Applied Materials Trades at a Record 16x Sales While Its Biggest Market Shrinks

---

Start with what is not in dispute, because Applied Materials is a great company and pretending otherwise would forfeit the argument. On May 14, 2026, the company reported fiscal second-quarter revenue of $7.91 billion, a record, up about 11% from a year earlier and 13% from the prior quarter. Gross margin came in at 49.9% on a GAAP basis and 50.0% adjusted — the richest in more than two decades. Management raised its outlook for the year, guiding fiscal third-quarter revenue to roughly $8.95 billion give or take half a billion, and told investors it now expects its semiconductor-equipment business to grow more than 30% in calendar 2026, with advanced packaging up more than 50%. These are not the numbers of a company in distress. They are the numbers of a company at the top of its game, in the middle of the largest capital-spending wave the chip industry has ever seen.

And yet the stock tells a story the income statement does not quite support. Over the year into mid-June 2026, Applied Materials rose roughly 216%, more than tripling. By June 16, its price-to-sales ratio had pushed past sixteen — by the reckoning of at least one valuation tracker, the highest in the company's history — while the trailing price-to-earnings multiple sat near fifty-five. For a business whose end demand is set not by consumers refreshing phones but by a handful of chipmakers deciding, year to year, how many tens of billions of dollars of fab capacity to build, those are extraordinary numbers. This is a company that has been doing essentially the same thing since the 1960s. It has never been a secular grower in the way software is. It is the supplier of capital equipment to the most violently cyclical industry in the modern economy. The question this article asks is simple: what happens to a cyclical business when the market stops pricing it as cyclical?

The cycle priced as a secular

There is a category error at the heart of the Applied Materials trade, and it is worth naming precisely. Wafer-fab equipment — the lithography-adjacent etch, deposition, ion-implant, and inspection tools that Applied sells — is the textbook example of a capital-goods cycle. When chipmakers are flush and confident, they order machines by the hundreds and the equipment vendors print money. When demand softens, or when too much capacity has already been built, those same chipmakers slam the brakes, and orders evaporate with terrifying speed. Applied has lived through this perhaps a dozen times. Its revenue has fallen year over year in 2001, 2008, 2012, 2019, 2023 — each time on the back of a downturn in the underlying chip cycle. The machines do not wear out on a smooth schedule; they are bought in bursts and then not bought at all.

What is happening now is that the AI build-out has stretched the current upswing long enough, and made it large enough, that investors have begun to extrapolate it as a straight line. The narrative has shifted from "Applied is enjoying a strong year in a cyclical business" to "Applied is the picks-and-shovels play on a secular AI supercycle that has no obvious end." That reframing is what justifies a sixteen-times-sales multiple. But the reframing is a bet, not a fact. The leading-edge logic and high-bandwidth-memory capacity being built for AI today will, at some point, satisfy demand — or overshoot it. Memory capex in particular has always been the most lumpy and self-correcting corner of the industry; DRAM and NAND makers are famous for building too much, crashing prices, and then starving the equipment vendors for a year or two. Pricing a cyclical at a secular multiple does not make the cycle disappear. It simply means that when the cycle turns, the multiple and the earnings fall together — the dreaded double-compression that makes semiconductor-equipment drawdowns so brutal.

The denominator that is quietly shrinking

Now look at the geography the headline glosses over. For years, China was Applied Materials' single largest market — at the peak, close to 40% of total systems-and-services revenue. Chinese chipmakers, racing to build out domestic capacity in trailing-edge and mature nodes, bought Applied's tools voraciously, and a meaningful slice of those years' "growth" was really a one-country surge. In fiscal 2025, China had fallen to roughly 28% of revenue. By the second quarter of fiscal 2026, it represented around 24–27% of systems-plus-services revenue depending on the cut. The largest customer geography in the company's history has lost something like a third to a half of its share of the pie in a few short years, and management is explicit that it expects China to remain depressed for several more quarters.

This matters in a way that the consolidated revenue line disguises. When a company's biggest market is contracting and total revenue is still hitting records, the rest of the business — leading-edge foundry-logic and AI-driven memory — has to be growing fast enough to fill the hole and then some. That is genuinely impressive while it lasts. But it also means the reported growth rate is flattered by a base effect: the China decline is now far enough along that its drag on the year-over-year comparison is beginning to ease, which makes the surviving businesses look like they are carrying the company more cleanly than they will once China stops shrinking and becomes a flat, structurally smaller contributor. The denominator is being quietly re-based lower. Investors extrapolating 30%-plus equipment growth should ask how much of the easy comparison comes from a China collapse that is mostly behind the company rather than a leading-edge boom that is durably ahead of it.

The policy switch someone else controls

Compounding the China problem is that its trajectory is not in Applied Materials' hands at all. It is set in Washington. The company has told investors to expect roughly a $600 million revenue hit to fiscal 2026 from expanded U.S. export restrictions, on top of around $110 million of product that went unshipped in the prior fiscal fourth quarter under an affiliate rule — call it roughly $710 million of revenue that policy, not the market, removed from the table. And the risk is asymmetric and ongoing: proposed legislation circulating in Congress would tighten the screws further on equipment sales to China, and each new tranche of controls lands as a surprise the company can only react to.

For a forensic investor, this is the moat-versus-loophole distinction in sharp relief. Applied Materials has a genuine technological moat in many of its product lines. But a meaningful share of its recent China revenue depended on a regulatory window — the ability to sell mature-node and certain advanced tools to Chinese buyers — that the U.S. government has been steadily closing. Revenue that exists at the sufferance of an export-control regime is not the same quality as revenue protected by patents and switching costs. It can be legislated away in an afternoon, and it has been, repeatedly. A company priced for perfection is carrying a line of business whose size is decided by people who do not work there and do not have its interests at heart.

When GAAP beats adjusted, look harder

Here is the detail that should make a careful reader sit up. In most quarters, a company's adjusted (non-GAAP) earnings come in higher than its GAAP earnings, because management strips out stock compensation, restructuring charges, amortization, and other real-but-inconvenient costs. The adjusted number is the flattering one; that is the whole point of it. In Applied Materials' second quarter of fiscal 2026, the relationship inverted: GAAP earnings per share were $3.51, while non-GAAP earnings per share were just $2.86. GAAP beat adjusted by sixty-five cents — a gap of more than 20%.

That inversion does not happen by accident. It happens when something large and non-operating flows through the GAAP income statement that management then excludes from its adjusted figure — almost always a gain. In Applied's case, the company holds strategic equity investments, including a roughly 9% stake in the Dutch advanced-packaging specialist BE Semiconductor (Besi) that it accumulated in 2025. Under accounting rules, unrealized gains on such marketable equity stakes flow through net income each quarter. When the investee's stock rises, Applied books a paper gain that inflates GAAP profit without a single extra tool being sold. Strip that gain back out — which is exactly what the lower adjusted EPS does — and you get a cleaner read on the actual operating business. The headline $3.51 is real in an accounting sense, but a chunk of it is a mark-to-market on someone else's share price, not earnings from making semiconductor equipment. An investor anchoring on the bigger GAAP number is paying an earnings multiple on profit that the company itself does not consider part of its core performance, and that will reverse the moment Besi's stock falls.

The customers who can turn off the tap

Applied Materials sells to a customer base so concentrated it can be listed on one hand: TSMC, Samsung, Intel, the leading memory makers, and a thinning roster of Chinese fabs. This is the nature of the industry — there are only a few firms on Earth that build leading-edge chips, and they are the only ones who can afford the most advanced tools. But concentration cuts both ways. The same handful of customers that drives record revenue when they are all spending can crater it when even one of them pauses.

The current upswing leans heavily on TSMC, which is guiding 2026 capital expenditure toward $52–56 billion, up roughly a third year over year, to feed AI demand. That is the engine. But it is one engine. TSMC's capex plans are a forecast, revisited quarterly, and the history of this industry is littered with capex guides that were cut hard when end demand wobbled. Samsung and Intel have both, at various points in the past two years, deferred or restructured their most ambitious fab plans. A single large customer trimming its build-out by ten or fifteen percent would knock a visible hole in Applied's order book — and at sixteen times sales, the stock has no cushion for that. The market is pricing the synchronized, simultaneous, sustained spending of every major chipmaker at once, indefinitely. That is the best case, not the base case.

The trailing edge and the home-grown rival

There is a slower, more structural threat lurking underneath the AI headline, and it concerns the part of Applied's business that built it: ICAPS — the IoT, communications, auto, power, and sensor chips made on mature and trailing-edge nodes. For years this was a reliable, high-margin franchise, and much of the China surge was ICAPS demand. But two things are happening at once. First, the trailing-edge build-out, especially in China, has front-loaded an enormous amount of capacity, and that pull-forward is now unwinding — there are only so many mature-node fabs the world needs, and a lot of them got built in 2022–2025. Management's own guidance that China and worldwide ICAPS will be merely "flat to slightly higher" for the calendar year is a polite way of saying the growth has gone out of it.

Second, and more dangerous over the long run, domestic Chinese equipment makers are climbing the ladder fastest precisely in the lower-end, trailing-edge tools where Applied once had the field to itself. Export controls that lock Applied out of selling advanced tools to China simultaneously hand its Chinese competitors a captive home market and a powerful incentive to localize — to build their own etch and deposition machines for the mature nodes. The very policy that protects America's leading-edge advantage is incubating a generation of Chinese rivals at the trailing edge, and the trailing edge is where the volume lives. This is the long, quiet erosion that a 16x-sales multiple, fixated on the AI leading edge, prices at zero.

Buybacks, dividends, and the flattering per-share math

One more piece of the per-share story deserves scrutiny. Applied Materials is a prodigious returner of cash, buying back enormous quantities of its own stock and paying a growing dividend. Buybacks are not free money; they are a use of capital that shrinks the share count and thereby lifts earnings per share even when total net income is flat. Over the long sweep, Applied has retired a large fraction of its shares, and a meaningful slice of its multi-year EPS growth is arithmetic — fewer shares dividing the profit — rather than a bigger profit. That is a legitimate and shareholder-friendly strategy in a cash-rich business. But it is worth separating in the mind: the growth rate of the company's actual earning power is lower than the growth rate of its earnings per share, because buybacks are doing part of the lifting. When you pay fifty-five times earnings, you want to be sure you are buying growth in the business, not growth in the denominator. A buyback executed at a record valuation, moreover, is a poor use of cash — the company is retiring shares at sixteen times sales, the most expensive level in its history, which is precisely when buybacks destroy rather than create per-share value.

What the bulls genuinely get right

It would be intellectually dishonest to pretend the bull case is hollow, because it is not — it is one of the strongest in the sector, and a fair forensic account has to concede where the optimists are right. Applied Materials is a genuinely dominant company with a real, durable technological moat across deposition, etch, and process control. It sits at the absolute center of the most important capital-spending wave in technology history: the build-out of AI infrastructure requires more leading-edge logic and more high-bandwidth memory, and both require Applied's tools. Management is not bluffing when it says equipment demand should grow more than 30% this year — TSMC's capex guide and the broader foundry-logic and DRAM build-out genuinely support it, and Applied has explicitly positioned itself in the three areas (leading-edge foundry logic, DRAM, advanced packaging) that now drive more than 80% of WFE spending growth. Advanced packaging, where Applied's collaboration with Besi targets hybrid bonding, is a structurally growing field that may indeed be less cyclical than the historical equipment business, because it scales with chip complexity rather than raw wafer volume. The 50% gross margin is the richest in a generation and reflects real pricing power and mix improvement, not financial engineering. The company gushes cash, carries a fortress balance sheet, and returns capital generously. And the China decline, painful as it is, has the perverse benefit of de-risking the revenue base — the more the company's growth comes from leading-edge AI rather than from a politically exposed China, the higher the quality of what remains. If the AI build-out runs for another three or four years without a memory glut or a foundry capex cut, Applied's earnings will grow into a good part of today's multiple, and the bears will look foolish. None of this is a weak company. The entire argument is about price, cyclicality, and the quality of the headline number — not about whether Applied Materials is a fine business. It is.

The asymmetry that should worry you

Put the pieces together and what emerges is not a short thesis on a bad company but a caution about an asymmetry. On the upside, the bull case is already substantially in the price: a record sixteen times sales and fifty-five times earnings assume that AI equipment demand grows at thirty-percent-plus rates not just this year but for years, that China stabilizes rather than deteriorates further, that no major customer trims its capex, that memory does not glut, and that the GAAP earnings — flattered by investment gains — are a fair representation of earning power. For the stock to work from here, most of those things have to go right at once. That is the definition of priced for perfection.

On the downside, the list of things that could go wrong is long and each is plausible: a single export-control tranche, a TSMC or Samsung capex trim, a memory inventory correction, a reversal in Besi's stock that flips the GAAP earnings tailwind into a headwind, or simply the market deciding that a forty-year-old cyclical does not deserve a software multiple. When semiconductor-equipment cycles turn, they turn fast, and the multiple compresses at the same time the earnings fall. A stock that has tripled in a year has a very long way to fall before it reaches anything resembling a historical-average valuation. The bull case requires the future to cooperate completely. The bear case requires only that the cycle, eventually, behave like a cycle.

The kicker

None of this means the music stops tomorrow. AI capex is real, Applied's tools are essential, and a powerful upswing can run longer and higher than any sober analyst expects — that is the nature of manias and of genuine booms alike, and they are maddeningly hard to tell apart in real time. But the job of a forensic investor is not to call the top; it is to notice when the price has stopped reflecting the business and started reflecting a story. At sixteen times sales, with its largest market in retreat, its headline earnings flattered by a stock-market gain, and its growth dependent on a handful of customers all spending at once, Applied Materials is no longer priced as the supplier of machines to a cyclical industry — it is priced as the industry's destiny.

The cycle does not care what multiple you paid; it only knows that it has never, in forty years, failed to turn.