Applied Materials Bets Big on AI Chip Boom at SEMICON West 2026

Applied Materials used its Tuesday, October 13, 2026 breakfast at SEMICON West to argue that artificial intelligence has started a multi-year spending cycle in wafer fab equipment, and that the Santa Clara company is in position to capture a large share of it. Chief Executive Gary Dickerson said demand across leading-edge foundry logic, DRAM and advanced packaging is running ahead of what the industry can build, with chips "never more valuable" and fabs running near full utilization. Shares fell 8.01 percent that day to $457.69, even though the stock is up 128 percent over the past year. That is one sign the market has already priced in a lot of the optimism and is now weighing the constraints management flagged right alongside its growth targets.

"The industry is now focused on capacity innovation, yield and output innovation because demand is much bigger than supply," Dickerson said. He added that Applied supplies materials for more than 60 percent of the transistors and wiring on a chip, giving it "an enormous opportunity and responsibility" to make chips faster and more energy efficient.

What Applied is really selling

Applied makes the deposition, etch, polishing and metrology tools that turn silicon wafers into chips. The pitch at SEMICON West 2026 was that three markets matter most right now: leading-edge foundry logic, DRAM and advanced packaging. The company expects about 75 percent of incremental wafer fab equipment growth through 2030 to come from those three areas, and it already holds the number one position in each. Management is targeting more than 2 percentage points of wafer fab equipment market share gains by 2030.

The numbers it put on the table are substantial. Ten-year revenue compound annual growth of 12.2 percent to $34.3 billion. Non-GAAP gross margin up 680 basis points to 50 percent. Non-GAAP earnings per share growth of 22 percent a year, nearly double revenue. Free cash flow averaging 21 percent of revenue and return on invested capital above 30 percent. For 2030, the targets are 54 percent non-GAAP gross margin, 40 percent non-GAAP operating margin, more than $12 billion in Applied Global Services revenue, and more than $2 billion in display and photonics revenue.

Bryce Hill, the chief financial officer, said the margin path comes from a better product mix and operating leverage rather than steady gains. "About 100 basis points of annual gross margin improvement on average, though not in a straight line," he said, adding that 200 to 300 basis points of operating leverage are needed to reach the 40 percent operating margin goal.

The EPIC bet

A day before the event, Applied opened its EPIC center in Sunnyvale, California. The Enabling Process Innovation Collaboration facility brings chip makers, equipment firms and research institutions together under one roof. Intel, TSMC, Micron, KIOXIA, SK Hynix and Samsung were among those present. The facility has more than 100 test vehicles and customer reticles, and management claims the parallel development model can cut innovation time by about half compared with a traditional sequential process.

Chief AI Officer Yu Cao framed the company's view in blunt terms: "We are early in the AI revolution. The opportunity ahead of us is larger than what's behind us, and every layer of it starts with materials engineering." That is a claim about Applied's relevance, but it also carries an assumption that the current AI demand curve holds through the decade, which is the part that is not yet tested.

Services, software and the connected fab

Tim Deane, who runs Applied Global Services, pitched the services arm as a second growth engine. AIX, the company's Actionable Insight Acceleration platform, has 37,000 chambers connected, about 80 percent of them remotely, and the company expects to exceed 90,000 by 2030. Eighteen quarters of data show 30 percent faster service resolution on connected tools and fewer escalations. Services revenue grew 20 percent in calendar 2026, and the company expects a mid-teens compound annual growth rate toward the $12 billion 2030 target.

A useful example: the company says it can now predict electrostatic chuck failures 30 days ahead with 90 percent accuracy. That turns an unplanned outage into a planned maintenance window. Whether customers actually see those yield gains at scale is harder to verify than the headline suggests, but the renewal rate above 90 percent and service contracts averaging more than 2.5 years indicate customers are buying the pitch.

The bottlenecks management admitted

The candor was the more interesting part of the session. Management said demand is not the constraint; adding capacity is. Clean room space is the biggest near-term limit. Supply chain lead times, hiring, manufacturing fixtures, test capacity, energy availability and financing for fab construction all follow. The company is tracking more than 100 fab projects worldwide and added more than 10 new projects in each of the last two quarters.

China is a separate problem. Access has fallen from 100 percent in the early 2020s to about 70 percent in 2025. Applied cannot serve leading-edge, DRAM or NAND in China because of trade restrictions. It can still compete at the 28-nanometer node. That leaves a meaningful chunk of the world's second-largest chip market off limits, and it is not clear how much of the 30 percent share loss can be recovered elsewhere.

Hyperscaler capital spending could exceed $1 trillion, and management now expects the semiconductor industry to reach between $1.7 trillion and $3.0 trillion by 2030, up from an earlier $1 trillion estimate. Applied also said agentic AI compute uses 2 to 4 times more wafers than generative AI pods, which is the arithmetic behind the higher forecast. Those figures are forward-looking and broad enough that they should be treated as directional, not precise.

Why the market still sold the stock that day

A 128 percent gain over a year leaves little room for disappointment. At $457.69 the stock sits at 39.4 times earnings and a $363 billion market capitalization, and one third-party analysis rated it overvalued relative to its fair value estimate. The stock had set a 52-week high of $739.67, so Tuesday's decline also reflects a pullback from levels that assumed flawless execution. The targets are for 2030. A lot can change between now and then, and clean room square footage is not something Applied can build itself.

What comes next

Keith Wells, who leads the Imaging and Process Control Group, expects $3 billion in systems and service revenue in 2026 in metrology and inspection. Applied plans new optical inspection products in the first half of 2027 aimed at advanced packaging. On the logic side, gate-all-around transistors require atomic-level materials engineering with 10-nanometer spacing between nanosheets, and DRAM is moving from 6F-squared to 4F-squared, which Applied says expands the served market by more than 40 percent on the way to 3D DRAM. Whether those technology bets pay off depends on customer adoption speed, not just Applied's roadmap. For now, the company has staked a clear claim: it is the materials and integration winner of the AI chip build-out, and the next four years will show whether that claim survives contact with capacity, policy and competition.

For more on equipment and capacity planning, see Reuters. Browse our coverage by sector at semiconductors or scroll the archive.