller.
  • Sanmina's segment gross margins moved in the same direction independently: IMS adjusted gross margin rose 270 basis points year on year, which the addition of a whole business explains without any engineering fee.
  • Management has guided Q4 adjusted operating margin at 7.5%-8.0%, barely below the record, which would be odd if the record depended heavily on a fee that stops.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release does not split engineering fees from volume margin; shelf life moves to the first-quarter report
current state · 2026-07-27written 2026-09-06
C36facthigh confidence
AMD entered a manufacturing services agreement with ZT Systems at closing with an initial term of five years, and named Sanmina its United States new product introduction manufacturing partner of choice for AMD AI rack and cluster-scale systems. No volume or minimum is disclosed anywhere.
current state · 2026-06-27written 2026-09-06
C10management claimhigh confidence
The next-generation accelerated compute programme contributes no revenue to the Q4 FY2026 outlook, on revenue-recognition timing, and management expects it to begin contributing in Q1 FY2027 and ramp over that year.
against this
  • Management describes the programme as on track and says customer validation is going well, so the exclusion is presented as accounting timing rather than a delay.
  • Pre-production work is already being paid for as engineering services, so the relationship is generating revenue even though the product is not.
  1. 2026-11-03editedresults date corrected to 9 November: Sanmina announced on 19 October it would report then, not on 2 November
  2. 2026-11-09editedthe release is silent on when the accelerated-compute programme first contributes; the call was not yet on the vendor, so the claim stays a management claim and its shelf life moves to the first-quarter report
forwardwritten 2026-09-06

What would change our view

  1. Attractive→NeutralThe price rises within 20% of fair value

    It needs:

    • ○The share price rises to about $204, 80% of what we think the business is worth, so it is no longer cheap enough.checked daily
  2. Attractive→UnattractiveThe price runs 20% above fair value

    It needs:

    • ○The share price rises to about $306, 20% above what we think the business is worth.checked daily
  3. Attractive→UnattractiveThe programme or the concentration turns

    It needs any of:

    • ○AMD names a second manufacturer for Helios volume, or Sanmina's share of the programme falls.watched as it happens
    • ○The ten largest customers pass 75% of sales while total sales stop growing.checked at the next results, due by 1 Feb
  4. Attractive→NeutralThe ramp slips

    It needs any of:

    • ○First-quarter sales fall below $3.8B, or the full-year forecast is cut below $15B.checked at the next results, due by 1 Feb
    • ○Management says the AI rack volume ships in the second half or later.watched as it happens
We would also be wrong if
  • Free cash flow falls below fiscal 2026's $541M while sales grow.

What is coming

  • November 2026 Fiscal 2026 annual report (10-K): Full-year customer concentration, the earn-out's year-end fair value against AMD's receivable, and whether the competition section names anyone.
Every dated event ahead
  • November 2026 Fiscal 2026 annual report (10-K): Full-year customer concentration, the earn-out's year-end fair value against AMD's receivable, and whether the competition section names anyone.
  • January 2027 2027 proxy statement: What the fiscal 2026 bonus and the fiscal 2024 PSUs paid on a year of acquired growth, and how much the chief executive still owns.
  • 1 February 2027 Q1 fiscal 2027 results: Whether first-quarter revenue reaches the $3.8B to $4.1B guide at a 7.0% or better adjusted margin, and whether the accelerated-compute programme starts contributing in the quarter management said it would.
Since we wrote this
Written
6 Sep
Price the thesis read
$198.47
4 Sep
  • 18 NovEarningsAnnual report read: value raised, still AttractiveThe 10-K showed lower customer concentration than at June and a larger earn-out owed to AMD; with the analysts now at the guide's middle, our value rose to about $255.
  • 10 NovEarningsThe call held the late-year ramp; still AttractiveManagement still weights the AI sales to the second half and gave no sign of a slip; re-based on fiscal 2026 the price is about 67% of a value near $225.
  • 9 NovEarningsFiscal 2027 guide of $17.5B to $19.5B moves the view to AttractiveThe guide beat the $16B target and clears the 8-times test at the November price, a third below September's. Conviction stays Medium until the call on rack timing is read.
  • 9 NovThesis reviewStill Attractive: priced at 69% of a value near $196Re-valued in three stages at an 8% return, the method now drawn on every thesis: the price sits past the 20% margin of safety.
  • 3 NovThesis reviewResults date corrected to 9 November; view heldThe thesis expected the fiscal 2027 guide on 2 November; the company scheduled its call for 9 November, so nothing has been tested yet.
  • 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
  • 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
  • 26 SepThesis reviewJournal completed from the written record; view heldThe opening entry lacked the view and conviction the thesis was written with, and what was read; both are filled from the original file and its commit. Nothing in the argument moved.
  • 6 SepThesis writtenNeutral, medium convictionWritten six weeks after the June-quarter results, when the shares fell 17.5% despite a raised full-year earnings forecast.

Also checked 6 times with nothing new to act on.

Sources 23

Where this came from

Sources - 23

Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.

  1. S01primarycompany filing2025-11-13
    the business description and segment structure, the Competition section naming only classes, the fiscal calendar convention, the ZT acquisition subsequent-event note with the five-year manufacturing services agreement, and the strategic-transaction risk factors
    backsC01C02C25C36C49C53C54
  2. S02primarycompany filing2026-07-27
    the end-market and geographic split, the customer concentration table, the purchase price allocation and intangible detail, the segment gross profit table, the full cash flow statement, the tax provision and the AMD partner-of-choice language
    backsC01C02C04C05C06C47C17C20C22C24C36C38C48C53C54
  3. S03primarycompany release2026-07-27
    the balance sheet, the GAAP to non-GAAP reconciliation line by line, the fourth-quarter and full-year guidance including the raise, and the period end date that fixes the 14-week quarter
    backsC05C08C15C16C20C22C24C30C31
  4. S04primarytranscript2026-07-27
    Sanmina Q3 fiscal 2026 earnings call
    the core Sanmina and ZT Systems revenue split and the Q4 split by half, the segment adjusted margins, the non-recurring engineering attribution, the $16B fiscal 2027 statement, the leverage target, and four analyst questions on AMD share, the legacy run-off, the components margin and the India joint venture
    Our core Sanmina business revenue of $2.4 billion grew 17% versus the same period a year ago... Our ZT Systems business revenue of $1.1 billion was at the midpoint of our outlook... we are more confident than ever in our ability to achieve revenue of $16 billion-plus in 2027
    backsC08C09C10C14C19C35C38C46
  5. S05primarycompany release2025-11-03
    fiscal 2025 revenue, margins, adjusted earnings of $6.04 and free cash flow of $478M, and the Q1 fiscal 2026 guidance the company then beat at the top of both ranges
    backsC33
  6. S06primarycompany filing2026-08-05
    the $4.4B AMD paid for all of ZT Systems, the sale of only the manufacturing business to Sanmina, and the $348M earn-out receivable AMD says is probable of being received - the counterparty's own view of the same contract
    backsC03
  7. S07primarycompany filing2026-02-27
    a competitor's filed economics on the same demand - Hardware Platform Solutions at $5.0B and 41% of revenue, its diversified segment growing 1% - and its own risk factor describing aggressive pricing dynamics
    backsC26
  8. S08primarycompany filing2025-08-31
    the second competitor's scale in this end market - Intelligent Infrastructure at 41% of $29.8B against 32% - and group gross margin of 8.9%, which brackets what this work earns
    backsC27
  9. S09primarycompany filing2026-06-30
    a buyer's own capital spending, $173.0B of property and equipment purchased in twelve months against $107.7B - the demand rate measured by the party paying for it
    backsC41
  10. S10primarycompany filing2026-03-31
    the second buyer's equipment build - computer equipment and software at cost up $58.0B in nine months, to $190.9B
    backsC41
  11. S11primarygovernment data2026-05-19
    an independent measure of the scale of data centre load, and its projected path to 22%-33% of commercial building electricity by 2050
    Across all cases, servers alone accounted for an estimated 7% of commercial sector electricity consumption in 2025.
    backsC40
  12. S12primarygovernment data2026-01-13
    the rate behind the driver, measured outside the industry, and the EIA's own refusal to quantify the data centre share of it
    EIA expects U.S. electricity use to grow by 1% this year and 3% in 2027. This increase would mark the first time since 2007 that power demand has risen for four years in a row and the strongest four-year growth period since 2000.
    backsC40
  13. S13primarycompany release2026-07-23
    the physical configuration of the rack Sanmina is preparing for production - 72 accelerators in 18 Open Rack Wide four-GPU compute trays
    Each rack integrates 18 open rack wide-aligned 4-GPU compute trays (72 total GPUs) that provide consistent building blocks for cluster deployment and expansion.
    backsC42
  14. S14secondaryvendor data2026-09-06
    Research pack, SANM, 2026-09-06
    price and enterprise value, the ratio series against six-year medians, the analyst panel split, FINRA short interest at the 2026-08-14 settlement, the measured next-session reaction to the Q3 print, the eight-quarter earnings surprise history, and the Form 4 insider transactions by code and filer
    backsC29C30C32C33C35
  15. S15primarycompany filing2026-01-23
    beneficial ownership as of 2025-12-31, the directors' tenure, the fiscal 2025 bonus plan and its payouts, the PSU design and the fiscal 2023 PSU vesting, and the request for 1.2M more plan shares
    backsC49C50C51
  16. S16primarycompany filing2023-11-16
    the fiscal 2021 to 2023 income statements and cash flow statements: the cost lines per sales dollar, the share repurchases, and the $215.8M received for a stake in the India business
    backsC53C54
  17. S17primarycompany release2026-11-09
    fourth-quarter and fiscal 2026 results, the non-GAAP reconciliation, the balance sheet, cash flow, and the first-quarter and fiscal 2027 outlook
    backsC12C13C14C18C21C28C45C56C57C59
  18. S18primarycompany filing2026-11-09
    trailing revenue, operating profit, stock pay, tax, debt, cash and the diluted share count
    read, but no claim rests on it
  19. S19secondaryvendor data2026-11-09
    the analysts' average revenue and operating profit for the next three fiscal years, and the close
    read, but no claim rests on it
  20. S20primarytranscript2026-11-09
    Sanmina Q4 fiscal 2026 earnings call
    the timing of the cloud and AI revenue, the platforms inside and outside the fiscal 2027 guide, the margin comparison and the one-time tariff recoveries, the capital spending and working capital plan, and five analysts' questions
    from a timing perspective, still expected to be back-end loaded to the second half. Still early days. We wanted to give a full year perspective... The third platform that I mentioned is not included in our guide.
    backsC60C61
  21. S21primarycompany filing2026-11-17
    the full-year customer concentration (ten largest 63%, two customers at 16.7% and 10.1%), the earn-out liability of $261M at fair value with its $150M remeasurement charge, the 55.2M diluted shares, and the five years of operating capital the fair value reads
    backsC07C23C59
  22. S22primarycompany filing2026-11-04
    the $348M earn-out receivable AMD still carries as probable of being received at 26 September 2026
    backsC23
  23. S23secondaryvendor data2026-11-18
    the analysts' average revenue and operating profit for fiscal 2027 and 2028, three analysts each, and the 17 November close
    backsC58C59
Rendered from content/SANM/thesis.json · thesis reviewed 2026-11-18. This page explains a company; it does not say what to pay, when to buy or how much.