FORM · FormFactor, Inc. · Stock research

Current Thesis

The fundamental leg keeps printing and the price keeps refusing to hold it. Since the last note, FormFactor rallied out of the early-August shelf into the $130s — StockStory put shares at $129.57 on 2026-08-12, CEO Michael Slessor filed open-market sales at $128.72–$130.54 on 2026-08-14, and GuruFocus reported a 5.0% session gain to $138.16 on 2026-08-17 — then gave the entire move back. On 2026-08-19 the stock fell 6% to $117.27 with no company news, alongside Teradyne -5% to $383.80, Cohu -6% to $55.73 and Aehr -10% to $111.01 (24/7 Wall St, 2026-08-19). The adjusted close on 2026-08-21 was $114.16, 28.6% below the 52-week high of $159.93, and fractionally under the 2026-08-05 close of $114.44. Two weeks of record-quarter follow-through produced a net lower close. The narrative an investor is buying — probe-card content growth as HBM4 layer and pin counts raise test intensity and probe-card replacement frequency — is intact in the reported numbers and is not currently setting the price. NVIDIA's 2026-08-26 print is the only dated input inside the next 30 days that can reset the complex.

Bullish and bearish views on FormFactor, Inc.

The model's bull view on FormFactor, Inc. (FORM), in brief: Q2 2026 revenue was $258.242M against $239.999M consensus and adjusted EPS $0.82 against $0.61 (2026-07-29 release) — a company record, up 14% QoQ and 32% YoY, with management attributing demand to DRAM, Foundry & Logic and systems tied to HBM and co-packaged optics. The bear view: The mid-August advance retraced completely. Both cases follow in full.

Bull Case

• Q2 2026 revenue was $258.242M against $239.999M consensus and adjusted EPS $0.82 against $0.61 (2026-07-29 release) — a company record, up 14% QoQ and 32% YoY, with management attributing demand to DRAM, Foundry & Logic and systems tied to HBM and co-packaged optics.
• The Q3 guide raises rather than digests: $260–280M revenue against $247.354M consensus, adjusted EPS $0.77–$0.95 against $0.63, GAAP EPS $0.66–$0.84 against $0.51 (2026-07-29). StockStory framed the midpoint on 2026-08-12 as roughly $270M revenue and about $0.86 adjusted EPS, pushing the annualised revenue run rate past $1B for the first time.
• The drawdown is off a doubled base. 24/7 Wall St put FormFactor at +123% year-to-date through the 2026-08-18 close before the 2026-08-19 selloff; StockStory had it +119% year-to-date at $129.57 on 2026-08-12. This is a correction inside an up-year, not a broken advance.
• 2026-07-28: expanded strategic partnership with Keystone Microtech covering probe-card assembly, test and distribution in Taiwan — capacity adjacent to the foundry and OSAT base, disclosed one day before the print.
• Evercore ISI held Outperform with a $150 target on 2026-07-30, against a 2026-08-21 close of $114.16. Aggregated sell-side targets sit above market: WallStreetZen shows a $146.67 average across six analysts with a $175.00 high; a second aggregator shows $147.56 across sixteen.
• The 2026-05-11 Investor Day 2030 model ($1.6B revenue, 55% gross margin, 23% opex, $5.00 non-GAAP EPS) now has two quarters of revenue slope behind it rather than a slide.

Bear Case

• The mid-August advance retraced completely. From reported levels in the $130s on 2026-08-14 and 2026-08-17, the stock closed 2026-08-21 at $114.16 — below the 2026-08-05 close of $114.44. A record quarter and a raised guide have so far failed to build a higher low.
• On 2026-08-19 the whole AI-test basket moved as one. 24/7 Wall St described the day as sector-wide profit-taking after outsized year-to-date gains (Aehr +510%, Cohu +139%, Teradyne +109%, XSD +61%) and characterised it as positioning rather than a rates problem; a companion piece the same day noted Applied Materials and Lam Research falling 4–5% despite falling Treasury yields. FormFactor's own bookings did not set its price that session.
• Insiders sold into the strength. Slessor's Form 4 records three lots on 2026-08-14 — three sessions before the reported local peak. Dennis 1,000 at $115.8399 (2026-05-19), Rebeca Obregon-Jimenez 3,828 units at $126.475 (2026-05-21).
• Targets were cut into the beat. TD Cowen kept a Hold and took its number to $100 on 2026-07-30 — still below the 2026-08-21 close of $114.16 — while Evercore trimmed to $150 from a higher level on the same morning. Marking the multiple down faster than the estimates go up is what a de-rating looks like in real time.
• Margin measure remains unreconciled in the coverage. The Motley Fool's 2026-07-30 write-up cited Q2 gross margin of 43.97% against a non-GAAP guide of 49.5% ±1.5 points given 2026-04-29. Those are different measures and the GAAP/non-GAAP spread for this name is normally wide, but no reconciled non-GAAP figure surfaced in headline coverage.
• Customer concentration is unchanged: probe cards into the top three HBM makers, so SK hynix, Micron and Samsung capex timing is the revenue vector, and one qualification push-right becomes a company-level gap.

Setup & Price Structure

• Reference close 2026-08-21: $114.16, 28.6% below the 52-week high of $159.93, with a three-month price change of -11.5% and RSI(14) at 53.5. Two weeks earlier (2026-08-07) the adjusted close was $117.39 with RSI(14) at 57.2 and a three-month change of -20.6%.
• RSI at 53.5 after a roughly 17% slide off the reported mid-August highs is neutral. There is no oversold washout here and no overbought extension — momentum has been discharged without a capitulation print.
• The shelf is defined by two closes eleven sessions apart: $114.44 on 2026-08-05 and $114.16 on 2026-08-21. That band is the structure the post-Q2 recovery is standing on. Below it, the nearest reference the prior coverage flagged is the $110 area.
• Overhead references, in order: the 2026-08-19 level of $117.27, the $128.72–$130.54 zone where the CEO sold on 2026-08-14, the reported $138.16 print of 2026-08-17, and the 52-week high of $159.93 set in June 2026.
• The narrative is maturing. Dating it: the story is still delivering upside surprises (2026-07-29 beat and above-consensus guide), so it has not failed; but the flow has moderated and rotated. The 2026-08-19 session moved every AI-test name together on no company news, retail-facing coverage clustered heavily through August (Motley Fool 2026-07-30, Alphastreet 2026-08-11, StockStory 2026-08-12, GuruFocus 2026-08-17, 24/7 Wall St 2026-08-19), and the Russell 1000 mechanical bid from 2026-06-29 is spent. It is well known, still working, with a bid that is being tested rather than expanding.
• Crowding and positioning observables, stated plainly: +123% year-to-date through 2026-08-18; five separate retail-facing outlets publishing on the move within three weeks; 8,656 shares sold by the CEO on 2026-08-14 near the local high; a rating spread running from a $100 Hold to a $150 Outperform, both set on 2026-07-30; and a sector-defining print (NVIDIA) landing four sessions after the reference close.

Catalyst Calendar (next 30 days)

• 2026-08-26 — NVIDIA Q2 FY2027 results (quarter ended 2026-07-26). The single dated read on AI-accelerator and HBM demand pacing inside the window. Nothing company-specific falls inside 30 days, so the complex trades off this.
• ~2026-09-24 (est.) — Micron fiscal Q4 2026. Nearest direct read from a named customer on HBM capex, HBM4 qualification timing and DRAM pricing. Sits just outside the 30-day window.
• ~2026-10-28 (est.) — FormFactor Q3 2026 print. Grades the $260–280M revenue and $0.77–$0.95 adjusted EPS guide issued 2026-07-29 and gives the first reconciled margin read on the capacity ramp.

What Would Change Our Mind

The August shelf is the object under test. Two closes — $114.44 on 2026-08-05 and $114.16 on 2026-08-21 — have now defined a floor that the whole post-Q2 advance is resting on, and the mid-August excursion into the $130s failed to convert it into a higher low. A weekly close below $110 forfeits that shelf and reframes the July–August recovery as a failed rally inside a topping structure rather than a base.

Three non-price conditions would carry the same information. First, NVIDIA's 2026-08-26 print landing with an AI-capex digestion message from the largest downstream buyer of HBM — that repriced the complex once already on 2026-07-02 and again on 2026-08-19, both times without FormFactor news. Second, DRAM revenue declining sequentially at the Q3 print, which would say the HBM4 test-intensity story stopped compounding at exactly the point the 2030 model needs it to accelerate. Third, another round of target reductions after an in-line or better Q3 — a repeat of 2026-07-30, when both a bull and a bear cut into a record beat, would confirm the multiple is being marked down independently of the estimates.

On the other side, reclaiming and holding the $128–$130 zone where insiders sold on 2026-08-14 would say the distribution has been absorbed and the June high is back in play.

Correlation Notes

• FormFactor trades inside an AI test-and-probe basket, and on 2026-08-19 that basket moved as one unit: FORM -6%, Teradyne -5%, Cohu -6%, Aehr -10%, with no FormFactor-specific news. Idiosyncratic fundamental news is being overridden by complex-level flow on those days.
• The same pattern ran on 2026-07-02, when a chip selloff cut FormFactor roughly 24% from its late-June high within days.
• The upstream driver is memory capex, not general semis. Micron and SK hynix shares fell in the same August AI-chip selloff, and those two plus Samsung are the direct customer set for probe cards into HBM.
• NVIDIA functions as the demand proxy: its guide sets the market's assumed HBM bit-growth path, which sets probe-card replacement demand two steps downstream.
• Russell 1000 inclusion on 2026-06-29 attached a passive holder base, which adds index beta to what was previously a more idiosyncratic small/mid-cap tape.