Hook
The code didn't fail. The policy did.
Applied Materials reported another quarter where China revenue—once the company's growth engine—contracted further into the rearview mirror. The headline reads "worsening challenges," but the ledger tells a sharper story. Over the past 12 months, AMAT's China exposure has dropped from roughly 30% of total revenue to the low-20s, a bleed that traces directly through the U.S. Department of Commerce's Bureau of Industry and Security (BIS) export control framework. This isn't a market cycle. This is a structural amputation.
Tracing the bleed through the gateway: every denied license application, every suspended service contract, every "we cannot ship this spare part" email—they all compound into a single, measurable outcome. The world's largest semiconductor equipment maker is being systematically priced out of the world's largest semiconductor equipment market.
Context
Applied Materials is not a chipmaker. It's the company that makes the machines that make the chips. Its deposition tools (PVD, CVD, ALD), CMP systems, and ion implantation equipment are foundational to every advanced wafer fab on Earth. When TSMC builds a 3nm line, AMAT equipment fills the cleanroom. When Samsung ramps GAA transistors, AMAT process recipes define the film stacks. When SK Hynix expands HBM production for AI accelerators, AMAT's advanced packaging tools are on the floor.
The company holds the #1 position globally in thin-film deposition (~35-40% share), CMP (~60%+), and ion implantation (~50%+). In etching, it trails Lam Research and Tokyo Electron but remains a top-three player. Its R&D budget runs approximately $3 billion annually—about 12-14% of revenue—keeping it at parity with Lam and TEL, slightly below ASML's intensity.
The problem: China is the largest semiconductor equipment market on Earth, representing roughly 30% of global demand. And AMAT, as the most American of American equipment makers, is the most exposed to Washington's tightening export control regime.
The current rules restrict AMAT from shipping equipment that enables advanced logic (16nm/14nm and below) and advanced memory (128+ layer NAND, sub-18nm DRAM) to Chinese customers. License applications for such shipments are routinely denied. The trend line is not improving—it's worsening, as the article's title correctly notes.
Core: The Systematic Teardown
Let me be precise about what's happening, because the market narrative tends to blur the mechanics.
The Revenue Bleed
AMAT's China revenue has been in decline since the October 2022 export controls first landed. The trajectory:
- FY2022: China represented ~28-30% of revenue
- FY2023: Dropped to ~25%
- FY2024: Further erosion to ~22-23%
- Current run rate: Trending toward the high-teens
This isn't a demand problem. Chinese fabs want AMAT equipment. The problem is supply-side: the U.S. government has decided that certain AMAT products cannot be sold to certain Chinese customers, period. The company's sales team in Shanghai and Beijing can generate all the pipeline they want—the compliance team in Santa Clara will kill the deal.
The Service Contraction
Here's what most analysts miss: the export controls don't just stop new sales. They also degrade the installed base.
Chinese fabs that purchased AMAT equipment before the controls still need spare parts, software updates, and process optimization support. But the BIS rules increasingly restrict even these aftermarket services for advanced nodes. AMAT's field service engineers in China are operating with one hand tied behind their backs—they can service mature-node tools, but advanced-node support requires license approvals that may never come.
This creates a slow-motion erosion of customer stickiness. Chinese fabs that can't get AMAT service will eventually design AMAT equipment out of their future roadmaps. The installed base becomes a stranded asset, and the next fab expansion will go to domestic Chinese toolmakers or non-U.S. suppliers.
The Compliance Tax
Every major U.S. equipment maker now runs a shadow compliance organization. AMAT has dedicated teams tracking BIS rule changes, reviewing end-user certificates, and building jurisdiction-specific product matrices. This is not free. It's a tax on every transaction, adding weeks to deal cycles and requiring legal review on routine orders.
The cost isn't just financial—it's strategic. AMAT's product roadmap must now account for "exportable versions" of tools, which means engineering resources diverted from pure innovation to compliance-driven design modifications. The company is effectively building two product lines: one for the free world, one for the restricted list.
The Opportunity Cost
Here's the number that keeps me up at night: China's AI chip sector is exploding. Huawei's Ascend series, Cambricon, Biren—these companies are designing advanced AI accelerators that need leading-edge manufacturing. Their fabs (SMIC, Hua Hong) are expanding capacity. But they can't buy AMAT's best tools.
The demand is real. The wallet is open. And AMAT cannot participate.
Meanwhile, TSMC, Samsung, and Intel are building new fabs in Arizona, Ohio, Japan, and Germany—all funded by government subsidies (CHIPS Act, European Chips Act, Japan's semiconductor revival plan). AMAT will supply those fabs. But the revenue per fab is lower than what China's massive buildout would have generated, and the timeline is longer.
The Competitive Vacuum
Every dollar AMAT loses in China doesn't disappear—it gets redistributed. The beneficiaries:
- Chinese domestic toolmakers: Naura (北方华创), AMEC (中微公司), Piotech (拓荆科技) are making real progress in deposition and etching for mature nodes. They're not at AMAT's level for advanced nodes, but they don't need to be—they just need to be "good enough" for China's self-sufficiency push, backed by the $47 billion National IC Industry Investment Fund (Big Fund Phase III).
- Non-U.S. suppliers: Tokyo Electron (Japan) and ASML (Netherlands) face their own export restrictions, but they have more flexibility in certain product categories. TEL's coater/developer and etch tools can sometimes substitute for AMAT equivalents in non-restricted applications.
- The gray market: Let's not pretend it doesn't exist. Used AMAT equipment flows into China through third-party channels, often stripped of advanced features but still functional. This doesn't help AMAT's revenue, but it does keep Chinese fabs running on AMAT hardware—which creates a strange long-term dependency that could theoretically be monetized if controls ever ease.
The Financial Mechanics
AMAT's gross margin sits around 47-48%, healthy but below ASML's ~51%. The company generates $8-9 billion in operating cash flow annually, with free cash flow of $6-7 billion. ROIC runs 25-30% against a ~10% WACC—genuine value creation.
But here's the hidden tension: the market values AMAT at 25-30x trailing earnings, pricing in AI-driven growth. What's not fully priced is the China ceiling. If China revenue stabilizes at 15-18% of total (down from 30%), AMAT's long-term growth rate drops by 1-2 percentage points annually. That's the difference between a 15% EPS CAGR and a 12% one. In a DCF model, that's a 15-20% valuation haircut.
Contrarian: What the Bulls Got Right
Now let me steelman the other side, because the bears aren't entirely correct.
Export controls as a profit filter
Here's an uncomfortable truth: not all revenue is good revenue. China business, particularly in mature nodes, tends to be lower-margin and more price-competitive. Chinese fabs are notorious for squeezing suppliers on cost, and domestic toolmakers are willing to sell at breakeven to gain market share.
By being forced out of China, AMAT is effectively being forced to focus on its highest-margin customers: TSMC, Samsung, Intel, SK Hynix, Micron. These customers pay premium prices for leading-edge equipment and value the service ecosystem AMAT provides. The revenue mix shift could actually improve gross margins by 100-200 basis points over time.
The AI supercycle is bigger than China
The AI-driven demand for advanced logic and HBM is not a cyclical blip—it's a structural shift. NVIDIA's next-generation GPUs require more deposition and etch steps per wafer than any prior generation. GAA transistors (3nm and below) need 20-30% more process steps than FinFET. Backside power delivery adds another layer of complexity. CoWoS advanced packaging is the bottleneck for AI accelerators, and AMAT is a key supplier.
This demand is concentrated in Taiwan, South Korea, the U.S., and Europe—all accessible markets. The AI supercycle could add $3-5 billion in annual revenue to AMAT over the next 3-5 years, more than offsetting China losses.
The "China premium" in valuation
Some investors argue that AMAT deserves a higher multiple precisely because it's less exposed to China. The logic: China exposure is a political risk, and removing it makes earnings more predictable. A company with zero China revenue is easier to model, easier to underwrite, and less likely to face sudden regulatory shocks. This "de-risking premium" could support a higher PE ratio.
The service revenue moat
AMAT's installed base outside China is massive and growing. Service revenue (which carries 60%+ gross margins) now represents a significant chunk of total revenue. As the installed base expands with new fabs in the U.S., Europe, and Japan, the recurring revenue stream compounds. This is a subscription-like model hiding inside a hardware company—and it's largely insulated from export controls.
Takeaway
History is a Merkle tree, not a narrative. The branches are verifiable: China revenue declining, service contracts shrinking, compliance costs rising. But the root—the fundamental question—is whether AMAT can grow into a post-China world fast enough to offset the structural loss.
The answer is probably yes, but the path is narrower than the market believes. AMAT's growth now depends on three variables: the pace of AI-driven fab expansion in the free world, the speed of Chinese domestic toolmaker progress, and the political stability of the U.S.-led alliance system. Any of these could break.
Silence is the loudest bug report. And the silence from AMAT's China segment is deafening.
The company will survive. It will even thrive. But the Applied Materials of 2030 will be a different company than the one that rode China's semiconductor boom from 2015 to 2022. The question isn't whether AMAT can adapt—it's whether the global semiconductor industry can afford the cost of this decoupling.
Precision is the only apology the truth accepts. The truth here is that export controls have created a permanent structural break in the global semiconductor equipment market. AMAT is learning to live with it. China is learning to live without it. And the rest of us are left to trace the bleed through the gateway, watching history verify itself one transaction at a time.
The next signal to watch: AMAT's quarterly China revenue percentage. If it drops below 15%, the market will finally start pricing in the new reality. If it stabilizes in the high-teens, the bulls might be right that the worst is over. Either way, the code didn't fail. The policy did. And the ledger will remember.