KLA Investment Thesis

Published on 03/13/21 | Saurav Sen | 3,448 Words

The BuyGist:

  • This is our thesis as posted on January 21, 2020 - now in magazine format.
  • The Thesis Summary (somewhat redacted) is available to all readers. 
  • Thesis details - including 1) BUY/HOLD/IGNORE conclusions 2) whether we still hold the stock, and 3) Valuations - are for members only, accessible upon login. 
  • Please see our subscriptions plans for full access, if you haven't already.

Thesis Summary

As of January 21, 2020: KLA is the biggest Process Control company in semiconductor mfg., with a >50% market share. In the next 3-5 years, its market dominance should increase with the proliferation of EUV & other processes  that promise to extend Moore’s Law.

Competitive Advantage - The Castle: 

  • Core Competency? Quality control in semiconductor manufacturing.
  • Product Differentiation? Strong. KLA market share is above 50%. Focus is on EUV.
  • Growth Drivers? New production methods: EUV in Logic, 3D stacking in Memory.

Durability of Competitive Advantage - The Moat:

  • Competition/Threats? Competition is weak, but high customer concentration.
  • Moat? Entrenched nature of long life-cycle products. Services business growing.
  • Cash Flow Volatility? Volatile. Capex needs and margins are cyclical.

Management Qau

  • Strategy? Good. Focused on EUV process control and Services.
  • Return on Equity? High. Above 50%. With high margins and manageable debt.
  • Sustainable Free Cash Flow? *****Redacted.

Competitive Advantage: The Castle

Core Competency? Quality control in semiconductor manufacturing.

KLA is the big 800-pound gorilla in semiconductor process control. What does that mean? Process Control in this world means:

  1. Spotting defects
  2. Yield Management

Semiconductor manufacturing is a precision sport. It involves “drawing” lines with electrons and light on tiny surfaces, measuring a few nanometers (1 nanometer = 1 billionth of a meter). Companies that make (fab) semiconductor chips are always under pressure to churn out defect-free wafers so they can sell more. That’s what KLA’s machines help them do. KLA’s clients – like Intel, Samsung, Micron and TSMC – use KLA’s machines to spot defects and make sure that production goals are met.

KLA has confusing ways of slicing and dicing their businesses. They report 2 segments – Products and Services – which doesn’t tell us very much. In their Analyst Day, however, they described the Products segment in terms of 3 main categories:

  1. Inspection
  2. Metrology
  3. Control Software

I like these categories better.

Inspection is what it sounds like – KLA’s machines that scan Masks and Wafers to spot errors. Semiconductor manufacturing is microscopic. A blueprint of sorts is drawn on something called a Photomask. This is the chip design of the integrated circuits (ICs). This design must be inspected before it’s put into actual production on something called Wafers. And then obviously, the Wafers must go through their own quality control process to make sure they are absolutely flawless. In these microscopic environments, any little discrepancy ruins the production process. Errors may come from dust, misalignments in the IC “fabrication”, materials failures…the list goes on. KLA machines are in demand because they help spot these microscopic defects when a semiconductor fab executes IC designs.

Metrology is a smaller part of their business. This refers to KLA machines that spot defects in the physical architecture of the chips. In their 10K, KLA gives us some examples: pattern dimensions, film thicknesses, layer-to-layer alignment, pattern placement, surface topography etc. The way I understand it is that Inspection is about checking the way the transistors are connected – the little, thin lines that conduct electricity. Metrology is about checking how chips are constructed and stacked. KLA’s Metrology tools are used heavily in the Memory industry now, where memory chips are being stacked on top of one another to, well, pack in more memory in a device. We went over this in some detail a while ago in “Investing in AI Hardware”

There is a third, albeit smaller, category which I’m going to lump into their Inspection segment – Reticle Manufacturing. Reticles are like a “first cut” in wafer production. They are renditions of incomplete IC designs that then get etched onto a Photomask, which further gets etched onto a whole barrage of wafers. Reticles are basically like a bigger Photomask but with just a part of the complete design which must be “projected” onto the final wafer (the actual product) using some refracted light. KLA also engages in reticle production, which is slightly separate from “inspection”. But the point of a reticle also is, ultimately, quality control in the final wafer production. 

The Services business is also what it sounds like. KLA machines are complicated and must be maintained through their 3-5-year lifecycles. This makes the Services business quite lucrative – the more KLA machines permeate the semiconductor landscape, the more lucrative the Services business gets.

Overall, KLA’s core competency is spotting microscopic defects. They’re the biggest in what they do. 

Product Differentiation? Strong. KLA market share is above 50%. Focus is on EUV.

KLA’s market share is an impressive 50% (or more). As far as I know, they’re the only pure-play Semiconductor Process Control firm. Their competitors usually have other base businesses. More on that in the Competition/Threats section below. But for now, the point is that their impressive Market Share is evidence that they sell products customers want.

To a layman like me, their product differentiation has two vectors:

  1. They do only Process Control. They don’t manufacture other machines.
  2. They are investing heavily in EUV (Extreme Ultraviolet Lithography) process control machines.

#1 s about conflict of interest – KLA doesn’t have one. Common sense suggests that a customer like Intel or TSMC would prefer to buy third-party inspection tools rather than buying them from the maker of the tools that need inspection. The likes of ASML and Applied Materials sell both fabrication and inspection tools.

#2 is KLA’s big bet. And frankly, it’s ours too. We were upfront about it in “Investing in AI 2020” that we missed out on the ASML rally. ASML, as we had identified, is a supremely comfortable company – a true global dominator. They’re the only company in the world that make EUV machines – a machine that is at the heart of extending Moore’s Law. For more details, please read our ASML thesis.

But EUV is a new process. It’s not as tried and tested as older processes. But for chips to get smaller and become more powerful, EUV is the only viable solution on the horizon. The fact that it needs to tested and re-tested repeatedly for defect and yield is good news for KLA. The company has made special tools to inspect EUV production. And the best part is that KLA’s EUV inspection tools are selling well. It’s reasonable to bet that if EUV becomes mainstream – which we expect it will – KLA is going to be the inspection tool of choice at most fabs.

In an EUV-dominated world of say, 2022, I expect KLA to have a dominant market share.

Growth Drivers? New production methods: EUV in Logic, 3D stacking in Memory.

In the worldview we published last week – Investing in AI & Big Data 2020 – our basic thesis was that Process Control has never been more important than it is now. That’s because chips are getting smaller and they’re testing the limits of Moore’s Law. This comes at a bad time – just as the volume of data in our civilization is growing exponentially. We discussed this in some detail in the worldview “The Grand Inflection Point”.  In that worldview, we had pointed out that the semiconductor industry needs to resort to novel solutions to beat this problem. EUV and 3D-stacking are 2 of those solutions.

We already talked about EUV in the previous section. As EUV permeates, KLA’s new products will be more in demand. The same goes for 3D-stacking. It’s another way to extend Moore’s Law. It involves stacking chips on top of each other to increase processing power, rather than packing in more transistors on a single surface. The latter is becoming harder to do as the world moves from 7nm to 5nm. Stacking them on top of each other is a workaround. But some consumer products designs may not allow stacking. After all, it thickens the end-product. 3D-stacking, from what I’ve seen, has mainly permeated Memory chipmakers rather than the Logic guys.

The point is that novel solutions are never smooth. They give rise to errors and defects. And that’s KLA’s cue. The lollapalooza moment will arrive when both Memory and Logic chipmakers need to apply BOTH EUV and Stacking. That’ll be fun – for KLA.

The word on the street is that KLA should see a good 2020 and beyond. What does that mean? Well, one of its main clients – TSMC – is one of our top holdings. And in their latest earnings call a few days ago, their management said 2 things that bode well for KLA:

  1. They are going all-in into EUV.
  2. They are increasing their Capital Expenditure.

TSMC makes chips for their clients – like AMD, Nvidia, Qualcomm, Apple etc. Obviously, TSMC is seeing increased orders from these clients, which is why they’re investing more into R&D and Production. If KLA is the big gun in the Process Control arena, which they are, then the stars are aligning for them.

Overall, KLA’s revenue and cash flow growth is dependent on Wafer Frontend Equipment (WFE) growth. And the buyers of WFE are companies like TSMC, which is the largest fab in the world right now.

Durability of Competitive Advantage: The Moat

Competition/Threats? Competition is weak, but high customer concentration.

Let me list some of KLA’s main competitors:

  1. Applied Materials
  2. Hitachi
  3. Lam Research
  4. ASML

Each of these competitors has a base-business that’s different from Process Control. That’s good for KLA and that’s why they have 50% market share. I don’t think competition is the main threat. But customer concentration is.

KLA’s main customers are semiconductor fabs, both in memory and logic. After perusing their 10Ks, it’s clear that their main customers are (and have been):

  1. TSMC
  2. Samsung
  3. Micron

Maybe, we can add a couple of other fab names (like Intel) and they’ll account for almost all of KLA’s sales. This is a risk. But the mitigation is this: TSMC, Samsung and Micron are going nowhere. They are playing in some of the biggest megatrends – AI, 5G, IoT, Autonomous Vehicles – and they are winning. If they need EUV and Stacking, then so be it. KLA is likely to remain their choice for Process Control tools.

Another offsetting fact is KLA’s popularity among new fabs in China. If you’ve been keeping up with the Trade War drama, then you know that China now finds its dependence on US technology to be too risky. So, it’s setting up its own semiconductor fabs, mostly in the Memory space. But they still need equipment, which only some specialized companies make. KLA is one of them. In that sense, KLA has a nice in-built trade war hedge as well.

Moat? Entrenched nature of long life-cycle products. Services business growing.

KLA’s Moat comes from 3 factors:

  1. Its products have a long lifecycle and they’re entrenched in an intricate production process.
  2. The company is investing heavily into EUV inspection.
  3. It’s investing heavily in its Services offering.

At the end of the day, KLA’s customers want a reliable set of tools that ensures smooth production – defect-free and high yield. If KLA’s tools keep doing that, EUV or not, along with top-notch Services offerings, it’s hard for me to see KLA get uprooted from its throne.

Cash Flow Volatility? Volatile. Capex needs and margins are cyclical.

The fact that KLA is investing heavily in EUV inspection and other new products such as E-beam (where it’s a late entrant). As we approach the death-end of Moore’s Law, things are getting more complicated. Chips are getting smaller but are to be made with different processes. EUV is still in its nascent stages. TSMC, Intel, Samsung and the other fabs are still figuring it out. The point is that the capex needs of all of them and their equipment providers (like KLA) is still up in the air. As of now, all these companies have committed to increased capital expenditures. KLA is no exception.

Cash Flow volatility won’t come from competition because KLA has a wide Moat, as discussed above. It will come from the natural teething process from new processes, which are necessary due to the confluence of these factors:

  1. Data compounds at an even faster rate.
  2. Moore’s Law is facing its natural end.
  3. 5G takes off.
  4. IoT is now a real thing.
  5. AI makes some giant leaps.
  6. Data compounds even faster.
  7. EUV is now mainstream.
  8. Processing Power increases.
  9. AI makes some bigger leaps.

This list can go on and on. As that happens, KLA and its customers will need to figure it out as they go. Right now, EUV and Stacking seem to solve the problem over the next couple of years. So, they’re all spending money on those. But in 2023, it could be some other novel solution. KLA would need to provide the appropriate inspection for that new process. That requires capital.

Management Quality: The Generals

Strategy? Good. Focused on EUV process control and Services.

Management is focused on the right things:

  1. Hold on to market share by…
  2. Investing in EUV inspection tools.
  3. Investing in Services.

As I mentioned earlier, those are 2 big Moat wideners. And its Management’s job to widen the Moat.

KLA recently held an Investor Day, and that’s my main source of information on Management. 2 things stood out to me:

  1. The top Management team is mostly home-grown. Most of them have worked at KLA for more than a decade.
  2. They’ve consolidated all their product under one roof: Process Control.

#2 is interesting because KLA has recently made an acquisition called Orbatech, which makes inspection tools for things like LED screens and something called MEMs. The latter stands for Micro Electro-mechanical systems. They combine electrical and mechanical processes – like the screen of your smartphone. I can see the “synergy” here, but LEDs and MEMs are a far cry from semiconductors. But all of this is under the Process Control division. Why? Because, the end-markets are so diverse now. Management is thinking ahead to other use cases of high-power computing like Autonomous Vehicles and Augmented/Virtual Reality.

Acquisitions, however, come with a downside. They cost money. And money can come from internal cash flow or from a fresh round of capital-raising. KLA’s management have taken the latter route. And that’s a negative because it means either:

  1. Issuing more debt
  2. Issuing more equity

Either way, it hurts the valuation of the company unless the acquisition pays for itself. That takes years. And most acquisitions don’t pay off. But in this case, KLA’s debt levels are manageable. And the company buys back shares to counter equity dilution due to acquisitions.

Overall, Management has done a good job of investing in the right areas. Personally, I hope they take it easy with the acquisitions.

Return on Equity? High. Above 50%. With high margins and manageable debt.

ROE is the best easy report card we have with which we can judge Management Quality. On this measure, KLA Management get full points. KLA’s ROE is about 54%, which is a fantastic number. This is mostly driven by margins. KLA’s debt level is not low but it’s manageable with a Long Term Debt to Equity ratio of 1.1.

KLA’s CEO mentioned during their earnings call that they use Gross Margin as their measure of competitiveness. It indicates to them both pricing power and product demand. It makes sense. Based on this measure, it appears KLA has both pricing power and demand. This margin obviously flows down to the Free Cash Flow level if costs are well controlled. KLA has a Free Cash Flow margin of around 25%. This is among the highest I’ve seen in any industry.

Sustainable Free Cash Flow? About $1.8 billion. Translates to $227 per share.

The assumptions behind this $1.8 billion free cash flow number are as follows:

  1. Assumed Revenue increases to $6 billion in 2023 (from $4.3 billion over last 12 months).
  2. Assume EBITDA margin improves slightly to 40% from 39%.
  3. Assume Capital Expenditures same as last 12 months.
  4. Assume no benefit from working capital swings.
  5. Assume no additional debt.
  6. Assume similar tax rate.

The big assumption here is #1: Revenue. It’s quite simple. Management expect revenue to increase to the range of $7 to $7.5 billion by 2023. Revenue is at $4.3 billion now, so this assumption is quite aggressive. Management cites the inevitable growth in EUV, a comeback in Memory and an increase in process intensity as growth drivers. It’s a believable story. It’s rare that we come across a company with >50% market share in an industry where growth is almost inevitable. But I took the mid-point in my assumptions – between $4.3 and $7.5 billion – and rounded it up. I don’t know if KLA will get to $7.5 billion in 2023, but the path to $6 billion looks utterly achievable.

The rest of the assumptions are straightforward. Of course, I expect more acquisitions. But I don’t expect Management to dilute shareholders anymore. I expect them to use buybacks to counter that.

Overall, the KLA story is a solid one. The big underlying assumption is the proliferation of EUV over the next few years. Will that happen for sure? I can’t say. But the facts are:

  1. There is no other technology in sight that can extend Moore’s Law.
  2. The biggest fab in the world – TSMC – is all in. They’re investing heavily in EUV.

Given all this, there seems to be some Margin of Safety in KLA’s stock price. It’s a comfortable company at a comfortable price.

We use cookies on this site to ensure the best service possible.