FormFactor
The maker of the tiny bed of springs that touches every chip on a silicon wafer, so a chipmaker can find the bad ones before paying to package them.
Our viewUnattractive Medium conviction
The scores
The profile · 1-4
Growth3/4 ↑
Four quarters above 30% and organic, but it narrows to HBM at one customer and the record is under two years.
Business quality3/4 →
Qualification is a real barrier with filed evidence, but it did not stop the competitor owning GPU wafer test outright.
Financial position4/4 ↑
Net cash of $337 million, one amortising building loan, an undrawn revolver, and the plant funded from operations.
Execution record3/4 ↑
Four straight beats and the prior target model reached early, but the record of this margin level is two quarters.
Per share capture3/4 ↓
Share count creeps up 2.7% a year, outrun by growth; the offsetting buyback is paused for the plant, not abandoned.
Potential2/4 ↑
GPU, custom ASIC and optics are large, but none is contracted and all three are somebody else's ramp decision.
Risk profile3/4 →
One customer at 24.3% and 65% of revenue shipped to two jurisdictions - disclosed, long-standing and survivable.
Valuation2/4 ↓
Not Fair: meeting the company's own 2030 plan and exiting at 25 times now returns slightly less than nothing, so the good case is required.
Evidence quality3/4 →
Everything central is filed, but the 51% baseline, the GPU timing and the 2030 model exist only on one call.
The Verdict
We rate FormFactor unattractive. It makes probe cards, custom contacts that test every chip on a wafer before it is packaged. The business is solid, and demand for the memory used beside AI chips lifted its margin sharply. But management itself puts the lasting margin below the latest figure, and memory makers are already shifting away from the product that drove it. The shares have more than tripled in a year, and meeting the company's own long-term plan exactly would not be enough. So it is full, not cheap. The view hinges on new products, not yet shipped, beating that plan.
The long caseAI chips need more high-bandwidth memory, which takes more and harder testing on each wafer; but memory makers are shifting away, so FormFactor's new products must more than replace it.
The evidence under this - 8 claims
C01facthigh confidence
A probe card is a custom-built array of microscopic contacts on a circuit board that presses onto a finished wafer so a tester can exercise every chip before it is cut out and packaged.
C03facthigh confidence
Second-quarter 2026 gross margin was 50.7% on a GAAP basis and 53.3% on a non-GAAP basis, against 37.3% and 38.5% a year earlier.
C40facthigh confidence
The shares returned 257% over the 252 sessions to 25 September 2026 and sit 19.1% below the high of that window.
- 2026-09-27editedRe-struck on the refreshed pack's 252-session window to 25 September 2026: the return fell to 257% as the window's start moved off the low, and the distance below the high narrowed to 19.1%.
C04management claimmedium confidence
Management says the recurring gross margin is now about 53% at Q3 volumes, up from about 51% to 51.5% a quarter earlier, and expects 2027 gross margin at similar levels as higher volume and efficiency offset the Farmers Branch ramp; the reported 56.5% in Q3 included about 380 basis points of tariff refunds and other favourable items that will not recur.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- Management also says Farmers Branch will be accretive to gross margin once ramped, so 51% is presented as a floor rather than a ceiling.
- The company has beaten its own gross margin guidance in each of the last four quarters, most recently by 32 basis points above the high end.
- The 51% figure appears only on the call; no filing states it, so it is a number that can be revised without amending anything.
- The 53% figure is stated on the call and in no filing, so it can be revised without amending anything.
- Management says a new factory ramp typically costs margin and that it expects to offset that with volume; the Q4 outlook of 54.0% includes about 70 basis points of refunds, so it is about 53.3% on the baseline.
- 2026-10-28editedQ3 2026 release shows a margin and a Q4 outlook above the 51% baseline management gave; the claim stays as management's statement and records the counter-evidence
- 2026-10-29editedthe Q3 call raised the baseline gross margin to about 53% from about 51% and put the Q3 one-offs at about 380 basis points
C09management claimhigh confidence
Management guides third-quarter DRAM revenue comparable to the second quarter but with a significant shift from HBM to DDR, because customers earn more starting DDR5 wafers than HBM4 wafers.
against this
- A mix shift inside a flat DRAM line is not a demand loss, and management frames it as customers optimising profit in a shortage rather than as weakness.
- DDR5 probe cards are still FormFactor's product; the company sold DDR into record DRAM quarters through the second half of 2025 before HBM accelerated.
C22estimatemedium confidence
Meeting the 2030 target exactly and exiting at 25 times it returns about minus 1% a year from $129.69, the 25 September 2026 close; at 30 times about 3.5% a year, at 35 times about 7.3%, and about 39 times is needed for 10%.
derived$5.00 of 2030 non-GAAP earnings per share at 25, 30 and 35 times gives $125, $150 and $175, compounded over the 4.27 years from 25 September 2026 to the end of 2030; 10% a year over the same span needs about $195, or 39 times $5.00.
against this
- If the target is beaten as the prior one was, the arithmetic runs on a larger number: $5.70 in 2030 at 28 times is about 5% a year.
- The exercise assumes no capital returned; a resumed buyback after Farmers Branch would add to the per-share outcome.
- 2026-09-27editedRe-struck at the 25 September 2026 close of $129.69: at 25 times the plan met now returns slightly less than nothing a year, and about 39 times is needed for 10%.
C21estimatemedium confidence
At $129.69, the 25 September 2026 close, the enterprise is about $9.77 billion: roughly 36 times the non-GAAP operating profit the third-quarter guide annualises to at the 51% baseline, and 26 times the 2030 target's $5.00.
derivedMarket value $10.11bn from the pack less net cash of $337m (cash and investments $349.0m against $11.6m of debt, 10-Q). Q3 revenue guide midpoint $270m annualised to $1.08bn, at 51% non-GAAP gross margin less the annualised $70m quarterly non-GAAP opex guide, gives $271m. $129.69 divided by the $5.00 target.
against this
- Annualising one guided quarter is a snapshot of a business growing double digits sequentially, so the forward multiple falls fast if the growth holds.
- The 2030 target is the company's own floor-shaped plan and the street is already above its implied path, so 26 times a stale target overstates what is being paid for.
- 2026-09-27editedRe-struck at the 25 September 2026 close of $129.69 after a 12.8% rise from $114.95: market value from the refreshed pack, net cash and the annualised guide unchanged.
C05management claimhigh confidence
The third-quarter guide of 54% non-GAAP gross margin included tariff refunds; the quarter's refunds were $8.6M, about 300 basis points, and about $2M more (around 70 basis points) is expected in Q4, after which management expects no ongoing benefit from refunds.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- The refunds are the return of tariffs actually paid from 2025 into 2026 and later ruled unlawful, so they are a recovery of real cost rather than a windfall.
- Management disclosed the amount unprompted in the prepared remarks rather than letting the 54% stand, which is the opposite of the behaviour this claim is used to worry about.
- 2026-10-29editedthe Q3 call gave the refund actually received ($8.6M, about 300 basis points) and the roughly $2M still due in Q4
The company
Chipmakers build hundreds of chips on one silicon wafer. Before cutting them apart and paying to package each one, they need to know which work. A probe card is the tool: a circuit board of microscopic spring contacts that press onto the wafer so a tester can check every chip.
Each card is built for one customer's chip design. The biggest buyers are the memory maker SK hynix, 24.3% of revenue in the June quarter, and Taiwan Semiconductor, 11.0%.
The evidence under this - 4 claims
C01facthigh confidence
A probe card is a custom-built array of microscopic contacts on a circuit board that presses onto a finished wafer so a tester can exercise every chip before it is cut out and packaged.
C26facthigh confidence
Probe cards are customised to each customer's wafer and chip design, and the annual report names the high capital investment and the time and cost of customer evaluation as significant barriers to entry in MEMS probe cards.
C10facthigh confidence
SK hynix was 24.3% of second-quarter 2026 revenue and 26.8% of the first half, and was a greater-than-10% customer in all eight quarters disclosed in the fiscal 2025 annual report.
C43facthigh confidence
Taiwan Semiconductor was 11.0% of second-quarter revenue, and management says that figure contained no GPU business at all.
How it makes money
Cards wear out and every new chip design needs its own, so customers buy again and again, on ordinary orders placed against the wafers they plan to make. There is no backlog and no minimum volume, so a customer can change the mix from one quarter to the next. A smaller business sells the machines that hold a wafer under test; it had a record quarter at $48.5M.
The evidence under this - 2 claims
C09management claimhigh confidence
Management guides third-quarter DRAM revenue comparable to the second quarter but with a significant shift from HBM to DDR, because customers earn more starting DDR5 wafers than HBM4 wafers.
against this
- A mix shift inside a flat DRAM line is not a demand loss, and management frames it as customers optimising profit in a shortage rather than as weakness.
- DDR5 probe cards are still FormFactor's product; the company sold DDR into record DRAM quarters through the second half of 2025 before HBM accelerated.
C33facthigh confidence
Systems revenue reached a record $48.5 million in the second quarter, up 44.1% on the year and 74% on the quarter, driven by Triton, the co-packaged optics test platform.
Where the money comes from, stream by stream
| Stream | Share of sales (Q2 2026) | How it earns it |
|---|---|---|
| Probe cards: Foundry & Logic | 47.2% | Custom probe cards for processors, networking, mobile and graphics chips, sold to foundries and chip designers for each new design. |
| Probe cards: DRAM | 32.9% | Probe cards for memory wafers, high-bandwidth and standard DRAM, sold mainly to the three large memory makers. |
| Probe cards: Flash | 1.1% | Probe cards for flash memory wafers, a small and shrinking line. |
| Systems | 18.8% | Probe stations, thermal and cryogenic systems and the Triton optics test platform, sold to labs and fabs as equipment. |
Who runs it
Mike Slessor has run FormFactor since 2014, after arriving with a company it bought; Aric McKinnis, its long-time controller, became finance chief in 2025. Insiders own under 1% of the shares and have been selling, not buying. Most of the cash bonus rides on adjusted operating profit each quarter; half the share awards pay only if the stock beats other chip stocks over three years. Nothing rewards cash flow or profit per share.
The evidence under this - 3 claims
C49facthigh confidence
Mike Slessor has been president since October 2013 and chief executive since December 2014, having joined as chief executive of MicroProbe when FormFactor bought it in October 2012; Aric McKinnis, corporate controller since August 2019, became chief financial officer in August 2025, replacing Shai Shahar, chief financial officer since 2018.
C50facthigh confidence
At 16 March 2026 the nine directors and executive officers as a group beneficially owned 615,370 shares, under 1% of the 77,954,440 outstanding; Mike Slessor's 486,669 shares, about 0.6%, were the largest insider holding; BlackRock (15.3%), Vanguard (13.6%) and EARNEST Partners (7.8%) were the holders above 5%, as last reported on their 13G filings.
against this
- BlackRock's 15.3% is taken from a 13G reflecting holdings at 31 December 2023, so the large-holder figures are older than the insider ones.
- Insiders made 70 open-market sales and no purchases in the year to 12 September 2026 (C42), so the insider figure has fallen since the record date.
C51facthigh confidence
FormFactor's 2025 annual bonus paid 70% on quarterly adjusted operating income against plan and 30% on corporate goals (revenue, gross margin and expenses; growth; quality), and scored 59.9% with the goals at zero; long-term awards are half time-vested RSUs and half PRSUs on three-year shareholder return against the S&P Semiconductors Select Industry Index.
against this
- Adjusted operating income excludes stock pay, restructuring and acquisition costs, so the bonus never sees the $32 million to $40 million of 2026 restructuring (C38) or the dilution in C24.
- No metric pays on cash flow, return on capital or a per-share figure, and half the equity vests on time alone; the 2022 PRSUs still paid 77% after a negative 12% three-year return.
Why it wins, and who could beat it
Why customers come back
Every card is designed around one customer's chip and has to be approved on that customer's own test line. Switching supplier halfway through a product generation means approving a new one, at the customer's cost and risk, and the annual report names that time and cost as a major barrier.
It shows in who stays: SK hynix has bought more than a tenth of FormFactor's sales in every one of the eight quarters the annual report discloses.
The evidence under this - 2 claims
C26facthigh confidence
Probe cards are customised to each customer's wafer and chip design, and the annual report names the high capital investment and the time and cost of customer evaluation as significant barriers to entry in MEMS probe cards.
C10facthigh confidence
SK hynix was 24.3% of second-quarter 2026 revenue and 26.8% of the first half, and was a greater-than-10% customer in all eight quarters disclosed in the fiscal 2025 annual report.
The threat
Technoprobe, a rival using similar technology, holds the business of testing graphics processors for artificial intelligence; FormFactor only qualified and shipped its first production cards there in the third quarter of 2026. Technoprobe's sales grew 42.4% in the first half of 2026, against FormFactor's 31.9%. In ordinary memory, management admits there is lots of competition.
The evidence under this - 4 claims
C27factmedium confidence
Technoprobe designs and produces probe cards on MEMS technology similar to FormFactor's and holds the GPU wafer test business FormFactor has only just qualified into.
C13management claimhigh confidence
GPU probe cards were primarily a competitor's business; FormFactor shipped probe cards for one GPU design to the leading foundry in multimillion-dollar quantities in Q3 2026 and expects significant share gains in 2027, constrained by its ability to produce in volume.
sourceFormFactor Q2 2026 earnings call transcriptprimaryTechnoprobe S.p.A corporate websiteprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- This is a damaging admission volunteered by the chief executive, which is the most credible kind of company statement; the incentive runs the other way.
- Qualification at a GPU maker is a multi-year gate the company has now passed, so the same fact reads as an option that is not yet in any reported number.
- 2026-10-29editedthe Q3 call confirmed the first production GPU shipments (one design, multimillion-dollar volumes) and that supply, not demand, limits custom-chip volume
C28factmedium confidence
Technoprobe reported first-half 2026 revenue of EUR 464.1 million, up 42.4%, against FormFactor's 31.9% over the same half.
C09management claimhigh confidence
Management guides third-quarter DRAM revenue comparable to the second quarter but with a significant shift from HBM to DDR, because customers earn more starting DDR5 wafers than HBM4 wafers.
against this
- A mix shift inside a flat DRAM line is not a demand loss, and management frames it as customers optimising profit in a shortage rather than as weakness.
- DDR5 probe cards are still FormFactor's product; the company sold DDR into record DRAM quarters through the second half of 2025 before HBM accelerated.
The rivals: all private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | How far along | Priced against it |
|---|---|---|---|
| Technoprobe S.p.APriced against it: Listed abroad | On sale | Listed abroad | |
| Micronics Japan Co., Ltd. and Japan Electronic Materials CorporationPriced against it: Listed abroad | On sale | Listed abroad | |
| Siemens EDA Tessent MemoryBISTPriced against it: Listed abroad | On sale | Listed abroad |
How big it could get
How big this could get
Spending on the fast memory stacked beside artificial-intelligence chips is forecast to grow from $33B in 2025 to $86B in 2027. Nobody publishes how much of that goes on probe cards, so the market cannot be sized; FormFactor billed $167.9M for memory cards in the first half of 2026. A newer line, testing chips that send data as light, management puts at about $400M a year by 2030.
The evidence under this - 3 claims
C31estimatemedium confidence
HBM revenues are forecast to rise from $33 billion in 2025 to $86 billion in 2027, a compound rate of 60.5% - Gartner's figure, carried in SK hynix's SEC-filed prospectus.
against this
- Forecasts of this steepness are made at a cyclical extreme and the same prospectus forecasts the whole memory market growing 192.7% in a single year, which is not a rate anything sustains.
- FormFactor's own third-quarter guide already shows customers rotating wafer starts away from HBM4 to DDR5, which is the demand mix moving against the forecast's composition inside the forecast period.
C32facthigh confidence
First-half 2026 DRAM probe card revenue was $167.9 million and Foundry and Logic was $233.0 million.
C34management claimmedium confidence
Co-packaged optics revenue is expected to exceed $20 million in 2026, against a served market management puts at about $400 million by 2030.
against this
- Management raised this from a $10m to $20m range set at the start of the year and says it cannot forecast 2027 quarter by quarter, so the number is moving faster than the company can bound it.
- The $400 million served market is FormFactor's own sizing from its investor day and no party outside the company publishes a co-packaged optics test market at all.
Where the growth is coming from, business by business
Where the growth is coming from
| Business | Share of revenue | Growth: last quarter | Last half-year | Last year |
|---|---|---|---|---|
| Foundry & Logic Product Group | 47% | +22.4% | +26.1% | +11.3% |
| DRAM Product Group | 33% | +49.0% | +58.6% | +34.9% |
| Systems Product Group | 19% | +44.1% | +11.5% | +8.0% |
| Flash Product Group | 1% | -48.4% | -11.8% | +21.7% |
| Whole company | 100% | +31.9% | +31.9% | +18.0% |
The filed split through 2026-06-27. These rates are not carried forward: the reading below says why.
How we read it: The blend, narrowed: DRAM's June-quarter rise was three-quarters HBM, a market Gartner sees compounding 60%, but Q3 DRAM is guided flat as HBM rotates to DDR and one customer is a quarter of revenue: 3 holds.
The results
The numbers
Is it growing?Steady
Sales grew 18% over the twelve months to Jun 2026, faster than the 2.8% in the year to Dec 2025.
Is it profitable?Healthy
Operating margin is 13%, about where it was (12%) in the year to Dec 2023.
Does growth pay off?Flat
Running costs take 30¢ of each sales dollar, up from 29¢ in the year to Dec 2021.
Does it earn good returns?Low
Each dollar put into the business earns 9.5% a year after tax, about where it was (10%) in the year to Dec 2021.
Is it making cash?Strong
It turned 15% of sales into free cash, $136M over the last twelve months.
Is the balance sheet safe?Net cash
It holds $346M of cash against $11.6M of debt, $334M more cash than it owes.
Are shares being diluted?Stable
The share count rose 0.6% in 4.5 years, and pay in shares costs 4.0% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-04. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Jun 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $770M | $748M | $663M | $764M | $785M | $902M |
| Sales growth | +11% | −2.8% | −11% | +15% | +2.8% | +18% |
| Profit | ||||||
| Gross margin | 42% | 40% | 39% | 40% | 39% | 43% |
| Operating margin | 13% | 7.3% | 12% | 8.5% | 7.3% | 13% |
| Net margin | 11% | 6.8% | 12% | 9.1% | 6.9% | 13% |
| What growth costs, per sales dollar | ||||||
| Research and development | 13% | 15% | 17% | 16% | 15% | 13% |
| Sales and marketing | - | 7.8% | 7.9% | 7.0% | 6.5% | 6.0% |
| General and admin | - | 6.8% | 8.7% | 8.8% | 8.0% | 7.0% |
| Pay in shares | 3.8% | 4.2% | 5.8% | 5.2% | 4.9% | 4.0% |
| Capital spending | 8.6% | 8.7% | 8.4% | 5.0% | 13% | 4.8% |
| Cash | ||||||
| Cash from operations | $139M | $132M | $64.6M | $118M | $115M | $180M |
| Free cash flow | $72.9M | $66.5M | $8.57M | $79.1M | $11.7M | $136M |
| Balance sheet | ||||||
| Cash and short-term investments | $276M | $238M | $328M | $360M | $275M | $346M |
| Debt | $24.4M | $15.4M | $14.4M | $13.3M | $12.2M | $11.6M |
| Net cash | $252M | $223M | $314M | $347M | $263M | $334M |
| Returns and shares | ||||||
| Return on capital | 10% | 5.8% | 8.7% | 6.0% | 4.6% | 9.5% |
| Shares (diluted) | 79.1M | 78.2M | 78.2M | 78.4M | 78.3M | 79.6M |
What growth costs
Growth has not yet made FormFactor cheaper to run. Its research and overhead costs have risen about as fast as sales over six years, shrinking as a share in good years and swelling in bad ones. The recent jump in profit came from a better gross margin, not from spreading costs. The company's 2030 plan needs those costs to fall well below today's share, which has not happened before.
The evidence under this - 2 claims
C47facthigh confidence
Per sales dollar, research and development was 12.8 cents in fiscal 2020, 17.5 in 2023, 14.7 in 2025 and 13.4 over the four quarters to June 2026; selling, general and administrative 16.6, 20.1, 17.0 and 15.3; stock pay 3.4, 5.8, 4.9 and 4.0; capital spending 8.1, 8.4, 13.2 and 4.8.
sourceFormFactor Form 10-K, fiscal year ended 31 December 2022primaryFormFactor Form 10-K, fiscal year ended 27 December 2025primaryFormFactor Form 10-Q, quarter ended 27 June 2026primary
against this
- Factory start-up costs for Farmers Branch are a separate operating line outside both, $11.9 million or 2.4% of revenue in the first half of 2026 (S02), so the two lines understate what the plant costs today.
- Management guides 2026 capital spending to $140 million to $170 million (C19), so the trailing 4.8 cents is the low point before the second-half build, not a new level.
C48inferencemedium confidence
Growth has not yet bought operating leverage: research plus selling, general and administrative cost rose from $204.1 million in fiscal 2020 to $258.6 million over the four quarters to June 2026, 27% against 30% revenue growth, so it still took about 29 cents of each sales dollar; the margin gain since came from gross margin.
derivedC47's cost lines per sales dollar, revenue of $693.6 million in fiscal 2020 and $902.2 million over the four quarters to June 2026, and gross margin of 41.5% against 43.1% over the same span, all from the numbers card
against this
- Within a year the costs are largely fixed: they took 37.6 cents of each dollar in 2023 when sales fell 11%, and fell per dollar in the recent rise, so leverage appears in an up-cycle even if it has not accumulated across one.
- The 2030 target model sets operating expenses at 23% of revenue (C20), which would be the first sustained step down in the six-year record.
The last report (29 Jul)
Quarter ended June 2026, filed figures.
32%
Sales growth
against the same quarter a year earlier
14.2%
Growth from the quarter before
against 5.1% the quarter before that
22¢
Operating profit
per dollar of sales
In the September 2026 quarter revenue grew 38.4% to $280.5M, above the top of its guide, and the adjusted gross margin was 56.5% against 53.3% in June. Foundry and logic sales rose 10% on the quarter and the optical test line rose more than 20%.
The call said about 300 basis points of the margin were tariff refunds, so without them it was between 52.5% and 53%. Management raised its lasting margin to about 53% from 51%, and the December guide of 54% includes about 70 basis points of refunds.
What it did to the thesisIt beat the test we set, and the margin without refunds was above the lasting level we had used, weakening that part of the case. We hold the view because the price still asks for more than the company's own plan, and a second quarter is yet to come.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $0.82 against $0.61 expected; $0.71 as filed, the gap being what the company adds back.
The evidence under this - 4 claims
C54facthigh confidence
In the third quarter of 2026 FormFactor's revenue was $280.5 million, up 38.4% on the year and 8.6% on the quarter, above the $270 million plus or minus $10 million guide. Non-GAAP gross margin was 56.5% (GAAP 54.7%) against 53.3% in Q2, non-GAAP earnings were $1.04 a share, Foundry and Logic grew 10% on the quarter and Systems grew more than 20%. The fourth-quarter outlook is revenue of $285 million plus or minus $10 million and non-GAAP gross margin of 54.0% plus or minus 1.5 points.
sourceFormFactor Q3 2026 results release, Exhibit 99.01 to the 8-KprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- Of the 56.5%, about 300 basis points were tariff refunds ($8.6M) and about 380 basis points in all were refunds and favourable items; without them the margin was between 52.5% and 53%, about 150 basis points above the 51% baseline given last quarter (the call).
- The chief executive says productivity gains from the existing footprint are moderating as it nears capacity limits, so growth from here leans on the Farmers Branch plant.
- 2026-10-29editedthe Q3 call split the refunds out of the 56.5% margin, answering the question the release left open
C04management claimmedium confidence
Management says the recurring gross margin is now about 53% at Q3 volumes, up from about 51% to 51.5% a quarter earlier, and expects 2027 gross margin at similar levels as higher volume and efficiency offset the Farmers Branch ramp; the reported 56.5% in Q3 included about 380 basis points of tariff refunds and other favourable items that will not recur.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- Management also says Farmers Branch will be accretive to gross margin once ramped, so 51% is presented as a floor rather than a ceiling.
- The company has beaten its own gross margin guidance in each of the last four quarters, most recently by 32 basis points above the high end.
- The 51% figure appears only on the call; no filing states it, so it is a number that can be revised without amending anything.
- The 53% figure is stated on the call and in no filing, so it can be revised without amending anything.
- Management says a new factory ramp typically costs margin and that it expects to offset that with volume; the Q4 outlook of 54.0% includes about 70 basis points of refunds, so it is about 53.3% on the baseline.
- 2026-10-28editedQ3 2026 release shows a margin and a Q4 outlook above the 51% baseline management gave; the claim stays as management's statement and records the counter-evidence
- 2026-10-29editedthe Q3 call raised the baseline gross margin to about 53% from about 51% and put the Q3 one-offs at about 380 basis points
C05management claimhigh confidence
The third-quarter guide of 54% non-GAAP gross margin included tariff refunds; the quarter's refunds were $8.6M, about 300 basis points, and about $2M more (around 70 basis points) is expected in Q4, after which management expects no ongoing benefit from refunds.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- The refunds are the return of tariffs actually paid from 2025 into 2026 and later ruled unlawful, so they are a recovery of real cost rather than a windfall.
- Management disclosed the amount unprompted in the prepared remarks rather than letting the 54% stand, which is the opposite of the behaviour this claim is used to worry about.
- 2026-10-29editedthe Q3 call gave the refund actually received ($8.6M, about 300 basis points) and the roughly $2M still due in Q4
C13management claimhigh confidence
GPU probe cards were primarily a competitor's business; FormFactor shipped probe cards for one GPU design to the leading foundry in multimillion-dollar quantities in Q3 2026 and expects significant share gains in 2027, constrained by its ability to produce in volume.
sourceFormFactor Q2 2026 earnings call transcriptprimaryTechnoprobe S.p.A corporate websiteprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- This is a damaging admission volunteered by the chief executive, which is the most credible kind of company statement; the incentive runs the other way.
- Qualification at a GPU maker is a multi-year gate the company has now passed, so the same fact reads as an option that is not yet in any reported number.
- 2026-10-29editedthe Q3 call confirmed the first production GPU shipments (one design, multimillion-dollar volumes) and that supply, not demand, limits custom-chip volume
What management did with the money
No shares were bought back in the first half: the cash is going into a new factory at Farmers Branch, with $140M to $170M of spending planned for 2026. There is $349.0M of cash and investments against $11.6M of debt. The company's own 2030 plan implies 85 to 89 million shares, against 79.6 million now.
Insiders in 2026Q2: 12 purchases and 44 sales, from their Form 4 filings.
The evidence under this - 4 claims
C25facthigh confidence
No shares were repurchased in the first half of 2026, $70.9 million of authorisation remains, and management says cash is being prioritised to the Farmers Branch ramp.
C19management claimhigh confidence
Management guides 2026 cash capital expenditure for Farmers Branch and capacity additions to $140 million to $170 million, which leaves $115 million to $145 million for the second half.
against this
- Texas has awarded a $24.2 million grant from its Semiconductor Innovation Fund against this spend, and management says the site is largely self-funded from operating cash flow.
- First-half operating cash flow was $106.8 million and rising, so the company can carry the spend without borrowing; the $150 million revolver is undrawn.
C23facthigh confidence
Cash and investments were $349.0 million against total debt of $11.6 million at 27 June 2026, with the $150 million revolving facility undrawn.
C24estimatelow confidence
The 2030 target's own arithmetic implies about 85 million to 89 million diluted shares, against 79.6 million in the second quarter of 2026.
derived$1.6bn at a 32% non-GAAP operating margin (55% gross less 23% opex) is $512m, plus about $14m of interest income on the current cash balance, taxed at the guided 15% to 19%, divided by the $5.00 target. Diluted share count from the Q2 income statement.
against this
- The target model may simply assume more interest income or a lower tax rate than the current guide, either of which lowers the implied share count.
- The company has $70.9 million of buyback authorisation and an explicit policy of offsetting dilution, which it has used before; the pause is stated as temporary.
The price, and where we disagree
In short. Full. Meeting the company's own 2030 plan exactly would not be enough. The price needs new products, which have not shipped yet, to beat it.
Share price
$129.69
close,
Price for each $1 of yearly profit
87.6
its 6-year normal 42.3
Company value for each $1 of cash profit
58.5
its 6-year normal 24.3
The shares have fallen 17.6% since the case was written and the September-quarter results are out: the call showed the margin without refunds near 53%, above the 51% management had given.
The next test is the December-quarter results, due in early February 2027: whether the margin stays above 51% for a second quarter without refunds, and whether foundry and logic sales keep growing as the new Texas factory starts.
The evidence under this - 4 claims
C40facthigh confidence
The shares returned 257% over the 252 sessions to 25 September 2026 and sit 19.1% below the high of that window.
- 2026-09-27editedRe-struck on the refreshed pack's 252-session window to 25 September 2026: the return fell to 257% as the window's start moved off the low, and the distance below the high narrowed to 19.1%.
C04management claimmedium confidence
Management says the recurring gross margin is now about 53% at Q3 volumes, up from about 51% to 51.5% a quarter earlier, and expects 2027 gross margin at similar levels as higher volume and efficiency offset the Farmers Branch ramp; the reported 56.5% in Q3 included about 380 basis points of tariff refunds and other favourable items that will not recur.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- Management also says Farmers Branch will be accretive to gross margin once ramped, so 51% is presented as a floor rather than a ceiling.
- The company has beaten its own gross margin guidance in each of the last four quarters, most recently by 32 basis points above the high end.
- The 51% figure appears only on the call; no filing states it, so it is a number that can be revised without amending anything.
- The 53% figure is stated on the call and in no filing, so it can be revised without amending anything.
- Management says a new factory ramp typically costs margin and that it expects to offset that with volume; the Q4 outlook of 54.0% includes about 70 basis points of refunds, so it is about 53.3% on the baseline.
- 2026-10-28editedQ3 2026 release shows a margin and a Q4 outlook above the 51% baseline management gave; the claim stays as management's statement and records the counter-evidence
- 2026-10-29editedthe Q3 call raised the baseline gross margin to about 53% from about 51% and put the Q3 one-offs at about 380 basis points
C09management claimhigh confidence
Management guides third-quarter DRAM revenue comparable to the second quarter but with a significant shift from HBM to DDR, because customers earn more starting DDR5 wafers than HBM4 wafers.
against this
- A mix shift inside a flat DRAM line is not a demand loss, and management frames it as customers optimising profit in a shortage rather than as weakness.
- DDR5 probe cards are still FormFactor's product; the company sold DDR into record DRAM quarters through the second half of 2025 before HBM accelerated.
C13management claimhigh confidence
GPU probe cards were primarily a competitor's business; FormFactor shipped probe cards for one GPU design to the leading foundry in multimillion-dollar quantities in Q3 2026 and expects significant share gains in 2027, constrained by its ability to produce in volume.
sourceFormFactor Q2 2026 earnings call transcriptprimaryTechnoprobe S.p.A corporate websiteprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- This is a damaging admission volunteered by the chief executive, which is the most credible kind of company statement; the incentive runs the other way.
- Qualification at a GPU maker is a multi-year gate the company has now passed, so the same fact reads as an option that is not yet in any reported number.
- 2026-10-29editedthe Q3 call confirmed the first production GPU shipments (one design, multimillion-dollar volumes) and that supply, not demand, limits custom-chip volume
How we read the multiple
Buyers pay 87.6 times last year's earnings, about twice its own normal. Counting its cash, the business costs about $9.77B, about 36 times the yearly adjusted operating profit this quarter's forecast points to.
The 2030 plan is $1.6B of sales and $5.00 of adjusted earnings a share. If it is met exactly:
- At 25 times earnings in 2030, owning the shares returns slightly less than nothing a year.
- At 30 times, about 3.5% a year.
- Buyers would need to pay about 39 times for a 10% yearly return.
Against its own history. Far dearer than its own normal, which is fair only if the higher margin lasts, and management's own figure says part of it will not.
The evidence under this - 4 claims
C21estimatemedium confidence
At $129.69, the 25 September 2026 close, the enterprise is about $9.77 billion: roughly 36 times the non-GAAP operating profit the third-quarter guide annualises to at the 51% baseline, and 26 times the 2030 target's $5.00.
derivedMarket value $10.11bn from the pack less net cash of $337m (cash and investments $349.0m against $11.6m of debt, 10-Q). Q3 revenue guide midpoint $270m annualised to $1.08bn, at 51% non-GAAP gross margin less the annualised $70m quarterly non-GAAP opex guide, gives $271m. $129.69 divided by the $5.00 target.
against this
- Annualising one guided quarter is a snapshot of a business growing double digits sequentially, so the forward multiple falls fast if the growth holds.
- The 2030 target is the company's own floor-shaped plan and the street is already above its implied path, so 26 times a stale target overstates what is being paid for.
- 2026-09-27editedRe-struck at the 25 September 2026 close of $129.69 after a 12.8% rise from $114.95: market value from the refreshed pack, net cash and the annualised guide unchanged.
C20management claimhigh confidence
The target model introduced in May 2026 is $1.6 billion of revenue, 55% non-GAAP gross margin, operating expenses at 23% of revenue and $5.00 of non-GAAP earnings per share, by 2030.
against this
- The prior target model of $850 million, 47% gross margin and $2.00 of earnings per share was reached on a run-rate basis in the first quarter of 2026, so this management's targets have been floors rather than ceilings.
- The vendor panel already has $3.75 for 2027, three quarters of the way to a 2030 goal set sixteen months earlier, which suggests the model is stale rather than stretching.
C22estimatemedium confidence
Meeting the 2030 target exactly and exiting at 25 times it returns about minus 1% a year from $129.69, the 25 September 2026 close; at 30 times about 3.5% a year, at 35 times about 7.3%, and about 39 times is needed for 10%.
derived$5.00 of 2030 non-GAAP earnings per share at 25, 30 and 35 times gives $125, $150 and $175, compounded over the 4.27 years from 25 September 2026 to the end of 2030; 10% a year over the same span needs about $195, or 39 times $5.00.
against this
- If the target is beaten as the prior one was, the arithmetic runs on a larger number: $5.70 in 2030 at 28 times is about 5% a year.
- The exercise assumes no capital returned; a resumed buyback after Farmers Branch would add to the per-share outcome.
- 2026-09-27editedRe-struck at the 25 September 2026 close of $129.69: at 25 times the plan met now returns slightly less than nothing a year, and about 39 times is needed for 10%.
C39factmedium confidence
Trailing valuation stands at more than twice its own six-year medians: price to sales of 11.2 times against 4.7, and price to earnings of 87.6 against 42.3.
sourceFormFactor research pack, frozen 2026-09-27secondaryFormFactor Form 10-Q, quarter ended 27 June 2026primaryFormFactor Q4 and fiscal 2025 results release, Exhibit 99.01primary
- 2026-09-27editedRe-read from the refreshed research pack at the 25 September 2026 close: both trailing multiples rose with the price against unchanged six-year medians, from roughly twice to more than twice.
Priced against its rivals
None of the listed rivals above sells the same thing to the same buyers as a large part of what it does. A price comparison needs at least two, so the price is judged against its own history instead.
Private or foreign rivals that cannot be priced: Technoprobe S.p.A, Micronics Japan Co., Ltd., Japan Electronic Materials Corporation, MPI Corporation, TSE Co., Ltd., Korea Instrument Co., Ltd., STAr Technologies, Inc., Max One, GGB Industries Inc.
Where we disagree with the market
The market believes
FormFactor has become a higher-margin artificial-intelligence test business for good, worth far more for each dollar of profit.
We believe
The margin step is real at 51%, not 53%, and the price already pays for a plan whose growth engines have not shipped.
The market is reading the results release, which shows 53.3% and forecasts 54%. The lower lasting figure, the timing of the graphics-processor approval and the 2027 date for custom chips are only on the call.
We would be wrong if graphics-processor and custom-chip cards bring in real revenue in 2026, a third big memory customer arrives, or the 2030 target is raised.
What analysts say, and the evidence
Analysts positive
50%
of 10, 1 Sep
Four months earlier
27%
of 11, 1 May
The evidence under this - 5 claims
C03facthigh confidence
Second-quarter 2026 gross margin was 50.7% on a GAAP basis and 53.3% on a non-GAAP basis, against 37.3% and 38.5% a year earlier.
C05management claimhigh confidence
The third-quarter guide of 54% non-GAAP gross margin included tariff refunds; the quarter's refunds were $8.6M, about 300 basis points, and about $2M more (around 70 basis points) is expected in Q4, after which management expects no ongoing benefit from refunds.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- The refunds are the return of tariffs actually paid from 2025 into 2026 and later ruled unlawful, so they are a recovery of real cost rather than a windfall.
- Management disclosed the amount unprompted in the prepared remarks rather than letting the 54% stand, which is the opposite of the behaviour this claim is used to worry about.
- 2026-10-29editedthe Q3 call gave the refund actually received ($8.6M, about 300 basis points) and the roughly $2M still due in Q4
C04management claimmedium confidence
Management says the recurring gross margin is now about 53% at Q3 volumes, up from about 51% to 51.5% a quarter earlier, and expects 2027 gross margin at similar levels as higher volume and efficiency offset the Farmers Branch ramp; the reported 56.5% in Q3 included about 380 basis points of tariff refunds and other favourable items that will not recur.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- Management also says Farmers Branch will be accretive to gross margin once ramped, so 51% is presented as a floor rather than a ceiling.
- The company has beaten its own gross margin guidance in each of the last four quarters, most recently by 32 basis points above the high end.
- The 51% figure appears only on the call; no filing states it, so it is a number that can be revised without amending anything.
- The 53% figure is stated on the call and in no filing, so it can be revised without amending anything.
- Management says a new factory ramp typically costs margin and that it expects to offset that with volume; the Q4 outlook of 54.0% includes about 70 basis points of refunds, so it is about 53.3% on the baseline.
- 2026-10-28editedQ3 2026 release shows a margin and a Q4 outlook above the 51% baseline management gave; the claim stays as management's statement and records the counter-evidence
- 2026-10-29editedthe Q3 call raised the baseline gross margin to about 53% from about 51% and put the Q3 one-offs at about 380 basis points
C13management claimhigh confidence
GPU probe cards were primarily a competitor's business; FormFactor shipped probe cards for one GPU design to the leading foundry in multimillion-dollar quantities in Q3 2026 and expects significant share gains in 2027, constrained by its ability to produce in volume.
sourceFormFactor Q2 2026 earnings call transcriptprimaryTechnoprobe S.p.A corporate websiteprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- This is a damaging admission volunteered by the chief executive, which is the most credible kind of company statement; the incentive runs the other way.
- Qualification at a GPU maker is a multi-year gate the company has now passed, so the same fact reads as an option that is not yet in any reported number.
- 2026-10-29editedthe Q3 call confirmed the first production GPU shipments (one design, multimillion-dollar volumes) and that supply, not demand, limits custom-chip volume
C14management claimhigh confidence
Management expects no significant custom-ASIC probe card revenue in 2026, with the advanced MEMS transition at those parts ramping into 2027.
against this
- Management says the same transition already happened at GPUs two to three generations ago and that it is engaged with all the hyperscalers, so the pattern has a precedent rather than being a hope.
- FormFactor booked some custom-ASIC revenue in the second quarter already; the claim is about the discontinuity, not about zero.
What could go wrong, and how we would know
The risks
- One customer, a quarter of sales. SK hynix is 24.3% of revenue, and its share of the memory FormFactor tests fell from 64% to 50% in a year while Samsung's rose.
- Memory makers change the mix. Customers are moving wafers to ordinary memory because it pays them more, and management expects that shift in the third quarter.
- A factory still to fill. Farmers Branch ramps through 2028, and $25M to $30M of start-up cost moves into the cost of each card once production begins.
- A grant that may need repaying. An $18.0M California grant depends on California jobs and investment, while the company closes two California plants.
The evidence under this - 6 claims
C10facthigh confidence
SK hynix was 24.3% of second-quarter 2026 revenue and 26.8% of the first half, and was a greater-than-10% customer in all eight quarters disclosed in the fiscal 2025 annual report.
C11factmedium confidence
SK hynix's share of HBM revenue fell from 64% in the second quarter of 2025 to 50% in the second quarter of 2026, while Samsung's rose to 33%.
C09management claimhigh confidence
Management guides third-quarter DRAM revenue comparable to the second quarter but with a significant shift from HBM to DDR, because customers earn more starting DDR5 wafers than HBM4 wafers.
against this
- A mix shift inside a flat DRAM line is not a demand loss, and management frames it as customers optimising profit in a shortage rather than as weakness.
- DDR5 probe cards are still FormFactor's product; the company sold DDR into record DRAM quarters through the second half of 2025 before HBM accelerated.
C16management claimhigh confidence
Farmers Branch starts ramping at the end of 2026 and runs through 2027 and 2028; management expects it to be accretive to gross margin only on completion of the ramp, by the beginning of 2028.
against this
- Management says it plans to largely offset the ramp inefficiency with the same operational improvements that delivered the last four quarters of margin gain.
- The initial targeted capacity is roughly the whole of the current California probe card footprint, so the option value of the site is large relative to the cost of carrying it for a year.
C17management claimhigh confidence
Farmers Branch pre-production ramp costs of $25 million to $30 million sit in general and administrative expense in 2026 and move into cost of goods sold once production begins.
against this
- The reclassification moves cost between lines rather than creating it, so a reader tracking non-GAAP operating income sees the same dollars; only the gross margin percentage is affected.
- Management has quantified the 2026 cost and has not quantified the effect on the 2027 gross margin, which is the number that matters and the one it chose not to give.
C37facthigh confidence
The $18.0 million deferred grant is a California Competes Grant whose repayment depends on California headcount and capital investment over a five-year term, while the company closes two California plants.
What we still do not know
- Whether the 51% margin holds once the new factory's costs move into the cost of each card.
- How much chipmakers spend on probe cards for each dollar of that fast memory. Nobody publishes it.
- How big the graphics-processor and custom-chip orders will be. No volume has been disclosed for either.
The evidence under this - 7 claims
C04management claimmedium confidence
Management says the recurring gross margin is now about 53% at Q3 volumes, up from about 51% to 51.5% a quarter earlier, and expects 2027 gross margin at similar levels as higher volume and efficiency offset the Farmers Branch ramp; the reported 56.5% in Q3 included about 380 basis points of tariff refunds and other favourable items that will not recur.
sourceFormFactor Q2 2026 earnings call transcriptprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- Management also says Farmers Branch will be accretive to gross margin once ramped, so 51% is presented as a floor rather than a ceiling.
- The company has beaten its own gross margin guidance in each of the last four quarters, most recently by 32 basis points above the high end.
- The 51% figure appears only on the call; no filing states it, so it is a number that can be revised without amending anything.
- The 53% figure is stated on the call and in no filing, so it can be revised without amending anything.
- Management says a new factory ramp typically costs margin and that it expects to offset that with volume; the Q4 outlook of 54.0% includes about 70 basis points of refunds, so it is about 53.3% on the baseline.
- 2026-10-28editedQ3 2026 release shows a margin and a Q4 outlook above the 51% baseline management gave; the claim stays as management's statement and records the counter-evidence
- 2026-10-29editedthe Q3 call raised the baseline gross margin to about 53% from about 51% and put the Q3 one-offs at about 380 basis points
C17management claimhigh confidence
Farmers Branch pre-production ramp costs of $25 million to $30 million sit in general and administrative expense in 2026 and move into cost of goods sold once production begins.
against this
- The reclassification moves cost between lines rather than creating it, so a reader tracking non-GAAP operating income sees the same dollars; only the gross margin percentage is affected.
- Management has quantified the 2026 cost and has not quantified the effect on the 2027 gross margin, which is the number that matters and the one it chose not to give.
C31estimatemedium confidence
HBM revenues are forecast to rise from $33 billion in 2025 to $86 billion in 2027, a compound rate of 60.5% - Gartner's figure, carried in SK hynix's SEC-filed prospectus.
against this
- Forecasts of this steepness are made at a cyclical extreme and the same prospectus forecasts the whole memory market growing 192.7% in a single year, which is not a rate anything sustains.
- FormFactor's own third-quarter guide already shows customers rotating wafer starts away from HBM4 to DDR5, which is the demand mix moving against the forecast's composition inside the forecast period.
C32facthigh confidence
First-half 2026 DRAM probe card revenue was $167.9 million and Foundry and Logic was $233.0 million.
C13management claimhigh confidence
GPU probe cards were primarily a competitor's business; FormFactor shipped probe cards for one GPU design to the leading foundry in multimillion-dollar quantities in Q3 2026 and expects significant share gains in 2027, constrained by its ability to produce in volume.
sourceFormFactor Q2 2026 earnings call transcriptprimaryTechnoprobe S.p.A corporate websiteprimaryQ3 2026 earnings call transcript (FMP), held October 28, 2026primary
against this
- This is a damaging admission volunteered by the chief executive, which is the most credible kind of company statement; the incentive runs the other way.
- Qualification at a GPU maker is a multi-year gate the company has now passed, so the same fact reads as an option that is not yet in any reported number.
- 2026-10-29editedthe Q3 call confirmed the first production GPU shipments (one design, multimillion-dollar volumes) and that supply, not demand, limits custom-chip volume
C15management claimmedium confidence
At the large fabless CPU customer FormFactor describes its market share as low single digit, with capacity constraints acting as a governor until Farmers Branch comes online.
against this
- Low single digit share at a customer FormFactor was not qualified at two years ago is progress, not weakness, and management describes multiple design wins already booked.
- Management attributes part of the slow share gain to its own capacity constraint rather than to losing on merit, which is a claim that flatters the company and cannot be checked from outside.
C14management claimhigh confidence
Management expects no significant custom-ASIC probe card revenue in 2026, with the advanced MEMS transition at those parts ramping into 2027.
against this
- Management says the same transition already happened at GPUs two to three generations ago and that it is engaged with all the hyperscalers, so the pattern has a precedent rather than being a hope.
- FormFactor booked some custom-ASIC revenue in the second quarter already; the claim is about the discontinuity, not about zero.
What would change our view
- Unattractive→AttractivePriced so that meeting the plan is enough
It needs:
- ○The company's value falls to about 16 times the adjusted operating profit this quarter's forecast points to, where simply meeting the 2030 plan pays.about ≈$57.48 today, 56% below the last close
- Reopens the researchThe ramps beat the target model
It needs any of:
- ○Sales of cards for logic chips and foundries grow through 2027 on the graphics-processor approval, faster than the 2030 plan assumes.checked at the next results, due by 4 Feb
- ○The adjusted gross margin stays above 51% for two quarters running, rather than sliding toward 48%.checked at the next results, due by 4 Feb
What is coming
- 4 November 2026 Q3 2026 results: Whether non-GAAP gross margin holds near 51% once the $7M to $9M of tariff refunds are stripped out, DRAM stays flat through the HBM-to-DDR shift, and GPU cards ship.
Every dated event ahead
- 4 November 2026 Q3 2026 results: Whether non-GAAP gross margin holds near 51% once the $7M to $9M of tariff refunds are stripped out, DRAM stays flat through the HBM-to-DDR shift, and GPU cards ship.
- December 2026 Farmers Branch production starts: Whether the Texas plant starts on time; its start moves $25M to $30M of yearly start-up cost into cost of sales, testing the margin baseline.
- February 2027 Q4 2026 results and the FY2026 10-K: The 2027 margin path with Farmers Branch in cost of sales, the full-year capital spending against $140M to $170M, and whether buybacks resume.
- 24 April 2027 Share repurchase authorisation expires: Whether the board renews the programme meant to offset stock-pay dilution, with $70.9M unused while the share count climbs.
Since we wrote this
Written
12 Sep
Price the thesis read
$129.69
25 Sep
- 29 OctEarningsCall split out refunds; baseline margin raised to 53%Without about 300 basis points of refunds the quarter's margin was 52.5% to 53%; Unattractive holds on price until the Q4 print is the second quarter above 51%.
- 28 OctEarningsMargin 56.5% beat the 54% line; view held for the callMargin and the Q4 outlook sit above the 51% baseline and the price fell 17.6%, but the refunds in Q3 are not split out; Unattractive holds until the call or 10-Q shows the baseline.
- 23 OctNewsTeradyne is buying $300M of new FormFactor stock, not only partnering. View heldTeradyne agrees to buy $300M of FormFactor stock in a private placement (8-K filed 13 October); Farmers Branch opening release.
- 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.
- 27 SepPrice moveShares up 12.8% with no news; plan met now loses slightlyNo results or filing arrived. At the higher price, meeting the company's own 2030 plan and exiting at 25 times earnings returns slightly less than nothing a year, so the case against the price got stronger.
- 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.
- 12 SepThesis writtenUnattractive, medium convictionWritten six weeks after results in which management put the lasting margin at about 51%, below the 53.3% reported, with the shares up 290% in a year.
Also checked 3 times with nothing new to act on.
Sources 17
Where this came from
Sources - 17
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.
- S01primarycompany filing2026-02-20Item 1 business and product description, the named competitor list, the eight-quarter customer concentration table, China revenue share, manufacturing footprint and the Farmers Branch build-outbacksC01C10C12C26C27C37C47C52C53
- S02primarycompany filing2026-08-04the income statement, the balance sheet and cash flow, segment and end-market revenue, segment gross margins, customer concentration, the restructuring note, the FICT equity investment note and the California Competes deferred grantbacksC02C03C06C07C08C10C18C23C25C28C32C33C37C38C39C43C46C47C53
- S03primarycompany filing2026-07-29the GAAP to non-GAAP reconciliation, the non-GAAP gross margin series and the Q3 2026 outlookbacksC03C44
- S04primarytranscript2026-07-29FormFactor Q2 2026 earnings call transcriptmanagement's 51% sustainable gross margin, the composition of the Q3 tariff refunds, the 2030 target model, GPU and custom-ASIC timing, the DRAM mix shift, Farmers Branch ramp costs; and eight analyst questions, of which the first was on the margin baseline
So the recurring or sustainable element of gross margins, we believe, is more like 51% at the current volumes and mix.
backsC04C05C08C09C13C14C15C16C17C19C20C23C25C27C33C34C43C44 - S05primarycompany filing2026-02-04the fiscal 2025 full-year figures, the GAAP and non-GAAP earnings per share series, full-year free cash flow and the Q1 2026 outlookbacksC35C39
- S06primarycompany filing2026-04-29the Q1 2026 non-GAAP gross margin and earnings per share, and management's statement that the prior target model was reached on a quarterly run-rate basisbacksC45
- S07secondaryvendor data2026-09-27FormFactor research pack, frozen 2026-09-27price, market value, the 252-day return and range, volatility, the analyst panel split, the vendor multiple series against its own six-year medians, short interest and the insider filing aggregatebacksC06C39C40C41C42
- S08primarycompany filing2026-09-02SK hynix Form 424B4 prospectus SK hynixIDC's HBM and DRAM share figures, Gartner's HBM revenue forecast, and the absence of any probe card manufacturing in the customer's own description of itselfbacksC11C30C31
- S09primaryindustry data2026-09-12Counterpoint Research - Global DRAM and HBM Market Share, quarterly Counterpoint Researchthe quarterly HBM revenue share series showing SK hynix falling from 64% in Q2 2025 to 50% in Q2 2026
Q2 2026: SK Hynix 50%, Samsung 33%, Micron 18%
backsC11 - S10primarycompany release2026-09-12Technoprobe S.p.A corporate website Technoprobethe attacker's own description of its product and scope, from outside FormFactor
We specialise in the design, development and production of probe cards for the biggest technology brands
backsC13C27 - S11primarycompany release2026-09-12the built-in self-test substitute, described by the company selling it
It tests and permanently repairs all defective memories in a chip using virtually no external resources.
backsC29 - S12primarycompany filing2026-08-17the transaction codes and the footnote confirming an automatic Rule 10b5-1 plan adopted 19 August 2025backsC42
- S13tertiarynews2026-08-05the rival's half-year growth rate; its own press release and results presentation PDFs are image-only and could not be read, so this stays tertiary and carries no score
Technoprobe H1 FY26 net profit more than doubled to EUR 143.1 million; revenue rises 42.4% to EUR 464.1 million
backsC28 - S14primarycompany filing2026-04-02beneficial ownership as of 16 March 2026, director and officer biographies, the CD&A: the 2025 short-term incentive design and payout, the PRSU and RSU design and the 2022 PRSU resultbacksC49C50C51
- S15primarycompany filing2023-02-24the FY2020 to FY2022 income statement lines and capital spending, read through the frozen numbers card (content/FORM/financials.json)backsC47
- S16primarycompany filing2026-10-28the Q3 2026 income statement, GAAP and non-GAAP gross margin, segment growth commentary, cash flow and the Q4 2026 outlookbacksC54
- S17primarytranscript2026-10-28the tariff refunds inside the 56.5% margin, the raised baseline margin of about 53%, the roughly $2 million of refunds still due in Q4, the 2027 margin view, Farmers Branch capacity, the first GPU probe cards and the questions
The favorable impact of tariff refunds on gross margins in the third quarter was about 300 basis points, or $8.6 million.
backsC04C05C13C54