Why Semiconductors | Nickwar.com

Why Semiconductors?

It's not a stretch to say Semiconductors:

are the chief driver of technological innovation today and have been for the last 40 years.
are one of the, if not the most, crucial inputs to economic progress currently.
are the building blocks of everything we call 'modern life'.

As we found out in the 2020-2021 time period, it turns out you can't build the things that humans really need - from cars, to the internet, to microwaves, to credit cards - without semis. The world needs them now. It's hard to imagine a future where the world doesn't need more and more.

If you believe the future will include more electronics - or will be more digital - more connected - use less fossil fuels - use more compute, storage, memory --- then you also believe the world will need increasing quality and quantity of semis.

The technology trends people fantasize about (AI, autonomous cars, AR/VR, crypto, ML, Metaverse, touchless transactions, etc.) is predicated on semiconductor advances. Of course, a bunch of sophisticated software needs to be built on top, but it doesn't happen without continued advances in semis.

Over the last couple hundred years, we've had a number of engines for economic and productivity growth. Early industrial automation, steam power, coal, railroads, modern banking systems, oil, etc. In the late 20th century and the advent of the information age, we have semiconductors. Flowing from semiconductors, came the computer, the PC, the commercial internet, mobile, and now AI.

If you want to dive into semis, this episode of The Knowledge Project with the guys from NZS Capital is a great place to start.

So, with all that being said - what makes the semiconductor space attractive for retail public market investment?

Investing Overview

Let's start with industry returns, compared to other major industries. As of May 1, 2023, trailing 10 year returns - SMH dark green is semis:

And to keep myself honest, trailing 20 year returns:

So, the semiconductor industry is a hit on the performance angle. And covered in the first section, is a hit on broad predictions and secular growth trends. Before we dive in deeper to the investment stuff, let's take a look at the industry.

Industry Overview

Semiconductors are a capital intensive business. One of the main reasons the industry has been highly cyclical in the past. The CAPEX growth in the last 23 years has been substantial, 14% CAGR since 2000. Compared to global GDP growth of 3% over the same time period (world bank).

The performance of publicly traded companies in the industry shows a more precise and compelling growth story.

In particular, look at the growth in profit in the last two tabs and the rapid acceleration in the last few years.

And for the skeptical among us, here is the same data broken down per company - showing the same growth story.

That's robust growth over a long period of time and explosive growth in the last few years. Over the last 10-20 years, the industry has moved from a disjointed, redundant, vertically integrated make up towards a symbiotic, horizontal, connected system. While the early results from this consolidation are clear, getting precise details on the consolidation itself is tricky.

We know the number of foundries (contract manufacturing of the chips) has dramatically reduced, in large part to the foundry horizontal platform pioneer TSMC (Taiwan Semiconductor Manufacturing Company).

Another area of the value chain with clear consolidation is the equipment makers. According to Capital Group, the top 5 players currently control circa 75% of the market share - up from around 40% 10-15 years ago.

Estimates of the entire industry consolidation are looser. Here's Accenture's data on the different company sizes:

More total players and more consolidation - which is a rare combination. In fact, I can't think of another industry like this (in many respects, but this is a salient example).

When I analyze the industry data for publicly traded companies in the space, I see the industry has always been head driven. Although, it's worth noting there are sub-segments within, like the ones mentioned above, not broken out here. That's a manual labelling task I haven't gotten to yet and yes, I've tried using the LLMs.

First looking at the revenue and gross profit - highly correlated. The gross profit is shared with the tail of the industry more than the revenue.

In contrast to gross profit, the net income is more concentrated in the head. This is exacerbated once you get below the line in the P/L, as a fair amount of tail companies (here and across other industries) will be negative some years.

The insight here is the head of the industry takes the lions share of the profits. It suggests scale is an important factor - which goes back to the start of this foray. Consolidation, in this case, appears to be a net positive for investors. At least in the short term.

In the long term, I can see it playing out both ways. Less competition and variety is a potential negative - lowering the growth rate of the entire industry - should dampen returns over time. On the other hand, increased scale, decreased competition - increases reliability and profits, increases the capital inflows - should lead to higher returns.

A last word on this, these businesses across the value chain require A LOT of investment, financial, commercial, and R&D. In my view, the increasing scale of key players, industry reliance on what they're best at, and the capital inflows from more stable, higher profits - will increase innovation for the next few chapters of the industry. The key players will be able to do things smaller companies cannot. In the world of semis, they're pushing the limits of physics and material science forward constantly. Longer investment cycles are necessary to achieve this.

Industry Components

The origins of the industry were entirely vertical companies, like Intel, who designed, fabricated, and sold integrated end market products. Almost the entire supply chain as we know it now, was done in house - with a few exceptions.

Over time, the industry evolved towards a more horizontal approach. Semiconductors are a demanding product category. The pace of innovation and growth of use cases slowly eroded the viability of vertical business models. A unique trait of this horizontal industry is the interconnected and symbiotic nature. Each different area is highly collaborative with the neighboring areas.

Today, semi industry companies fall into 5 generalized buckets:

Digital - Largest segment, highest flying products and companies, very competitive, short product cycles.

These are the 'leading edge' chips frequently making waves.
Digital input -> digital output, rely solely on 0s and 1s.
CPUs, GPUs, baseband processes.
Companies such as Advanced Micro Devices (AMD), NVIDIA (NVDA) are leaders.

Analog - Low price and long catalogs, long product cycles, diversified customers and use cases, low glamour.

These are the boring chips that you rarely hear about.
Analog input -> digital output, process analog information (temp, pressure, color, audio, etc.).
Companies such as Texas Instruments (TXN), Analog Devices (ADI), NXP Semiconductors (NXPI) are leaders.

Memory - Highly concentrated producers, commodity like market, critically important, workhorse products.

Used to store data and programs for use in electronic devices.
Two main types, DRAM and NAND.
Companies such as Samsung (SSNLF), Micron (MU), SK Heinex (HXSCF) are leaders.

Foundries - Highly concentrated, operational excellence and scale businesses, high tech contract manufacturing.

Produce semis (almost all digital chips and lots of the other chips),
Enable the fabless/fablight business model that has flourished in the last 15-20 years.
Companies such as Taiwan Semiconductor Manufacturing Company (TSM), Global Foundries (GFS), Semiconductor Manufacturing International Corporation (SMIC) are leaders.

Equipment and Materials - concentrated and specialized, the main drivers of technological innovation, half science/half businesses.

Produce the capital equipment and materials required to produce semis.
This is where much of the innovation in the industry happens, or doesn't.
Companies such as ASML Holding (ASML), Applied Materials (AMAT), KLA Corporation (KLAC), Lam Research (LRCX) are leaders.

Electronic Design Automation Software - concentrated, necessary for designing digital chips, akin to Adobe/Figma/Canva for marketing graphics.

These products are currently verticalized, tailoring to specific needs (designing chips).
Given the complexity of what they do and the optionality in software business models, there is potential to expand horizontally. Although these companies are mature, so perhaps not.
Only 2 companies operate in this space at scale, as far as I know: Cadence Design Systems (CDN) and Synopsys (SNPS).

Investing - SP500 Comparison

This section is going to get numbers and ratios heavy. These are the ways, at least I, look at businesses and market segments to determine relative attractiveness. I start with some of the things above and then get into the fundamentals of the business(es).

At this point, semiconductors seem to be a very attractive section of the market and economy. Plus, the products they make are objectively awesome. These companies are underwriting the transition from the industrial age to the information age - which has and should continue to make the human experience better for all of us.

First stop on our train is income versus headcount. Before we get to the goods, let me explain the methodology here.

The date is May 4 2023, may the fourth be with you all. For the semiconductor portion, I pulled all actively traded companies in the semiconductor or semiconductor equipment and materials industries. 89 companies in total. For the SP500 portion, I pulled the entire index, 500 companies in total.

This is clearly a red apples to green apples comparison. However, it's an accessible one to do. Additionally, the SP500 companies typically represent some of, if not most of, the most consistently performant companies in the world. And certainly, a majority of the largest and most capitalized businesses in the world. So it's a good benchmark.

Back to headcount efficiency. This is one of my favorite ways to compare companies and business models. In short, how productive are the people investments for a given business. Let's take a look at the last 12 months!

SP500 wins the first round. It's an excusable loss for the semis, as the revenue base for SP500 companies is approximately 1400% higher. Still, not ideal.

For this next section, I'm sticking to medians to save space and it seem more fair, given both of these datasets have very fat heads.

Let's look at financial growth - first at TTM, then at T5-year for different sections of the financials.

This is a clear win for semis, in both the immediate past and trailing 5 years, strong outperformance. Growing faster, generating more cash. Except in my favorite place: FCF. We'll forgive them for that since the rest is so clear cut.

We're going to get weird with time periods here. Not quite a Christopher Nolan level of weird, but still buckle up!

This one is harder to call - still, I feel pretty good about ruling in favor of semis. The median dividend is completely excusable for semis, given the immaturity and small size of the median company compared to the SP500. Across most other per share growth metrics, it's semis.

Notable here is a few big deltas, namely in equity per share and net income per share. This is an attribute we'll see again - the semi companies have done a great job utilizing strong margins to increase shareholder value recently. And if you want dividends, the head of the semi segment has you covered - check the names from the industry overview.

Now let's take a peek at some spending and capital allocation. I'm flipping back to averages here as many of the medians are 0% for these items - therefore not very insightful.

And wooof! The SP500 empire strikes back. Or more aptly put, this the most challenging area for the semi segment. Capital intensive. The viz was dropped for the t5yr view due to the wild numbers over there. Clear win for the SP500.

The insight here, beyond the capital intensity, is these metrics are quite important for the semi companies to manage. Given the recent track record of revenue and profit performance, the massive growth opportunity for the end use cases, and as we'll see later ROIC, it's reasonable to expect the debt capital to be allocated well. Still, it drastically reduces the margin for error in execution.

The next stops on our train here loop back through the different areas of the businesses expressed above. The lens and sometimes metrics are different, but we're looking at pretty much the same stuff.

If you've been holding out for the traditional valuation metrics, saddle up!

Boom! Back come the semis with a decisive win. Growing much faster, higher risk/reward, more profitable - still trading at lower valuations.

What's particularly striking is the delta between the TTM and the T5yr view. The growth rate and profitability acceleration doesn't seem to reflect in the valuations. If you're an optimist like me, that's an inefficient lag - which translates to an opportunity.

And the EV/Sales ratio - still important, is explained by the higher debt levels for the semi companies.

We still have a few more metrics to go.... let's see what happens.

And what did happen here? I'm confidently declaring another win for semis. Although my gosh, I'd like to see FCF performance. Putting my treasured metric aside, the other three are instructive for our purposes on this page.

Let's start with ROIC. I believe Charlie Munger said 'it's hard for a stock price to outpace ROIC over the long term'. No matter who said it, ROIC is a great measure of the underlying performance of a business. As an investor, higher ROA/ROE/ROIC gives me confidence in the company's ability to put my capital to use effectively. Clearly, the semi industry at large has done that in the past.

Capex to revenue. This surprised me, for such high capital intensity businesses with tremendous recent growth, I thought this would be out of whack with a cohort such as the SP500. I'm glad to see it's not.

Debt to assets. Again this surprised me, but not as much. The semi industry is asset heavy/rich, one of the big reasons why it's a hard business to run and an even harder business to disrupt. Although semi companies have been loading up on debt, they manage it well and it's put to use in durable assets that produce over long periods of time.

Investing - Semi Specific Deep Dive

Across the value chain, making semiconductors requires lots of capital, long production batches, deep R&D investment, and long lead times. Mix that with the rapid innovation in semi products and most of the end markets, and you get cyclicality.

Over time, as the industry has grown and matured, companies are getting better at managing the cycles of supply and demand. Still, it remains a chief concern for all parties involved. Regulatory and global cold war type stressors have ramped up significantly in the last few years - adding new variables to the ecosystem. All this is to say, plenty of uncertainty exists in this area of the market.

If one wants to better understand the uncertainty, there's no better place to start than revenue/inventory spending cycles and this monster of a chart:

Here we see the cycles of the inventory play out in inventory to revenue. If we had more space or better glasses, we could look at the 1970s and 80s where the cycles were even rockier than the 90s and 00s. After that, the cycles have started to smooth out quite a bit. The industry has become more mature, reached higher scales, and the collaboration of the major players in the supply/demand dynamics has become tighter. Suddenly in 2022, the predictable oscillation left and was replace by the big scary/exciting spike.

If you reference back to the industry revenue and profit charts from the beginning of this journey, you'll recall the last few years the industry has experienced exponential growth. Growth roughly in line with the increase in inventory to debt. This is the biggest question in my mind and in the minds of all semi investors.

Is this a harbinger of a nasty cycle driven by oversupply or indicative of a step change in the industry?

I believe it's the latter. Am I confident enough to bet my life on that? Absolutely not. Am I confident enough to bet my personal capital on that? Apparently so.

I'll depart from this for a little while now, but the supply/demand balance should remain in the forefront of our minds as we interpret the rest of the industry deep dive....