What happened
Applied Materials reported fiscal third quarter results on 13 August 2026 showing total revenue of 9.115 billion dollars, up 25 percent from a year earlier and 15 percent higher than the previous quarter. Adjusted earnings came in at 3.50 dollars per share, a rise of 41 percent year on year and 22 percent sequentially. By any conventional measure it was the strongest quarter in the history of the company.
The shares fell 4.94 percent in after hours trading to 508.15 dollars. Management had also raised its outlook and guided toward another record quarter, citing accelerating demand from the global build out of artificial intelligence infrastructure, and investors sold anyway.
Two details explain the reaction. Management guided to flat gross margins in the fourth quarter, meaning the percentage of each sale left after the direct cost of making the product would stop improving. Investors had been extrapolating continued margin expansion, and a pause in that trend removes one of the drivers of future profit growth.
The second detail was China. Sales to Chinese customers accounted for roughly 28 percent of total revenue in the quarter, down from around 35 percent a year earlier. That decline reflects export restrictions and a shifting customer mix, and it raises the question of whether AI demand elsewhere can keep growing fast enough to offset a shrinking Chinese business. Adding to the pressure, the stock had already risen roughly 200 percent into the results, leaving very little room for a merely excellent quarter.
Why it matters
Applied Materials does not make chips. It makes the machines that make chips, which places it near the very start of the supply chain that produces every processor in every data centre, phone and car. That position makes its results one of the earliest and most reliable indicators of how much the technology industry actually intends to spend, as opposed to how much it says it intends to spend.
Revenue up 25 percent tells you the AI build out is not slowing. Companies do not order semiconductor manufacturing equipment speculatively. The machines cost tens of millions of dollars each, take months to install, and are ordered against factory capacity plans that stretch years ahead. A record quarter for equipment is a strong signal about chip supply in 2027 and 2028.
The share price reaction tells you something different and equally important, which is that expectations have become extremely demanding. A stock that has tripled over roughly a year is priced for continuous acceleration. When the company delivered acceleration in revenue but not in margins, the market took profits. This is what traders call selling the news, and it is a recurring feature of markets near records.
The China number is the strategic story. A fall from 35 percent to 28 percent of revenue in a single year represents a structural realignment of the semiconductor equipment market along geopolitical lines, and it is happening while overall demand is booming, which conveniently disguises it. If AI spending ever cools, that disguise disappears.
Explained simply
Applied Materials sells the picks and shovels of the AI gold rush. This quarter it learned that when everyone already assumes you will find gold, merely finding gold is not enough.
A share price is not a measure of how well a company is doing. It is a measure of how well a company is doing compared with what people already expected. If the market expects revenue growth of 25 percent and the company delivers 25 percent, the share price does very little, because that outcome was already reflected in the price people paid yesterday. Only the surprise moves the number.
Gross margin is the simplest way to see whether growth is profitable. If a company sells a machine for 10 million dollars and it costs 5 million to build, the gross margin is 50 percent. Rising margins mean each additional sale contributes more profit than the last, which is the most powerful thing that can happen to a growing business. Flat margins mean profit grows only as fast as sales, which is good but far less exciting.
The reason a shrinking China share matters is concentration. If more than a third of your sales come from one country and that share falls sharply, you must replace that revenue elsewhere just to stand still. Applied Materials replaced it and grew 25 percent on top, which is impressive, but the replacement demand is coming from AI data centre construction, a single spending cycle driven by a handful of very large buyers.
This is the core question hanging over the entire sector. AI infrastructure spending is real, enormous and currently accelerating. It is also concentrated in a small number of companies making long term bets. Equipment makers sit at the point in the chain where any change in those bets shows up first.
What it means for you
Most British investors own a slice of this story without having chosen it. A global equity tracker in a workplace pension typically holds somewhere around 60 to 70 percent American shares, with semiconductor and AI related companies among the largest positions. If you have a default pension fund and have never changed it, semiconductors are one of your larger exposures.
That is not necessarily a problem, but it is worth knowing the shape of the risk. If you also hold a technology focused fund or individual chip shares in an ISA, you may own the same theme three times over through different wrappers. Checking the top ten holdings of each fund you own takes about ten minutes and frequently produces a surprise.
For anyone tempted by the sector after a fall of nearly 5 percent, remember the context. The shares are down from a level reached after a rise of roughly 200 percent. A 5 percent dip in a stock that has tripled is not a bargain, it is a rounding error, and buying individual semiconductor companies means accepting a level of volatility that most retirement savings are not designed to absorb.
The practical takeaway for most people is simpler. Results like these confirm that the AI spending cycle is still running, which supports the earnings underpinning global share indices, and therefore supports pension balances. Continuing regular monthly contributions captures that without requiring any view on which individual company wins.
The bigger picture
The semiconductor equipment industry moves in long cycles, historically boom and bust affairs tied to memory chip prices and consumer electronics demand. The current cycle is different in character because the buyer is data centre capacity rather than the consumer, and because governments in the United States, Europe, Japan and China are all subsidising domestic chip manufacturing for strategic reasons. That combination has produced an unusually long and unusually well funded upswing.
The risk in every such cycle is the same. Capacity ordered today arrives in two or three years, and if demand has moderated by then the industry finds itself with expensive machines and no orders. The fall in Chinese revenue, from around 35 percent to 28 percent of sales, is an early sign that the market is fragmenting into separate regional supply chains, each building capacity somewhat independently of the others.
What to watch next is the fourth quarter margin outcome against the flat guidance, order commentary from other equipment makers, and the capital spending plans announced by the large cloud computing companies. Those plans are the demand that everything in this chain ultimately rests on.
