What this covers
This Business Breakdowns conversation pairs Zack with Dylan Patel to walk through Applied Materials, starting with how a chip is actually built: thousands of sequential steps across hundreds of tool types, so defect rates have to be far better than Six Sigma. It then places the company among its customers. TSMC, Intel and Samsung drive much of equipment demand, and TSMC alone spent over $30 billion on capex last year, about 80 percent of it on equipment. Intel's failure to get 10-nanometer to high yield is set against TSMC's co-engineering with toolmakers.
The middle of the discussion covers the company's structure. Unlike ASML, which is mostly lithography, Applied runs six or seven franchises, many only 3 to 5 percent of fab spending. CMP is the worked example: under 3 percent of fab cost, effectively a monopoly at the leading edge, and still not worth TSMC's risk of replacing, given the roughly 20 experts who would have to requalify a new tool. The financial picture follows: about 45 percent equipment gross margin, roughly a quarter of the business in services, about 80 percent of operating cash flow returned, and buybacks of 5 to 7 percent of the float in downturns. Revenue still fell about 35 percent peak to trough in 2008. Forward-looking points include gate-all-around transistors creating new process steps and the blocked Tokyo Electron merger.
The contested turns come last. Dylan argues that government subsidies, including the CHIPS Act's $52 billion, dampen cyclicality by making Applied a picks-and-shovels supplier. He also names China's domestic toolmakers, Naura and ACMR, as the biggest risk as they win share in cleans.
Applied Materials' advantage comes from owning narrow, hard-to-replace process-step franchises embedded in TSMC's yield-driven manufacturing, which sustains margins and lets it buy back stock through cycles; its main risk is subsidized Chinese competitors willing to sell at low margins.
- Each process step has few qualified suppliers and few qualified engineers, so switching suppliers is costly even when the tool is cheap
- Services, R&D through downturns, and counter-cyclical buybacks dampen the semiconductor cycle's effect on earnings
- Geopolitical subsidies flow to equipment makers and reduce cyclicality, while China's domestic equipment push is the main threat
Co-engineering with leading customers determines yield, progress, and durable partner advantage.
- Chipmaking involves thousands of sequential steps across hundreds of equipment types, so defect rates must be far better than Six Sigma, because a Six Sigma defect rate applied across tens of thousands of steps would make functional chips impossible.
“Six Sigma on tens of thousands of steps would actually kill you, right, in semiconductor manufacturing.”
- Intel's failure at advanced nodes is attributable in part to hubris, as it believed it needed less help from equipment companies than TSMC did.
“Intel got a bunch of hubris and thought they didn't need the equipment companies' help as much.”
- Yield is the lifeblood of chip manufacturing, and Intel's failure at 10-nanometer was that it could make chips but could not get them to high yield, while TSMC progressed through 14, 10 and 7 nanometer by co-engineering with equipment makers.
“Yes, they were able to make some chips with 10-nanometer, but they were never able to yield it, right?”
- Shifts to gate-all-around and nanowire transistor shapes at 3, 2 and 1.4 nanometers create brand-new process steps that Applied Materials can compete to enter and co-engineer.
“there's meaningful changes on changing the shape of the transistor, right? It's called gate oxide all around, it's called nanowire”
- For any individual process step, only about 50 to 100 engineers in the world have deep expertise, split among Intel, Samsung, TSMC and equipment makers, making the field highly concentrated in talent as well as suppliers.
“In any individual process step, there's only 50 or 100 engineers in the whole world who know such a level.”
- Equipment companies see technology transitions before the rest of the industry because their customers place orders for the next era years in advance, which is why Applied's investor days discussed the PC, mobile and AI eras early.
“they see things earlier than the rest of the industry, in fact.”
- Being a long-term partner that keeps investing when business is weak lets Applied Materials feed R&D learning back to less concentrated customers such as Texas Instruments, NXP and onsemi, not just the leading fabs.
“Now we can do better and better there as well. We can help them in so many ways because we have access to this.”
- Applied Materials kept investing in R&D through downturns, raising it to 20 percent or more of revenue during drawdowns, which builds long-term partner status that pays off when the cycle recovers.
“in these drawdowns, hey, it shot up to 20% plus. They took that long-term view.”
- Concentration exists on both the buyer and supplier side, because for a specific tool with a specific chemical reaction and nanometer precision there are often only one or two qualified companies, so customer concentration is also where the business gets its strength.
“when you narrow down to exactly the chemical reaction you're trying to induce at what nanometer precision and all that sort of stuff, you actually end up with, there's two, maybe one, right, companies in this field.”
- Intel, Samsung and TSMC each run one research fab where equipment makers such as Applied Materials, ASML and competitors embed their latest tools years before production, so the next generation of chips is designed and proven at high yields with suppliers.
“all of these companies, Intel, Samsung and TSMC have one research fab, right, which is where they conduct all of their research and development.”
- Co-engineering learned at concentrated leading-edge customers feeds back into the service of less concentrated customers such as Texas Instruments, NXP and onsemi, and smaller customers depend on Applied's services because they lack the in-house technicians that TSMC has.
“what about the smaller customers? They don't have that capability, but because you're offering it to them, you're helping them be a better business.”
Applied's broad franchise portfolio makes it a leader among a few giant equipment makers.
- Applied Materials is the largest semiconductor manufacturing equipment company in the world, selling tools to TSMC, Samsung and Intel and also servicing and supporting R&D for those tools.
“So Applied Materials is the largest equipment company in the world for manufacturing semiconductors.”
- About eight or nine semiconductor equipment companies sell over $1 billion of equipment a year, and the largest players sell over $20 billion a year, including Applied Materials, ASML, Lam Research and KLA.
“There's about eight or nine companies that sell over $1 billion worth of equipment a year”
- Unlike ASML, which is mainly lithography, Applied Materials spans six or seven equipment franchises, and lithography is only about 20% of fab equipment spend while many Applied niches are 3-5% each.
“With Applied Materials, you've got arguably six or seven franchises, right? Lithography is like 20% of a fab's spend”
- Chemical mechanical planarization (CMP) polishes wafers to atomic-scale flatness between process steps; Applied Materials has effectively a monopoly on CMP tools for the most advanced chips, though CMP is under 3% of total fab equipment cost.
“this is an area that Applied Materials is effectively a monopoly, especially on the most advanced chips, and they're the only one making the tools for it.”
- Market share should be measured within narrow process segments, because aggregate company revenue mixes many segments where competitors do not overlap, so ASML and Applied Materials barely compete directly.
“ASML and Applied Materials do not compete with each other at all. There's a few small niches, but for the most part, 90% plus of the business does not compete”
- Applied Materials' new pattern engineering tool aims to compete with lithography in some process steps, and success could take billions of dollars of revenue from ASML.
“they released a new tool. They call it pattern engineering.”
- Applied Materials expects to keep over 70 percent share in verticals such as epitaxial growth, CMP, ion implant, RTP, annealing and PVD.
“So these are some of the areas where Applied Materials is going to remain 70%-plus share.”
- A long period of mergers and acquisitions built Applied Materials' portfolio of equipment franchises, and this acquisition-led breadth is why it has many segments where it leads rather than one product line.
“there was a long period of M&A, right, where they acquired companies, they merged with companies and Applied Materials slowly built their portfolio”
- The Japanese CMP competitor sells into less advanced nodes, while Applied Materials is the only supplier selling CMP tools for 5-nanometer production.
“When I look at a CMP tool for 5-nanometer, actually, Applied Materials is the only one really selling into it.”
- Even in more competitive areas such as etch and deposition, customers use Lam Research's etch tool for some chemistries and Applied's tools for others, so competition is specialized by step rather than head-to-head.
“Lam Research, which is another competitor, their etch tool is better at this kind of etch with this kind of chemistry”
High margins, cash returns, and buybacks let Applied recover from downturns.
- Equipment sales carry roughly a 45 percent gross margin for Applied Materials, and about a quarter of the business is services on already-sold equipment.
“that is done generally by Applied Materials in 45%-ish gross margin range”
- Applied Materials operating margins are in the 20 percent range, reaching about 30 percent in 2021 and 2022, on revenue above $20 billion a year.
“So you end up with a company where their operating margins are in the 20% range”
- Applied Materials spends $2 billion to $3 billion a year on R&D, roughly 13 to 14 percent of revenue, with the spend expected to exceed $3 billion this year.
“you're looking at that being like 13%, 14% of their revenue is R&D.”
- TSMC earns about a 50 to 60 percent gross margin and Applied Materials and ASML each earn about 50 percent, and the suppliers accept these prices because of mutual respect rather than racing to the bottom.
“TSMC makes about a 50% to 60% gross margin. And Applied Materials makes about a 50% gross margin. And ASML makes about a 50% gross margin.”
- Applied Materials returns roughly 80 percent of operating cash flow to shareholders, since capex is under 15 percent of spending and R&D is the other major use.
“something like 80% of their cash flow that's generated from operating activities is actually just turned around and returned to shareholders”
- After a long period of acquisitions that built its portfolio, Applied Materials was blocked from merging with Tokyo Electron and then shifted from M&A toward organic growth, focusing on new equipment for future process nodes.
“It culminated in this acquisition of Tokyo Electron, which is Japan's biggest equipment company”
- Applied Materials has not had a two-year period of no growth since the dot-com bubble, meaning single-year declines were offset over two years.
“there hasn't been a time period at least since the dot-com bubble where, for a two-year period, they didn't grow.”
- In downturns Applied Materials buys back about 5 to 7 percent of its float, which permanently lifts EPS once the cycle recovers, as it did after the 2008 and 2017-2018 periods and the 2022 drop.
“they end up buying back like 5%, 6%, 7% of float, right, in that downturn.”
Government funding and subsidies drive new chip capacity and cushion Applied's cyclicality.
- COVID-19 lockdowns in 2020 broke just-in-time manufacturing in chips, and the resulting auto chip shortage woke up consumers and governments to how concentrated and fragile chip supply was.
“But with COVID-19 and the lockdowns in 2020, everything kind of changed”
- A Japanese chemical plant explosion in the late 1990s shut down production of about 80% of the industry's relevant chemical supply, showing how concentrated the supply base is.
“in the late '90s, there was a Japanese chemical plant that blew up, and it shut down production of like 80% of the industry”
- Every iPhone relies on TSMC in Taiwan, so the chip supply chain is geographically concentrated and politically vulnerable because of Taiwan's relationship with China.
“every single iPhone relies on Taiwan Semiconductor”
- In many chip segments with only two or three players, adding a fourth competitor requires nation-state-level subsidies, which is why governments are now funding new capacity.
“And for there to be a fourth player is it requires nation state-level subsidies, which is exactly what's happening”
- Applied Materials' revenue has fallen about 35 percent peak to trough in severe cycles such as 2008 and about 20 percent in smaller cycles such as 2018, so the business remains highly cyclical.
“peak to trough of business, they still draw down like 35% on revenue. If you look at the 2008 crisis, right, they drew down 35%.”
- Government subsidies reduce Applied Materials' cyclicality because fabs will spend subsidy dollars regardless of weak orders, so the company does not need to cut capacity spending in downturns.
“instead of cutting back because orders are weak, like we're going to cut back because orders are weak, but oh, hey, we have all these subsidy dollars”
- Geopolitical tension, which many view as a risk, is actually tremendously positive for Applied Materials because state money for chip capacity flows to picks-and-shovels suppliers.
“That's sort of the viewpoint you could take with a lot of the geopolitical tension, is it's actually tremendously positive for Applied Materials”
Industry growth and continued execution drive Applied's long-run revenue growth and share.
- Applied Materials has grown revenue above 10 percent annually for over a decade (about 15 percent over five years), with EPS growing near 20 percent on a sustainable basis helped by a roughly 30 percent share count reduction over eight years.
“They've taken down something like 30% of their shares in the last eight years or so.”
- Because the semiconductor industry grows faster than GDP, equipment grows faster than chips, and Applied Materials grows faster than equipment, the company's long-run growth depends mainly on continued execution.
“the equipment industry is consistently growing above semiconductor. And then Applied Materials is consistently growing above that.”
- A useful way to assess market share is to remove the services business, which is 25 to 30 percent of revenue, and look at equipment revenue growth relative to the industry.
“remove out the services business because the services business, while it is 25%, 30% of the company's revenue, it's not the CapEx”
- Applied Materials has held about 20 percent of total equipment spend for a long time, and ASML's EUV shift moved share only from roughly 18 to 22 percent, which is small relative to total spend.
“Applied Materials has been 20%, and they remain about 20%.”
- Moving from 5-nanometer to 3-nanometer adds about 30 percent more process steps, and AI model complexity is why customers still want shrinks despite weakness in mobile and PC demand.
“because it's like 30% more process steps when you go from 5 to 3 nanometers, right?”
Concentrated fab spending creates equipment demand that customers are reluctant to switch away from.
- Major fabs such as TSMC, Intel, Samsung, Micron and SK hynix spent about $100 billion on equipment last year and are expected to spend somewhat less, around $90 billion, next year.
“spent about $100 billion, a little bit less last year.”
- Even though a second-tier CMP supplier exists, TSMC would not replace Applied Materials because the 3 percent cost share is small compared to the opportunity cost of the roughly 20 experts who would have to requalify the tools across successive nodes.
“I'm going to spend those 20 engineers who are super focused on CMP”
- TSMC spent over $30 billion on capex last year, about 80 percent of it on manufacturing equipment, and Intel and Samsung spend similar amounts, making three customers responsible for much of equipment demand.
“TSMC spent $30 billion plus last year on CapEx, right? And of that, 80% of it was equipment related to manufacturing semiconductors.”
- The CHIPS Act provides $52 billion plus a tax credit of about $30 billion targeted at equipment purchases, and Japan, Europe and China are also spending tens to hundreds of billions, which Dylan says flows through to Applied Materials as the picks-and-shovels supplier.
“the U.S. has $52 billion in the CHIPS Act plus a tax credit, which is estimated to be something like $30 billion. And that tax credit is specifically on equipment purchases”
Services raise customer tool output and margins while cushioning revenue during downturns.
- Services should be viewed as helping customers extract maximum output from equipment rather than as a subscription, and this helps smaller customers without in-house servicing capacity.
“let's make sure you get every single chip you can out of that.”
- About one third of Applied Materials' business during downturns comes from services, because equipment owners must keep servicing tools or lose the value of their purchase, and service margins are higher than equipment margins.
“1/3 of their business on these drawdowns is going to be the services business because if you purchased equipment, you still need to service it.”
- Applied Materials' services model aims to maximize each tool's output for customers, which makes customers more profitable and more willing to buy additional equipment, rather than simply selling tools once.
“let's make sure you get every single chip you can out of that. ... that's going to make you more profitable, and that's just going to make you want to invest more in our equipment.”
Subsidized Chinese competitors threaten Applied's share and the oligopoly's pricing power.
- The bear case is a Chinese competitor that is subsidized and willing to sell at low prices and thin margins to gain share before profitability, which would break the oligopoly's pricing power that has never been challenged by a new entrant in decades.
“what happens if there's a competitor who sort of is, let's say, irrational by the Western sense of competition”
- The biggest risk to Applied Materials is China, where domestic equipment makers such as Naura and ACMR are gaining share in segments like cleans, and China builds 25 to 30 percent of new fabs, so switching in those segments can erode revenue.
“there's numerous Chinese equipment companies, Naura and ACMR”
Specialists splinter the chip industry into niches, competing without owning everything.
- Specialization works because each layer of the chip industry has abstraction boundaries, so a niche firm can compete without being a conglomerate that does everything, and EDA software firms can share R&D across many customers.
“because there's these abstraction layers between every single layer.”
- The semiconductor industry shifted from companies like IBM and Intel doing everything in-house to specialists, starting with equipment makers selling tools to chipmakers, and Intel's decision to sell chips rather than systems was an early step in that disaggregation.
“Intel, sort of one of the big first disaggregators of the industry by saying, hey, we're going to sell chips. We're not going to sell systems.”
- TSMC led the shift to being a pure foundry that only manufactures chips and does not design them, which let companies like NVIDIA operate as asset-light fabless designers and let Apple design its own chips and outsource fabrication.
“this is what allowed companies like NVIDIA, who started as a completely fabless company”
- Chip design data was once handed from designers to fabs on physical tape, a handoff still called tape-in, and as design went digital third-party EDA software firms emerged to serve many chip designers.
“It's called tape in, which is when you send the tape from the design guys to the fab guys.”
- As chip design became computerized, third-party electronic design automation companies emerged that do not make chips but make the design software, gaining efficiencies from shared R&D and a larger customer base across many chip designers.
“these EDA companies, electronic design assistance, these companies don't know how to make a chip. They don't know how to design one, but they know how to make the software for it.”
- Small specialized players are more nimble, more able to innovate and more focused than massive conglomerates, and this age-old pattern is why the semiconductor industry keeps splintering into specialized niches.
“The thing is about almost any industry as a small player is always going to be more nimble, more able to innovate, more able to focus on their one specific area rather than be a massive conglomerate”