FormFactor, Inc. (FORM) | The Buildout — AI Infrastructure
The Verdict
FormFactor makes the physical interface that touches a chip while it is being tested. Its probe cards carry microscopic contacts that land on a wafer's pads and connect the chip to test equipment, and its Systems segment adds probe stations, thermal chucks and cryogenic instruments. That matters for AI because the most expensive parts in the industry — HBM stacks headed for GPUs and custom accelerators — are worth rejecting before assembly, and because co-packaged optics moves data with light instead of copper. The company sells to semiconductor makers and scientific institutions, and its probe cards are specific to each customer's chip design, so its product mix moves whenever customers change what they build. Both of its reportable segments, Probe Cards and Systems, sit on the same rising test intensity and complexity at the intersection of high-performance computing and advanced packaging.
| Market Cap | — |
| Revenue (TTM) | $902M |
| Revenue Growth | +18.0% |
| EBITDA Margin (TTM) | 21.4% |
| Net Cash | $316M |
| Earnings Beats | 5 of 7 |
| P/E (TTM) | — |
| EV/EBITDA (TTM) | — |
What We Like
- Revenue reached a record $258.2M in Q2 FY2026, up $32.1M — roughly 14% — from Q1, the third consecutive record, and management says the company surpassed a $1B annualized revenue run rate.
- Non-GAAP gross margin reached 53.3% in Q2 FY2026, up from 49.0% in Q1, with management putting the durable baseline at about 51% — above the old $850M/47% target model, which it says it achieved on a run-rate basis in Q1 FY2026.
- Management reset its long-range model at the May 2026 Investor Day: $1.6B revenue, 55% non-GAAP gross margin and $5 non-GAAP EPS by 2030.
- The customer base widened. SK hynix was 29.5% of Q1 FY2026 revenue, Intel was 12.0%, and NVIDIA appeared at 10.2%; management also said first-quarter networking growth made a high-performance-compute leader a 10% customer for the first time.
- GPU probe cards are qualified at the world's largest foundry with production units shipping in Q3, and 2026 CPO revenue guidance was raised twice.
What We’re Watching
- Q3 FY2026 gross-margin guidance of 54.0% ±150bps includes $7M–$9M, about 300bps, of IEPA tariff refunds management does not expect to recur.
- DRAM revenue is guided comparable to the record Q2, but with a significant mix shift from HBM to DDR as customers chase DDR pricing.
- Capacity gates share. Management says Farmers Branch is a governor on market-share growth until it ramps, 2026 preproduction costs were raised to $25M–$30M, and gross-margin accretion from the site is primarily a 2028 event.
- Concentration is rising in receivables: two customers were 26.1% and 13.3% of gross accounts receivable at March 28, 2026, versus 15.8% and 10.8% at December 27, 2025.
The operating record looks intact and strengthening: three consecutive revenue records, a step up in the durable margin baseline, a raised guide, and a broader set of demand vectors across HBM, data-center CPU, GPU and co-packaged optics. Two things keep it from being clean. The margin headline outruns the margin run rate, and the next quarter's guide is helped by a one-time tariff refund. And most of the forward story — GPU volume, custom ASIC revenue, CPO scale and Farmers Branch capacity — has not landed in the numbers yet. The open question is whether a recurring margin baseline near 51% and the guided revenue growth can carry through a quarter where the DRAM mix turns less favorable.
Earnings Beat
FormFactor's Q2 FY2026 revenue was a record $258.2M, $18.2M above the midpoint of its $235M–$245M guide and up from $226.1M in Q1. GAAP gross margin was 50.7%; non-GAAP gross margin was 53.3%, against a management-stated recurring baseline of about 51%. Management said the quarter crossed two milestones on the path to its new target model: a $1B annualized revenue run rate and a 50% gross margin. Non-GAAP EPS was $0.82, up from $0.56 in Q1 and more than triple the year-ago quarter.
| Metric | Q2 FY2026 | Q1 FY2026 | Q2 FY2025 | YoY |
|---|---|---|---|---|
| Revenue | $258M | $226M | $196M | +31.9% |
| Gross margin | 50.7% | 47.9% | 37.2% | +1350bps |
| EBITDA | $65M | $56M | $24M | +175.1% |
| EPS | $0.71 | $0.26 | $0.12 | +502.0% |
| Systems segment revenue | $48.5M | $27.9M | n/a | — |
FormFactor's second quarter revenue, gross profit and earnings per share set all time records. And we achieved 2 important milestones on the path to the new target model we introduced in May. First, we surpassed $1 billion annual revenue run rate. And second, we exceeded a 50% gross margin.— Mike Slessor, Chief Executive Officer, 2026-07-29
Management tone: Management led with records and milestones, then pre-empted the quality-of-beat question: the CFO volunteered that the 53.3% gross margin included non-recurring items and put the recurring element near 51%. Versus the prior call, they raised the CPO outlook a second time, moved GPU from nearing qualification to shipping production units, and raised the Farmers Branch preproduction cost estimate against themselves. They declined to give quarter-by-quarter CPO detail for 2027 and did not share cycle-time or yield metrics. On competitive position they were specific about the limits, describing low-single-digit share at a fabless XPU customer and stronger share at two of three HBM customers.
Management Guidance
For Q3 FY2026 management guided revenue of $270M ±$10M, non-GAAP gross margin of 54.0% ±150bps, non-GAAP operating expenses of $70M ±$2M including about $7M of Farmers Branch preproduction, non-GAAP EPS of $0.86 ±$0.09, and a non-GAAP effective tax rate of 15–19%. The company said the margin increase over the roughly 51% baseline is driven primarily by $7M–$9M, about 300bps, of IEPA tariff refunds not expected to recur, and that the benefit of marginally higher volumes is expected to be offset by less favorable DRAM mix. Longer-range, management restated a target model of $1.6B revenue, 55% non-GAAP gross margin and $5 non-GAAP EPS by 2030, and held 2026 cash CapEx at $140M–$170M.
Trajectory
Revenue has grown for four straight quarters on the audited data, from $202.7M in Q3 FY2025 to $215.2M, $226.1M and then $258.2M in Q2 FY2026; the latest sequential increase was 14.2%, well above the 3.5% to 6.2% range of the prior three quarters. Gross margin over the same span moved from 39.7% to 50.7%. Management attributes the improvement to higher factory utilization, manufacturing spending discipline, improved yields and cost reductions from the Q1 restructuring — explicitly not pricing — and says about a third of the latest sequential margin gain came from items it does not expect to recur. The near-term governor is mix rather than demand: Q3 DRAM revenue is guided comparable to the record Q2 while shifting significantly from HBM to DDR, a customer decision management ties to DDR pricing.
The Model
The model projects FY+1 revenue of $1,044M with EBITDA of $268M, a 25.7% margin, rising to FY+2 revenue of $1,300M with EBITDA of $367M, a 28.2% margin. The near term rests on what is already shipping or guided — HBM probe cards, data-center CPU demand, CPO revenue guided above $20M for 2026, and a Systems segment recovering off an unusually weak first quarter — plus the Farmers Branch site management says begins production in Q4 2026. The FY+2 step-up depends on the items management dates later: GPU probe-card volume, the custom ASIC transition it calls a 2027 event, and CPO scaling toward the served market it estimates at about $400M by 2030. Capacity, not demand, is the constraint management names for all of them.
| Metric | FY2025 | Next FY (E) | Following FY (E) |
|---|---|---|---|
| Revenue | $785M | $1.0B | $1.3B |
| YoY Growth | — | +33.0% | +24.5% |
| EBITDA | $112M | $268M | $367M |
| EBITDA Margin | 14.3% | 25.7% | 28.2% |
Projections are the median of 5 independent model runs. The model’s revenue sits 17.1% above analyst consensus.
For Q3 FY2026 management guided revenue of $270M ±$10M, non-GAAP gross margin of 54.0% ±150bps, non-GAAP operating expenses of $70M ±$2M including about $7M of Farmers Branch preproduction, non-GAAP EPS of $0.86 ±$0.09, and a non-GAAP effective tax rate of 15–19%. The company said the margin increase over the roughly 51% baseline is driven primarily by $7M–$9M, about 300bps, of IEPA tariff refunds not expected to recur, and that the benefit of marginally higher volumes is expected to be offset by less favorable DRAM mix. Longer-range, management restated a target model of $1.6B revenue, 55% non-GAAP gross margin and $5 non-GAAP EPS by 2030, and held 2026 cash CapEx at $140M–$170M.
What Could Go Right — and Wrong
- GPU probe cards ramp past initial production units at the world's largest foundry through the second half of 2026 and into 2027.
- Hyperscaler custom ASIC programs convert from design wins to revenue in 2027 as next-generation parts require advanced MEMS probes.
- CPO revenue clears the guided $20M in 2026 and keeps scaling toward the roughly $400M served market management estimates by 2030.
- Farmers Branch comes online in Q4 2026 and ramps to initial target capacity by early 2028, removing the capacity governor on share gains and adding gross-margin accretion.
- The durable gross-margin baseline holds near 51% and climbs toward the 55% target model as volume, utilization and yield improve.
- The HBM-to-DDR mix shift lasts beyond one quarter, weakening the differentiated mix that produced record margins.
- Custom ASIC contributions slip past 2027, or hyperscaler programs stay in development.
- Capacity binds: sole-source substrates, complex circuit boards and interconnects, a purchase-order supply model with no long-term contracts, and industry-wide capacity constraints.
- Customer concentration rises further — SK hynix reached 29.5% of Q1 FY2026 revenue and receivables concentration jumped.
- Margin quality erodes: a meaningful share of recent gains is non-recurring, and Farmers Branch ramp inefficiencies run through 2027 before accretion arrives.
Looking Ahead
The next four quarters turn on converting qualified positions into revenue. GPU probe cards are qualified and shipping at the world's largest foundry, CPO revenue is guided past $20M by end-Q3 and significantly higher for the full year, and hyperscaler custom ASIC work is a 2027 event on management's own words. Against that, Farmers Branch is scheduled to begin production in Q4 2026 with a ramp through 2027 and 2028, and management says the added capacity is what removes the current governor on share gains. Management's own framing is that this is early innings.
- Q3 FY2026Q3 results land — Tests the $270M ±$10M revenue guide and the 54.0% ±150bps margin guide.
- Q3 FY2026GPU probe-card shipments — Production units ship for revenue after no GPU revenue in Q2.
- Q3 FY2026CPO crosses $20M — Management expects the 2026 bar cleared by the end of the quarter.
- Q4 2026Farmers Branch starts up — New Texas probe-card site begins production, then ramps through 2027–2028.
- 2027Custom ASIC revenue — Management says hyperscaler ASIC contributions are a 2027 event, not 2026.
- By 2030Target model — Management targets $1.6B revenue and 55% non-GAAP gross margin by 2030.
Financials
Annual Summary
| Metric | FY2024 | FY2025 | TTM | YoY |
|---|---|---|---|---|
| Revenue | $764M | $785M | $902M | +2.8% |
| Gross Margin | 40.2% | 39.5% | 45.6% | 62bps |
| EBITDA | $98M | $112M | $193M | +14.8% |
| EBITDA Margin | 12.8% | 14.3% | 21.4% | +148bps |
| Net Income | $70M | $54M | $116M | -21.8% |
| Free Cash Flow | $79M | $12M | $136M | — |
| Net Cash | — | — | — | — |
Key Ratios (Trailing)
- P/E TTM—
- EV/EBITDA TTM—
- EV/Revenue TTM—
- Price/FCF TTM—
- Gross Margin (TTM)45.6%
- EBITDA Margin (TTM)21.4%
- Net Margin (TTM)12.8%
- ROIC15.1%
- FCF Conversion70.5%
- SBC / Revenue4.0%
The Company
FormFactor supplies electrical and optical test and measurement technologies along the semiconductor product lifecycle, per its 10-K: probe cards, analytical probes, probe stations, thermal systems and cryogenic systems, sold to semiconductor companies and scientific institutions. A probe card is a contactor — microscopic contacts that land on a chip's pads and carry signals to and from test equipment — and it is specific to each customer's chip design, so the product mix moves whenever customers change what they build. In the AI buildout that makes FormFactor a picks-and-shovels supplier: its SmartMatrix full-wafer contactor tests hundreds of completed HBM stacks at once at the 10 gigabit-plus IO rate of HBM4, and its Systems products test the photonic wafers used for co-packaged optics.
The company reports two segments. Probe Cards covers probe cards and analytical probes, made at Livermore, California; Beaverton, Oregon; Yokohama, Japan; and a Farmers Branch, Texas site that is in build-out. Systems covers probe stations, thermal subsystems and cryogenic systems, made at Boulder, Colorado and at Thiendorf, Munich and Karlsruhe in Germany. More than 50 analytical probe models serve engineering and production testing, and probe cards use a range of architectures including micro-electromechanical systems, or MEMS, technologies.
Business Segments
Competitive Landscape
FormFactor's 10-K lists competitors by product line rather than naming a single rival: probe cards against Japan Electronic Materials, Korea Instrument, Micronics Japan, MPI Corporation, STAr Technologies, Max One, Technoprobe and TSE; analytical probes against GGB Industries and MPI; probe stations against MPI, Semishare, STAr, Tokyo Electron and Wentworth; thermal subsystems against AEM Singapore, ERS Electronic and Temptronic; and cryogenic systems against Bluefors, Lake Shore Cryotronics, Maybell Quantum, Montana Instruments and Quantum Design. On the calls management places its differentiation in high-speed HBM stack final test, where it says it has strong share at two major DRAM manufacturers, with lots of competition in core die insertions. It also says GPU MEMS probe cards are primarily a competitor's business today, a position it has just begun to contest after qualifying at the world's largest foundry.
- Technoprobe S.p.ANamed in the 10-K as a probe card competitor. The criticality assessment names Technoprobe and Micronics as the alternatives that would take 6–12 months to fully qualify if FormFactor could not deliver.
- Micronics JapanNamed in the 10-K as a probe card competitor, and one of the two alternatives named in the criticality assessment.
- MPI CorporationNamed in the 10-K as a competitor in probe cards, analytical probes and probe stations.
- Tokyo ElectronNamed in the 10-K as a probe station competitor; also a co-development partner on the Triton CPO production test system.
- AdvantestCo-development partner on the Triton CPO test system; the source's supply-chain graph also tags it as a competitor in CPO test systems.
Supply Chain
FormFactor sits one step upstream of the chipmakers, supplying the probe cards and test systems used in wafer and final test. Its customers are memory, foundry and logic companies. No neighbor in the source material named FormFactor directly.
More on FORM: Earnings recap