The AI compute arms race is in full swing, and most investors are focused on the “on-stage stars” like Nvidia and AMD. But to understand the entire compute supply chain, there is one link you absolutely cannot skip—semiconductor equipment. As the saying goes, “the ones selling shovels outearn the ones panning for gold”: the more chips are made and the more advanced the process, the thicker the order books held by equipment makers. Today, in plain language, we will take the US-listed semiconductor equipment “shovel seller” track apart piece by piece, looking at the players in lithography, etching, metrology, and testing, where demand comes from, and where the risks lie.
Why Is Equipment the “Lifeline” of AI Compute?
At its core, a chip stacks tens of billions of transistors onto a silicon wafer the size of a fingernail, and it is semiconductor equipment that “draws,” “etches,” and “deposits” those circuits. Simply put, no equipment means no chips—and no AI compute to speak of.
This AI wave is pulling equipment demand along on “two wheels.” On one side is advanced processing: as more advanced nodes such as 2nm and A16 ramp into mass production, the more transistors shrink, the higher the precision required from lithography, etching, and metrology tools, and the higher the investment per fab climbs. On the other side is advanced packaging and HBM memory: AI chips must “team up” with stacked HBM through 2.5D/3D packaging processes such as CoWoS to unleash their compute, triggering an explosion in demand for bonding, thinning, and inspection equipment. For equipment makers, every step forward in process technology and every layer of added packaging complexity means orders worth real money.
Even more crucially, equipment orders are often a leading indicator of the whole supply chain's health. Fabs usually place equipment orders a year or more before chip demand explodes, so understanding equipment makers' backlogs and lead times can sense the direction of compute capacity expansion earlier than staring at chip companies' earnings reports.
A Roundup of Core US-Listed Semiconductor Equipment Players
Lithography: ASML Stands Alone
Dutch lithography giant ASML (US-listed ADR, ticker ASML) is the world's sole supplier of EUV lithography machines—TSMC, Samsung, and Intel all rely on its machines for advanced processes. The next-generation High-NA EUV is entering mass production, with a higher price per unit and an even taller technical barrier, making this moat visibly deep.
Etching and Film Deposition: The Lam–Applied Materials Duo
Lam Research (LRCX) is a top-tier player in etching and film deposition: rising 3D NAND layer counts and TSV (through-silicon via) processing for HBM both depend heavily on etch tools. Applied Materials (AMAT) is the world's largest semiconductor equipment maker by revenue, with a product lineup covering deposition, etching, ion implantation, metrology, and nearly every other step—a classic “one-stop shovel seller” and often seen as a bellwether for industry conditions.
Process Control: KLA, the Goalkeeper of Yield
KLA (KLAC) specializes in inspection and metrology equipment, and chip yields depend to a large extent on the “sharp eyes” of its machines. The more advanced the process, the bigger the share of value captured by inspection—one of the underlying reasons KLA's profitability has kept strengthening in recent years.
Testing and Packaging: Blossoming on Multiple Fronts
Teradyne (TER) is the US-listed leader in test equipment: the more compute an AI chip packs and the longer it takes to test, the hotter the demand for testers. FormFactor (FORM) is deeply rooted in probe cards, while Onto Innovation (ONTO) holds a strong position in advanced packaging inspection. In addition, Japanese equipment makers such as Tokyo Electron, DISCO, and Advantest are not listed on US exchanges, but they remain names the supply chain cannot bypass and are worth adding to a watchlist.
Three Demand Engines: How Far Can This Upcycle Go?
Equipment demand today is driven jointly by three engines, and any one of them slowing down could affect the pace of the upcycle:
- Advanced-node capacity expansion: TSMC, Samsung, and Intel are investing in fabs intensively for the 2nm era, with new plants coming online one after another in Arizona (US), Kumamoto (Japan), Dresden (Germany), and elsewhere, while local fab-construction subsidy policies further underpin equipment purchases.
- HBM and advanced packaging: AI chips' thirst for HBM capacity and packaging capacity shows no sign of ending anytime soon; new processes such as TCB thermocompression bonding and hybrid bonding are pushing equipment unit prices and demand up together.
- Spillover from cloud capex: Leading cloud giants' annual capital expenditures remain persistently high, and this money ultimately passes down the “chips—fabs—equipment” chain layer by layer, with equipment makers one of the important final destinations.
Behind the Buzz: Four Risks You Must See Clearly
- Strong cyclicality: Equipment stocks are essentially an “amplifier” of capital spending. Once fabs slow their expansion, order pullbacks are often steeper than at chip design companies, and share prices swing harder too.
- Export control uncertainty: Major equipment makers all have some revenue exposure to the Chinese market, and any tightening of trade policy could directly hit earnings—a sword of Damocles hanging over the sector.
- High customer concentration: Advanced-node orders come mainly from a handful of leading fabs, and changes in these big customers' capex pacing are enough to sway the short-term fortunes of the entire industry.
- Valuation and expectations: After the sustained rally in the AI trade, sector valuations are no longer cheap, and the equipment heavyweight names in the Philadelphia Semiconductor Index are especially sensitive to any guidance that falls short of expectations.
How Should Retail Investors Position?
For newcomers tracking equipment stocks, focus on three cards: first, orders and backlog, the “reservoir” of future revenue; second, management's guidance on next-quarter shipments and gross margins; and third, changes in the share of China-region revenue. If you would rather not research individual stocks, you can also use semiconductor ETFs such as SMH and SOXX to hold equipment leaders indirectly and share the industry's dividends in a diversified way.
In terms of execution, avoid chasing highs with a full position in one shot. When the equipment industry's upcycle is running, its upside is astonishing, but drawdowns on the way down are just as fierce; building positions in batches, controlling position size, and holding through the cycle are the true prerequisites for actually making money on this “shovel seller” track.
On the whole, as long as the AI compute race keeps going, chip capacity expansion will not stop, and there will always be eager buyers for the “shovels” sold by semiconductor equipment makers. For investors who cannot make sense of chip design but still want a share of the AI dividend, this relatively higher-certainty segment of the supply chain deserves a place on your watchlist—study it patiently and buy on dips.
