AMAT Gross Margin Stalls at 50.4% and the Stock Drops 5%

The core results:

Line up four quarters and the shape of the year comes into focus. FY2026 Q1 posted revenue of $7.01 billion and adjusted EPS of $2.38, with Semiconductor Systems still down 8% year over year; Q2 was $7.91 billion and $2.86; Q3 was $9.115 billion and $3.50; and for Q4 the company guides to $10.25 billion plus or minus $500 million with EPS of $4.02 plus or minus $0.20 (respective 8-K filings). Revenue traveled from $7.0 billion to $10.25 billion in three quarters, and over the same stretch gross margin climbed from 49.1% to 50.4% — the company's own framing is that this is the 13th consecutive quarter of year-over-year gross margin expansion, roughly 300 basis points cumulatively over three years, achieved through value-based pricing on a tool-by-tool basis (Q3 FY2026 earnings call). Going from 49.1% to 50.4% sounds like barely a point, but applied to $10.25 billion of revenue it is more than $130 million of incremental gross profit, close to a third of an entire quarter's buyback.

So what does next quarter have to deliver for that guidance to hold together? There is some hard arithmetic here. The company guides Q4 gross margin to 50.4%, flat with Q3 (Q3 FY2026 earnings materials), while the midpoint revenue of $10.25 billion implies a 12.5% sequential gain, taking gross profit to roughly $5.17 billion. But EPS going from $3.50 to $4.02 is a 14.9% sequential gain — faster than revenue. Working backward from roughly 800 million diluted shares in Q3 (GAAP net income of $2.538 billion divided by EPS of $3.17), $4.02 implies about $3.2 billion of adjusted net income, which requires the operating margin to rise from 34.0% to roughly 34.8%. With gross margin frozen, those 0.8 points have to come out of the expense line: Q3 non-GAAP operating expenses were roughly $1.498 billion, or 16.4% of revenue, and Q4 needs that down to about 15.6% — meaning expenses may grow only about 7% sequentially while revenue grows 12.5%.

And this quarter the company just added more than 1,500 net new employees in manufacturing and customer support (Q3 FY2026 earnings call). The one number to watch in the next quarterly report is operating expenses as a percentage of revenue — it decides whether this expansion is trading short-term cost for long-term capacity, or whether the costs are running ahead of capacity that hasn't landed yet.

The demand-side evidence is solid on its own terms. Management raised its target for calendar 2026 Semiconductor Systems revenue growth from "above 20%" to "above 30%," expects advanced packaging revenue growth "above 70%," expects the metrology and inspection business to grow "above 50%," and reported DRAM-related growth of 52% year over year (Q3 FY2026 earnings call). More telling is the change in framing: the mature-node ICAPS business, previously described as flat to slightly up, is now guided to grow in both 2026 and 2027, and the same holds for mature nodes in China (Q3 FY2026 earnings call). CEO Gary Dickerson's language on 2027 was that "the AI-driven demand function will continue, and we expect another strong year," while pinning the ceiling on growth to a very physical constraint — available cleanroom space.

This company does not burn cash, so this chapter is about the quality of the profit rather than its survival.

Start with the gap between the two reporting conventions. Q3 GAAP EPS of $3.17 was below adjusted EPS of $3.50, a difference of $0.33; at roughly 800 million diluted shares that is about $260 million of after-tax adjustments for the quarter, under 3% of revenue, mostly routine items like stock-based compensation and amortization of acquisition intangibles. Move back one quarter and the sign flips: Q2 GAAP EPS was $3.51, actually higher than adjusted EPS of $2.86, a difference of $0.65 (8-K, 2026-05-14). Same company, and in two consecutive quarters the GAAP-versus-non-GAAP gap swung in opposite directions — anyone tracking year-over-year GAAP EPS alone would have been flattered by a non-operating gain in Q2 and spooked by amortization in Q3. The operating trend here can only be read through the adjusted numbers, and the credibility of those adjusted numbers has to be corroborated by cash flow.

The cash flow corroborates:

Capital returns contain a contrast worth pulling out. The quarter's $440 million of buybacks and $420 million of dividends total $860 million, only 37% of free cash flow, against the company's own long-term target of returning 80% to 100% of free cash flow to shareholders (Q3 FY2026 earnings call), with $12.8 billion of buyback authorization remaining. Cash building on the balance sheet, debt coming down, and return ratios below the company's own stated commitment all point to the same explanation: the money is being held back for capacity expansion and for the plan to double quarterly systems output before 2028 (Q3 FY2026 earnings call).

One more piece of arithmetic on what this means for an actual shareholder: $440 million of buybacks, at roughly the share price range that prevailed during the quarter, retires fewer than one million shares — on a base of 800 million, that is about 0.1%. At this price level, the buyback barely covers the dilution from stock compensation; counting on it to support the share price is a bookkeeping error. The saving grace is that the dilution runs at about the same magnitude, so existing holders are not actually being eroded — a very different situation from growth companies that keep the lights on by issuing stock.

The customer names are spelled out in the company's own filings. In a May 2026 8-K, AMAT disclosed in one stroke an innovation partnership with TSMC, an extension of its existing collaboration with Samsung Electronics, a long-term collaboration agreement with SK Hynix, and joint development of next-generation products with Micron (8-K, 2026-05-14). Those four plus Intel account for the overwhelming majority of global leading-edge logic and memory capital spending, so customer concentration is inherently extreme — though the company does not disclose single-customer revenue share under the SEC's customer-concentration convention, and no specific percentage appears in public materials. The geographic concentration that is disclosed is China at 26% of combined Semiconductor Systems and services revenue (Q3 FY2026 earnings call), and management expects China revenue to grow in calendar 2026, driven mainly by 28-nanometer mature-node foundry investment. That 26% is the single largest variable sitting outside this company's balance sheet.

Upstream, AMAT itself shows up on other companies' lists. The same May 8-K named test-equipment maker Advantest as a collaborator (8-K, 2026-05-14). The more interesting corroboration comes from its components and subsystems suppliers: Ultra Clean (UCTT) and Ichor (ICHR) both issued first quantified guidance on 2026-08-03, citing the same drivers of surging demand and shipment volumes; ACM Research raised guidance on 2026-08-07; and Onto Innovation issued first guidance on 2026-08-06 — when a leading equipment maker's upstream supply chain collectively raises guidance in the same week, it says the conversion of orders into shipments is genuinely happening rather than sitting on customer intent lists.

The most fragile link is one management identified itself: cleanrooms. Dickerson pinned the deciding variable for 2027 growth on "cleanroom availability" (Q3 FY2026 earnings call), and that has two meanings — the pace of customer cleanroom construction limits how fast tools can be installed, and AMAT's own cleanroom capacity is equally a ceiling on shipments. The $500 million Tampines campus it built in Singapore exists for exactly this reason; the campus opened in June 2026, "more than doubled" the company's cleanroom footprint in Singapore, and is already in volume production (8-K, 2026-08-13; company press release, 2026-06-10).

The soft spots in the contract structure deserve a clear statement. On the call the company noted that customers have provided eight quarters of rolling forecast visibility, with some commitments extending to 2030 (Q3 FY2026 earnings call). Rolling forecasts and commitments are customer planning constructs, not financially binding performance obligations; the company has not disclosed a corresponding backlog dollar figure, nor what share of those commitments carry penalty clauses. Reading "visibility through 2030" as "revenue locked in through 2030" is the easiest mistake to make in this cycle's equipment-stock narrative.

Across the same window, not a single company in this industry is talking the demand outlook down. ACM Research, Onto Innovation, Lam Research and ASML all gave explicit statements pointing to demand moving higher, with zero pointing lower (as of 2026-08-14, all 9 peers have reported). The differences are in the magnitude of growth:

Reading that table correctly requires accepting two corrections. Fiscal calendars are not synchronized across these companies, so the above are each company's reported figures within the same rolling window; and reported figures and guidance figures are two different rulers, which cannot be ranked side by side. What survives the corrections is this: 25% was produced on a quarterly base of $9.115 billion, while 36% was produced on a base less than a tenth that size, and the absolute incremental dollars behind those two numbers are not remotely in the same class. The value of a platform leader in an upcycle has always been in the absolute capture of the increment; high percentages at small companies more often reflect a single product line inflecting.

The divergence in direction is not in the numbers — it's in who stayed quiet. KLA and FormFactor disclosed neither a revenue growth figure nor a demand-direction statement in this window; KLA's disclosure covered only named relationships plus capital expenditure and diluted share-count conventions, and FormFactor's covered only a named relationship. Whether a demand statement appears in their next filings is itself a reading: continued silence would warrant more caution than a soft number.

The valuation comparison is equally clear. As of 2026-08-10, AMAT's forward P/E was roughly 33.1x versus KLA at roughly 38.4x (Yahoo Finance data); all three sector leaders sit in the 97th to 98th percentile of their own five-year P/E ranges. AMAT is the cheapest of the group and also the one carrying the most concentrated downgrade risk — its 26% China exposure is clearly higher than peers focused on leading-edge logic.

The traditional anchors have not stopped working here; they just need the right denominator. The cyclical character of equipment stocks means dividing market capitalization by trailing twelve-month earnings systematically overstates the multiple during an upcycle — that is where the current 43.7x comes from, since it still carries the FY2025 trough quarters, when Semiconductor Systems was shrinking year over year, in the denominator. To see what the market is actually paying, use the fiscal-year basis.

What do 39.7x on the current year and 31.2x on the next year mean? They mean the market has already prepaid for 27% growth in FY2027 — precisely the slope required to travel from $12.76 to $16.23. Yet the company's own statement about 2027 stops at the qualitative "another strong year," with no quantification, and explicitly hangs the ceiling on cleanroom availability, a variable it does not fully control.

Put differently: at this price, the earnings improvement has already been absorbed, and the remaining upside depends on 2027 guidance being delivered — guidance whose outline arrives no earlier than the Q4 report in November. That is the structural reason the stock could not rally on results day, and it is the real pricing content of that 5.1% decline: the market is not denying the 25% growth, it is discounting a 2027 that has not yet been quantified.

Semiconductor equipment is a classic four-year cycle, and locating your position in it takes three kinds of readings: the direction of end-product prices, the degree of capacity tightness, and the pace at which idle capacity is being absorbed.

On price direction, the increases are still spreading. DRAM contract average prices have posted sequential increases of more than 40% (Lam Research FY2026 Q4 earnings call). On mature nodes, TrendForce's 2026-06-30 assessment was that AI component capacity crowding plus foundry production cuts would extend mature-node price increases into 2027, with 2026 mature-node utilization already up to roughly 90%. Price transmission is reaching the equipment layer too — AMAT attributes three years of 300 basis points of gross margin expansion to value-based pricing on a tool-by-tool basis, which in the equipment industry means pricing power has moved from the buyer to the seller.

On tightness, the bottleneck has moved from "should we buy tools" to "where do we put them." Cleanrooms becoming the deciding growth variable, AMAT planning to double quarterly systems output before 2028, the new Singapore campus entering volume production the moment it opened, UMC raising 2025 capital expenditures from $1.5 billion to $2.0 billion and planning at least $5.0 billion over three years — that string of actions describes capacity chasing demand rather than demand chasing capacity.

Payback and inventory are where the cycle-position call really gets made. Management pegs advanced logic foundry, DRAM and advanced packaging together at roughly 80% of wafer fab equipment growth in 2026 and 2027 (Q3 FY2026 earnings call), meaning the contribution to this upcycle is extremely concentrated in three legs. NAND-related equipment recovery has been pushed out to 2027 through 2028, with wafer starts still declining; mature-node ICAPS has only just been revised from "flat" to two years of growth. The signature of a late-cycle phase is a broad-based rally, and right now one leg still isn't at the table while another has only just sat down — this looks more like the middle of the upleg than the top.

Two leading indicators of a peak you can track yourself:

This chapter covers week-scale signals and governs entry and exit timing over the next few weeks; readers holding for a year or more can skip straight to the next chapter.

Start with where the money is going. The deposition and etch sub-industry that AMAT sits in only just moved up from the weakest tier to the middle tier on 2026-08-12, and as of 2026-08-14 it has held that tier for three consecutive days, with the RRG quadrant in the improving zone — but the money-flow score is still -23 (2026-08-14). The detail is less forgiving: the sub-industry's latest money-flow score is -11, a weak-outflow reading, with 22 consecutive trading days of net outflows, $301.2 million of net outflows over the past 5 days and $2.3 billion over the past 20 days, while the 5-day price change over the same stretch was +1.5% (as of 2026-08-14).

Money has been leaving for 22 straight days while price has kept climbing — price is running ahead of the money. At the sector level that divergence usually converges one of two ways: money comes back to catch up, or price comes back to find the money. Historically there have been 5,890 instances of a move from the weakest tier up to the middle tier (covering 2021-05-14 through 2026-07-17), with a median 20-trading-day forward return of +0.0% and a 56% probability of being higher — the tier upgrade by itself provides essentially no expected return, it merely rules out "worst."

Where did the money go instead? The top three sub-industries by inflow are gold at 58, quality factor at 44 and silver at 44 (2026-08-14); the top three by outflow are retail REITs at -73, industrial REITs at -66 and tobacco at -62. At the sector-tier level, the only sector standing in the top tier is financials, with technology crowded into the middle tier alongside materials, energy, industrials, healthcare and consumer discretionary. The single most direct rotation pair signal is money moving from technology into financials, at a strength of 1.58 (2026-08-14). At the same time, materials and industrials complexes are forming: steel and iron ore on August 4, gold miners on August 7, copper miners and silver miners on August 10, defense technology on August 7, drones on August 10, space exploration on August 13, and the defense and aerospace index on August 14 all moved up into the top tier in succession. This is a textbook picture of money moving out of growth and into cyclicals and safe havens.

On the macro backdrop, the index level is still constructive: the market regime score is 54, and SPY at 776.32 sits 2.66% above its 20-day moving average of 756.19 (2026-08-14). But liquidity is tightening: net liquidity on the FRED basis is $5.79 trillion, down 0.8% week over week and in a contracting state, with the 10-year Treasury at 4.63% and the liquidity tier reading as cautious. When the index is still rising, liquidity is receding and money is running toward gold and financials all at once, high-multiple growth stocks are usually the first to bleed.

On price location, AMAT currently closes roughly 9% below its 50-day moving average while sitting roughly 31% above its 200-day. The short-term trend has broken while the intermediate trend is intact — this is a drawdown, not a trend reversal, and it is exactly the kind of location that tempts people to reach for it too early.

Long term (quarters to years): worth following. Three things stack up behind that conclusion — revenue up 25% year over year with next quarter's guidance midpoint adding another 12.5%; 13 consecutive quarters of year-over-year gross margin expansion with the operating margin hitting an all-time high of 34%; and free cash flow of $2.33 billion at a cash conversion rate above 100%. This is not a quarter propped up by one-time items, and the cash quality of the profit stands up to scrutiny. Three costs come with it: 39.7x on the FY2026 basis and 31.2x on the FY2027 consensus basis, a price that has already prepaid for another 27% of growth next year; China at 26% of revenue, the heaviest geopolitical exposure among peers; and gross margin held down by ramp costs during the expansion, with the company itself willing to say only "roughly flat at the corporate level." As a holding, this is a position driven by verification points — each quarter you need to confirm only two things, whether gross margin has resumed expanding and whether operating expenses as a share of revenue have come down. If both stay on track, keep holding; if either moves the wrong way for two consecutive quarters, it is time to reassess whether the multiple is still worth paying.

Short term (days to weeks): do not chase, wait for the money to come back. The sub-industry has now seen 22 consecutive trading days of net outflows, $2.3 billion cumulatively over the past 20 days, while the 5-day price change is still positive — price is running ahead of the money. The sector has only just escaped the weakest tier, but the historical median 20-trading-day return after comparable tier changes is 0.0%, so the upgrade offers no expected return. Meanwhile the largest single rotation is money flowing from technology into financials, net liquidity is contracting week over week, and the 10-year Treasury is at 4.63%. On price, sitting 9% below the 50-day moving average is the first pullback leg after a trend break — expensive to catch and low in win rate. Two things need to appear together: the sub-industry money-flow score turning from negative to positive with the consecutive-outflow day count resetting to zero, and price reclaiming the intermediate moving average. Until then this belongs on a watchlist, not a trading list.

The second half turns both conclusions into something checkable — how each of the two entry conditions is defined, which anchor each of the three reference levels is derived from, and exactly which numbers to watch daily, weekly and quarterly.

Entry condition one (money flow, trackable daily): the deposition and etch sub-industry's money-flow score turns from the current -11 to positive and holds above the zero line for 3 consecutive trading days, while the consecutive net-outflow day count resets from the current 22 days to zero. The point of this condition is to confirm the bleeding at the sector level has stopped — against a backdrop of 22 days of net outflows and $2.3 billion cumulatively over 20 days, any individual stock's advance is being made against the flow of money. This can be checked directly in the sector money-flow section of the daily review.

Entry condition two (hangs on the next quarterly report, derived from the arithmetic in chapter 1): in the FY2026 Q4 report (expected mid-November), non-GAAP gross margin comes in no lower than 50.4% and non-GAAP operating margin reaches or exceeds 34.8%. That second figure is equivalent to operating expenses falling from 16.4% of revenue in Q3 to below 15.6% — that is, expenses growing only about 7% sequentially while revenue grows 12.5%. Whether the company can pull that off in a quarter when it added 1,500 net new employees is the only quantified answer to the question of whether this expansion is investment or loss of control. Only when both conditions hold do you have money and fundamental confirmation at the same time.

Three reference levels, each anchored to something publicly verifiable:

The invalidation level is still roughly 24% below the current price, which leaves too much empty space in between, so it needs an early warning to go with it: if the price loses $473.03 while the sub-industry money-flow score remains negative, treat that as money and price turning together — do not add, cut any existing position to half the planned size, and let the November quarterly numbers decide what happens next. A price break alone, with the money-flow score already positive, does not trigger this warning — that is more likely sector rotation of ownership than the end of the trend.

Two numbers to watch daily: the deposition and etch sub-industry's money-flow score and its consecutive net-outflow day count (currently -11 and 22 days) — the score turning positive with the day count resetting to zero satisfies entry condition one; conversely, if the score slides from -11 through -30, the bleeding is accelerating and the lower edge of the watch zone has to be moved down. Also the price's position relative to the 50-day moving average (currently roughly 9% below) — converging to within 3% indicates repair is underway.

Three numbers to watch weekly: net liquidity on the FRED basis (currently $5.79 trillion, down 0.8% week over week, in a contracting state), where two consecutive weeks of positive week-over-week change is a broad positive for high-multiple growth stocks; the 10-year Treasury yield (currently 4.63%), where a break above 4.8% would directly compress tolerance for a 31x multiple; and the technology-to-financials rotation pair strength (currently 1.58), where a fall below 1.0 would indicate the bleeding out of growth is stopping.

Four numbers to watch quarterly: non-GAAP gross margin (currently 50.4%, guided flat; a resumption of sequential expansion is the positive confirmation); non-GAAP operating expenses as a percentage of revenue (currently 16.4%, needs to fall below 15.6%); AGS services revenue growth year over year (currently +22%; a rollover is the earliest reading of a cycle peak); and China's share of revenue (currently 26%; if policy causes it to drop below 15% in a single quarter, the full-year revenue base needs to be recalculated).

We will publish a follow-up note if any of the following occur: the company issues first quantified FY2027 guidance; gross margin or the operating expense ratio moves the wrong way for two consecutive quarters; the conventions around China-related export controls change; or the sub-industry money-flow score turns positive for 3 consecutive days, triggering entry condition one.

Catalyst calendar:

We watch every item on that tracking list daily — the sub-industry money-flow score turning positive, the consecutive net-outflow day count resetting to zero, the distance between price and the intermediate moving average converging. Any one of those triggering will be written up as a follow-up note and sent out with that day's daily review, so readers don't have to sit refreshing the data pages themselves. Every company figure in this report comes from an 8-K filing, a quarterly financial statement or an earnings call transcript, each labeled with its source and date so it can be checked line by line against the original. The money-flow, sector-tier and macro-liquidity figures come from the corresponding data pages on the site, with dates likewise shown in parentheses. The judgments in this report will change as the data changes, and when they change we will say plainly which number moved first.

This article was produced on the NextPick research workbench; every money-flow, tier and valuation figure it cites can be verified on the corresponding data surfaces of the site, and the moment any tracking condition listed in the report triggers, an addendum will go out together with that day's daily review.