Burn-in runs chips hot under power for hours so weak ones fail in the factory, not in a customer's server. Aehr does it to whole wafers before they are cut up.
The company · 2 · What exactly does it sell?
Machines are over half of sales
Each $1 of sales, fiscal 2026
57%30%13%
Burn-in machinesPlates for each chip designService
Each new chip design needs its own plates, so plates sell again after the machine.
The company · 3 · How does it make money?
Big machine orders from a few buyers
$109.3Morders waiting to ship
$150Mforecast for the year to June 2027
70%of last year's sales from five unnamed buyers, who can cancel
Nearly all this year's forecast is already ordered. Next year's is not.
Is it a good business? · 4 · What gives it an edge?
Know-how in powering a whole wafer
125 patents
plus trade secrets; no rival shown powering whole AI chip wafers
Intel
says many defects still get past the wafer stage
Semight
a Chinese rival that won the first patent case
One weak chip can spoil an expensive package, so catching it while still on the wafer saves the whole package.
Is it a good business? · 5 · What is the catch?
Testing after packaging is the main alternative
Teradyne, Advantesttest finished packages instead
AEMsells burn-in for finished AI chips
3 Asian makerssell wafer machines like Aehr's for car chips
Listed rivals: TER · 6857.T · AWX.SI · TRT · 7160.KL
Aehr's lead AI customer stopped testing after packaging, which is the switch the case needs others to make.
Is it a good business? · 6 · What does its record show?
Costs ran ahead of sales for four years
Sales, $M
50.8
65.0
66.2
59.0
50.0
72.0
May 2022May 2023May 2024May 2025May 2026last 12 mo
Overhead31 → 64 centsoverhead per $1, fiscal 2022 to 2026
Per $148.5%kept per $1 after making it, latest quarter
Orders1.4 timesorders against sales last quarter
The company's profit target needs that overhead per dollar to fall by more than half.
Is it a good business? · 7 · How much room is there to grow?
A big testing market; nobody measures burn-in
$15.3Bchip-testing machines a year, 2026 forecast
$20.8Bby 2028
$50.0MAehr's sales last year
Nobody outside the company measures burn-in, so the true size of Aehr's market is unknown.
Is it a good business? · 8 · Who is in charge?
Insiders sold into both rallies
Since 2012Gayn Erickson chief executive
About 4.6%owned by insiders
743,893shares sold in a year, none under a plan
Long-term awards pay on reaching a sales target by June 2027.
The story right now · 9 · Why do AI chips need burn-in?
One bad AI chip spoils a costly package
7xorders, the market says
$33.0Msales last quarter, up 200%
AI chips are packed with memory into very expensive packages. Finding a weak chip before packaging saves the whole package, which is why AI chip makers try wafer burn-in.
The story right now · 10 · Why does the price worry us?
This year is ordered; the price adds years that are not
Buyers order machinesNearly all this year.
Production orders followNot yet.
A buyer can cutAs in 2023.
The price assumes moreYears of orders.
In 2023 it forecast at least $100M, cut twice and delivered $66.2M.
The story right now · 11 · What changed lately?
A record quarter and a forecast tripled
$33.0Msales, up 200%
$109.3Morders waiting to ship
$150Mforecast for the year
-46.7%the shares from 5 to 22 October, with no news
Strong quarter
Nearly all the year is ordered. The production order we waited for has not come.
The story right now · 12 · What is everyone arguing about?
One good year, or several?
Many investors think
The AI burn-in winner, orders up sevenfold
We think
One year ordered, production orders still missing
Against us: a second AI chip maker is testing a machine.
The price and our view · 13 · What is a share worth?
The price sits above even our sunny future
$9.20what we think a share is worth · the last close is of that
$9.20our value
Attractive below $7.36
Neutral
Unattractive above $11.04
$3.44gloomy: growth halves and new money earns little
$9.20our estimate: experts' path for a young, cyclical business
$14.73sunny: growth beats forecasts by a quarter
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 AI burn-in, it is fragile
Priced for successthe kind of bet this is
0.6×if it works
up to 5%kept if it fails
Multiples of the company's value when the bet was last struck, 5 Oct 2026.
If the AI orders stop, shareholders keep little beyond the cash, and the price already pays as if they continue.
The price and our view · 15 · How could we be wrong?
A few buyers, and orders that can slip
A few buyersBiggest
Unnamed, and they can cancel.
Testing after packagingWatch
Still works for rivals.
Chinese copiesShowing now
A rival won the patent case.
Who buys nextUnknown
The lead AI customer is unnamed.
In 2023 one buyer's cuts took the forecast down twice.
The price and our view · 16 · How will we know?
January's report tests the raised forecast
The $150M forecast heldGood sign
A production order from an AI chip makerGood sign
A cut forecast or a cancelled orderBad sign
JAN 20278
Second-quarter fiscal 2027 results, 8 January. Companies report every three months; each report is a test of the story.
Our view also changes on price alone: if the shares fall to within a fifth of our estimate, we move to Neutral.
The price and our view · 17 · So, what do we think?
The only wafer burn-in shown on AI chips, priced past it
$9.20our value
Attractive below $7.36
Neutral
Unattractive above $11.04
UnattractiveHigh conviction
Orders cover nearly all of this year's forecast. But a few unnamed buyers carry it, rivals test finished chips, and the price sits above even our sunny future. Unattractive.
↑ ↓ to move · C for contents
What it does
More about the company
Who it is
Some chips are weak from birth and fail in their first weeks of use. Burn-in finds them by running chips hot and under power for hours in the factory.
Aehr's main machine does this to whole wafers, the discs chips are cut from, nine at once, through a contact plate built for each chip design. A second line, bought in 2024, handles finished chips.
Its biggest AI customer used to test chips after packaging and now does all its burn-in on the wafer instead.
How it makes money
Aehr earns most of its money selling the machines: $28.7M of last year's $50.0M. The contact plates made for each chip design brought in $14.9M, and are bought again whenever a customer changes its chip, so management treats them as repeat business. Service brought the rest.
The money comes from very few hands: five customers paid 70% of last year's sales, and none is named.
Growth and profit
What growth costs
Aehr has been spending ahead of its sales. Research, selling and office costs took 31 cents of each sales dollar in fiscal 2022 and 64 cents last year, on the same sales. The profit forecast needs sales to rise much faster than those costs.
Its record
The numbers
Is it growing?Fast
Sales grew 27% over the twelve months to Sep 2026, faster than the −15% in the year to May 2026.
Is it profitable?Losing
Operating margin is −5.6%, down from 15% in the year to May 2024.
Does growth pay off?Costly
Running costs take 47¢ of each sales dollar, up from 31¢ in the year to May 2022.
Does it earn good returns?Low
Shareholders' money earned 1.0% a year, down from 30% in the year to May 2022.
How long can it keep going?Long runway
It used $6.81M of cash over the last twelve months, so the $105M it holds lasts about 15.4 years at that pace.
How steady is it?Swingy
Yearly sales growth has ranged from −15% to +206% over the last 5 years, with sales falling in 2 of them.
Are shares being diluted?Diluting
The share count rose 21% in 4.3 years, and pay in shares costs 9.7% of sales.
Why it wins
The evidence
An AI chip is now packaged with several other chips and stacks of memory, so one weak chip can spoil an expensive package. Intel has said many defects still get past the wafer stage. Aehr says catching them earlier pays many times over, though that figure is its own.
Its edge is know-how in powering a whole wafer at once, kept mostly secret. Nobody else is shown doing it for AI chips; for car chips, a Chinese rival sells a similar machine and won the first patent ruling.
Who could beat it
The rivals in full
The main alternative is to test after packaging, as Teradyne's and Advantest's machines do; Aehr's lead AI customer stopped doing that. For finished AI chips, rivals such as AEM in Singapore already sell burn-in machines, and AEM says a large AI chip designer chose it. For car chips, three Asian makers sell wafer machines like Aehr's.
Each rival
System-level test after packaging, from Teradyne and AdvantestTER6857.T
The screening step: catch weak chips after they are packaged, which is what the lead AI customer did before moving to the wafer.
Semight and Nexustest nine-wafer burn-in systemsprivate
The silicon carbide wafer-level system itself, at Chinese prices, for the electric-vehicle chip buyer.
AEM's automated burn-in for AI processorsAWX.SI
The Sonoma package-level sale to an AI chip maker; AEM says a major fabless AI maker chose it as plan of record.
Trio-Tech's burn-in boards and systemsTRT
Package-level burn-in equipment and outsourced burn-in services for chip makers in Southeast Asia.
The silicon carbide wafer-level system, six to eight wafers at a time.
How big it could get
The market and who has signed
The market
Chip-testing machines of all kinds sell for about $15.3B a year, on the industry group SEMI's forecast for 2026, rising to $20.8B by 2028. Aehr sold $50.0M last year.
Nobody outside the company measures how much of that is burn-in, so the true size of Aehr's market is unknown. What is measured is the driver: TSMC, which makes most AI chips, expects its AI chip sales to grow by well over half a year until 2029.
Who has signed
Since May the lead AI customer ordered $22M of wafer machines, a large internet company $41M of the finished-chip machines, and a light-chip customer two more systems. Car-chip customers added more than $8M.
None of these buyers is named, and the company says most orders can be cancelled with small penalties.
Who runs it
Leaders, pay and the cash
Who runs it
Gayn Erickson has run Aehr since 2012, through the car-chip boom and the turn to AI. The founder, Rhea Posedel, is still chairman.
Insiders own about 4.6%. Bonuses paid nothing last year, and long-term share awards pay on reaching a sales target by June 2027. Insiders sold 743,893 shares in the past year, none under a pre-set selling plan, and bought none.
What they did with the money
Aehr sold new shares as its price rose, raising $97.4M in the year to May 2026, most of it at about $74 a share in April. It has no debt; cash was $105.1M in September after buying long-lead parts, and that cash funds the growth.
The last report
The latest report
Aehr's sales hit a record $33.0M and it raised its forecast to $150M
What it did to the thesis Strong quarter
The forecast rose instead of falling, the margin recovered and orders beat sales, so nearly all of the year is ordered. The production order we waited for has not come. The view moves from Unattractive to Neutral.
The report
In the quarter to 25 September 2026, Aehr's sales rose 200% to a record $33.0M and it made $7.2M. The share of each dollar kept after production costs recovered to 48.5% from 35.3% last year. Orders were $46.7M, 1.4 times sales.
Orders waiting to ship reached $109.3M, and Aehr raised its forecast for the year to June 2027 to $150M, triple last year. A large AI chip maker has decided to buy one wafer machine to test in a production setting; a production order would come later.
What's happening now
The background
The shares rose sharply over the past year, all of it from investors paying more for each dollar of sales. They hit $145.61 in August, fell by almost half in two weeks with no news, and again between 5 and 22 October, to $55.65. They closed at $51.47 on 2 November, before the results.
The results came after that close. The next test is the second quarter, due in January.
Where we disagree
Both sides in full
The market
Aehr is the AI burn-in winner: a forecast raised to triple last year, orders up sevenfold and the only wafer machine shown working on AI chips.
We believe
This year looks ordered. What the price adds on top rests on customers who have not ordered for production.
The argument
In 2023 Aehr forecast at least $100M of sales on one car-chip customer, cut that twice and delivered $66.2M; the shares fell from $54.10 to $9.83. Today's orders are cancellable, the buyers unnamed, and insiders sold into both rallies.
Against us: the forecast rose, orders beat sales, and a second large AI chip maker is buying a machine to test on its production line. If it orders for production, much of our view falls away.
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.
The three cases
Bear case$3.44
Growth runs at half the analysts' path, 30%, 10% and 20%, the margin settles a fifth lower at 8.3%, and new capital earns only the 8% rate: the downturn test.
Our value$9.20
Revenue grows 60%, 21% and 40% over three years, the analysts' path (C51), held to at most 60% a year, then fades toward 4%, closing 30% of the gap each year; the margin moves to 10.4%, the analysts' year-three figure after stock pay and on the filed basis (4.4% less).
Bull case$14.73
Growth runs a quarter above the analysts' path, 75%, 26% and 50%, the margin reaches 11.4%, and new capital earns 15%.
The bet
Seen as a bet
This is a bet that wafer burn-in becomes standard for AI chips, with a second large chip maker in production. Even then, valued at a rich but not extreme price for each dollar of sales, the company would be worth a little over half of the 5 October price this bet was struck at.
If the AI orders stop once the current orders ship, what is left is the cash and a small car-chip and light-chip business, a small fraction of today's value.
What could go wrong
Every risk
The risks
A few buyers
Three customers were half of last year's sales and the forecast rests on two unnamed buyers whose orders can be cancelled.
Rivals for finished chips
AEM and others already sell burn-in for finished AI chips, and one large designer chose AEM.
Margins under pressure
Gross margin recovered to 48.5% in the latest quarter from 35.3% last year, but suppliers of power parts raised prices 40%, management says.
Chinese copies
A Chinese rival sells a similar wafer machine and won the first court round; the appeal is pending.
Orders can slip
In fiscal 2024 one customer's slower orders cut the forecast by a third within nine months.
What we still do not know
Who the lead AI customer is, and how much more it plans to buy after the $22M order.
How big the burn-in market is: nobody outside the company measures it.
Whether the second AI chip maker in trials will order for production, and when.
Whether any memory maker will commit; management needs one before it builds the product.
What changes our view
Every route to a new view
Every route to a new view
→ Neutral
The share price falls to about $11.04, back within 20% of what we think the business is worth.
→ Attractive
The share price falls to about $7.36, 80% of what we think the business is worth.
The next test
It supports the case if
Revenue of $35M or more, bookings at or above revenue, non-GAAP gross margin of 45% or more and the $150M guide held: the price is paying about what is ordered.
It pressures the case if
Revenue under $35M, bookings below revenue, non-GAAP gross margin under 45% or the guide cut: the ordered base is not turning into sales.
We would also be wrong if
A memory maker commits to wafer burn-in with an order.
The verdict
Scores and how we weighed it
The scores
Growth
2 of 4
Business quality
3 of 4
Financial position
3 of 4
Execution record
2 of 4
Per share capture
3 of 4
Potential
2 of 4
Risk profile
2 of 4
Valuation
1 of 4
Evidence quality
3 of 4
How we weighed it
ValueNegative
The price sits far above our value and above the bull case.
BusinessMixed
A mixed business: a unique tool, with sales that fell for two years.
GrowthPositive
Orders cover nearly all of this year's raised forecast.
RisksNegative
A few buyers carry it, rivals offer burn-in for finished chips and margins are under pressure.
RivalsMixed
No filed rival compares directly; the rise came from the price, not sales per share.
The groupMixed
Chip-equipment names move together on news about chipmakers' spending.
Latest newsPositive
The last report raised the forecast and recovered the margin.
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.