Tuesday, September 13, 2022

Texas Instruments ($TXN)

Back in Q4 2020, we read The Chip: How Two Americans Invented the Microchip and Launched a Revolution:

Transistors were a great invention, but pretty quickly the circuit designs using them outpaces the ability to reliably and economically wire them together: the "problem of numbers". The solution was to integrate all of the circuit parts, in miniature, in a single, monolithic block of semiconductor material. "If all the parts were integrated on a single slice of silicon, you wouldn't have to wire anything together. Connections could be laid down internally within the semiconductor chip. No matter how complex the circuit was, nobody would have to solder anything together. No wires. No soldering. The numbers barrier would disappear." The idea was simultaneously invented, by Jack Kilby (of Texas Instruments) and Robert Noyce (of Fairchild).

Texas Instruments (TXN) is still around, and is now 71 years old. Most well-educated people are familiar with them because of their pocket calculators, even if not because of their founding role in the semiconductor industry. Despite the association with the calculators for math class, those are only a fraction of TI's revenues. The overwhelming majority of their revenue ($14 billion in 2021, 77% of total) comes from the sale of analog semiconductors:

Analog semiconductors change real-world signals, such as sound, temperature, pressure or images, by conditioning them, amplifying them and often converting them to a stream of digital data that can be processed by other semiconductors, such as embedded processors. Analog semiconductors are also used to manage power in all electronic equipment by converting, distributing, storing, discharging, isolating and measuring electrical energy, whether the equipment is plugged into a wall or using a battery. As the digitization of electronics continues, there is a growing need and opportunity for analog chips to provide the power to run devices and the critical interfaces with human beings, the real world and other electronic devices. Our Analog products are used in many markets, particularly industrial, automotive and personal electronics.

Our Analog segment includes the following major product lines: Power and Signal Chain. Power includes products that help customers manage power in electronic systems. Our broad portfolio is designed to manage power requirements across different voltage levels, including battery-management solutions, DC/DC switching regulators, AC/DC and isolated controllers and converters, power switches, linear regulators, voltage references and lighting products. Signal Chain includes products that sense, condition and measure real-world signals to allow information to be transferred or converted for further processing and control. Our Signal Chain products include amplifiers, data converters, interface products, motor drives, clocks, logic and sensing products.

The rest of the revenue is from the Embedded Processing segment ($3 billion of revenue in 2021, 17% of total):

Embedded Processing products are the digital “brains” of many types of electronic equipment. They are designed to handle specific tasks and can be optimized for various combinations of performance, power and cost, depending on the application. Our devices vary from simple, low-cost microcontrollers used in applications such as electric toothbrushes to highly specialized, complex devices such as motor control. Our Embedded Processing products are used in many markets, particularly industrial and automotive.

An important characteristic of our Embedded Processing products is that our customers often invest their own research and development (R&D) to write software that operates on our products. This investment tends to increase the length of our customer relationships because many customers prefer to reuse software from one product generation to the next.

Our Embedded Processing segment includes microcontrollers, digital signal processors (DSPs) and applications processors. Microcontrollers are self-contained systems with a processor core, memory and peripherals that are designed to control a set of specific tasks for electronic equipment. DSPs perform mathematical computations almost instantaneously to process or improve digital data. Applications processors are designed for specific computing activity.

We can also look at TI's revenue by the end market that their products are sold into: industrial (41%), automotive (21%), personal electronics (24%), communications equipment (6%), enterprise systems (6%), and calculators/other (2%).

Something interesting about TI's business is that it has a very diverse product portfolio (80,000 products) and a very diverse customer base (over 100,000). Only 60% of their revenue comes from the top 100 customers. What is great about a product portfolio like this is that it doesn't necessarily pay for a competitor to copy any single one of the chips that TI sells. In other words, each product is its own little competitive castle; you couldn't put TI out of business by copying just once of them. (As the company says, the "diversity and longevity of our products, markets and customer positions [provide] less single point dependency and longer returns on our investments.") 

Great comment about the business: "[O]nce TI wins an automotive socket, it typically occupies that slot for at least 5 to 7 years. For industrial applications the product cycle is even longer, stretching out decades."

A couple of attributes of TI have caught our attention, starting with the very first sentence of the MD&A section in the annual report:

We design, make and sell semiconductors to electronics designers and manufacturers all over the world. Technology is the foundation of our company, but ultimately, our objective and the best metric to measure progress and generate long-term value for owners is the growth of free cash flow per share.

Next are the margins. For the trailing twelve months ended Q2 2022, free cash flow was $5.9 billion, which was 30% of revenue. That's revenue to free cash flow conversion like a tobacco company. Over those 12 months, TI invested $3.2 billion in R&D and SG&A, invested $2.8 billion in capital expenditures, and returned $6.2 billion to owners (4.2% of current market cap). 

The gross margins are tobacco-like, too (70%):

The business quality combined with good capital allocation has led to a shrinking share count and rising EPS:

Interesting observations from someone else's writeup of TI:

  • Every electronic device has analog content - from cars to medical equipment and home appliances. Switching costs are high since analog content is designed into a system that, once designed, is hard or costly to change. Additionally, analog and embedded processing have a product life cycle that spans decades. TI's focus segments are automotive and industrial, where adoption of semiconductor content is on the rise. They are also moving more production to larger wafers, reducing the cost compared to competitors who rely on third-party foundries for high-volume applications. 
  • Analog is a subscription-like business with consistent revenue from long-lasting products. The chips are built on older technology, not at risk of being innovated away. Analog is a small cost to the overall system and unlikely to be the pain point for customers. Furthermore, diversity in both products and customers creates a barrier for new entrants and creates a 20-year head start for incumbents.
  • Competitors such as Analog Devices and Maxim Integrated do not have enough volume to justify investing in 300mm wafer fabs. They have instead opted for a different strategy where they outsource manufacturing to a foundry, arguing it helps them better control the cyclicality of the business. Competitors also cannot justify larger sales forces or the same product breadth, because of their relative scale. As an example, TI can spend more on R&D on an absolute basis while still spending less as a percentage of revenue. This should lead to better and larger product portfolio than that of competitors in the long run.
  • TI is the largest analog semiconductor company in the world with an estimated market share of 19%, up from 15% in 2011. TI’s continuous market share gain in an industry with high switching costs is a testament to their competitive advantage and management’s ability to execute its flywheel strategy of product diversity, market reach, and high-volume manufacturing. It is also worth noting that TI's market share has also been created mostly organically with its latest acquisition being in 2011 compared to Analog Devices who has been a serial acquirer...
  • The semiconductor industry spends on average 30% of their revenue on capital expenditures. They continue by saying most of these investments go into supporting leading-edge technology. TI’s products are built on older and cheaper technology that lasts for longer, allowing for lower capital investments. TI’s capex as percent of revenue over the last 5 years averages at 5% while management guides 6% long-term. In addition to capital required for equipment, TI requires less research and development costs. This is in part due to their use of older technology but also because their products are not sold predominantly based on power and performance. Thus, their products remain in market demand for decades compared to leading-edge companies requiring newly developed chips every 18-24 months.
  • New entrants would also find it hard to compete with TI since it is a catalog company with thousands of products where their number one product represents ~1% of their total revenue. A competitor would have to reproduce TI’s entire catalog to stand a chance competing with TI’s sales force and product portfolio.
  • TI is currently trying to build higher inventory levels than prior to the covid crisis. The reasoning is that there is a low risk of their products becoming obsolete, hence the worst case is that they have excess inventory that will take longer to consume. Concurrently, they can guarantee their customers product availability. This widens their moat and allows them to win over customers who will remain for a long time because of the long product life.

The one fault in this TI idea is the price of the company equity. it's not cheap: the FCF/EV yield is 5.3% based on year-to-date earnings. (It's priced like Lamar Advertising.) Some people think that TI is currently over-earning, because of over-ordering as a result of worries about supply chain problems. This was discussed on the Q2 2022 call:

I think many have reported and we can see in the filings that our customers have had, that there’s clear signs of inventory being built over the last several quarters. And there is discussions on how much inventory do they hold more permanently and those types of things. We’ll see how that behavior changes over time. I think there’s some places where that probably will stick and probably some places where it won’t.

On the other hand, some parts of the economy are doing really well (which makes sense under our value vs growth paradigm), as TI's CEO noted at a recent investor conference:

Chris Danley - TI was one of the first companies to see the slowdown, and now it seems like every week, somebody else sees it. What do you think was unique about you guys and your ability to see this coming first? 

Rich Templeton - When everybody says no, the sky is the limit, there’s no end, you’ve probably got to be concerned that they aren’t looking. There could be that through construct of now over 60% of our revenue direct, things like TI.com online to where--that’s just real time consuming nature, we just haven’t had something like that in prior cycles. Can they assist us in being able to see it, and this in addition to looking, those are probably the only ways I’d think about it being different.

It’s also a slightly different cycle in that it’s not a synchronous downturn, and just look at second quarter where one extreme, if you were in PCs, it was a pretty miserable quarter, and if you were in automotive, there was no change. Typically if you go back to the bigger changes in late ’08 and bigger changes in late 2000 and early ’01, those were highly synchronized end market corrections, and this one is a little different on that. [...]

I don’t spend that much time focused on cycles because to me, the more interesting discussion is what’s going to happen in 2025 and ’26 and ’27, because those are the choices you’re making, those are the decisions you’re making today to really get positioned for them. That dominates 90%-plus of what we do, so preparing in this downturn is actually about getting capacity online and getting these things built out so that you’re in a position in 2025 to make sure that you can run very hard, depending on what that market wants to be.

Now, the other thing. Chris - and I know you’re a historian in terms of trying to keep track of this stuff, to me potentially the different--the cycle you may want to go back and look at is ’95 - ’96 because 2000, 2001, the Y2K cycle emerged into really a pretty weak personal electronics market. The 2008 - 2009 cycle emerged into a pretty weak global economy because of the structural impact from the global financial crisis, and if you go back to ’95, ’96, you actually had what turned out to be an eight-year--seven, eight-year secular run with the growth of PCs and cell phones. Well, it isn’t going to be PCs and cell phones in this secular run, it’s going to be industrial and automotive because of semiconductor content.

We are not selling our oil and gas investments to buy TI today. But we are interested in assembling a buy list of great companies to invest in down the road.

Just today, it appears that Biden wrote a put option on oil at $80 per barrel. As someone on Twitter pointed out, next time they can sell crude at $130 per barrel, then refill it at $100, resell it at $160, refill it at $130, and so forth; a "government-sponsored ascending trading channel for oil"!

Investors double count, so if oil goes to $160 not only will our companies' profits increase, but the valuation multiples should go from laughably low to reasonable - maybe even bubbleish!


CP said...

I’m Chris Danley, your friendly neighborhood semiconductor analyst here at Citi. It’s my distinct pleasure to kick off the tech conference semiconductor-wise with the reference standard company in our opinion for semis, Texas Instruments. Today, we have the Chairman, President, CEO - there’s probably a few other monikers or initials after his name, he does so many things around there - Rich Templeton.

As those of you who have followed my research over the years know, TI has been one of our favorite semiconductor stocks for at least a decade, if not more, for a combination of really two factors: one, the market they’re in, high end analog, which we consider the best market in semis and one of the best markets out there; and then number two, this man standing to my left, Rich Templeton.

If you look at the statistics, which even a semiconductor sell-sider like myself can determine, the margins have gone up for, I believe, basically 20 years since Rich took over, hitting new highs every upturn and hitting higher lows in every downturn, and then more importantly he gives that cash that he generates back to you guys in the form of both dividends and buybacks. I think other than some other semiconductor companies that try and be a growth company, and Rich has never pretended to chase any of the various shiny new pennies in semiconductors, so it’s really my pleasure to welcome him

CP said...

Chris Danley

Yes, so on that note, I want to touch on a few things that you guys have done, sort of regardless of cycles. I remember sitting down with you 12, 13, 14 years ago, the last big downturn, and you guys were buying up used equipment, buying up any piece of capacity you possibly could to be able to not let your customers down. As we look at what’s happened in this cycle, we did have some shortages out there and we’ll talk about the possible fixes. I guess I get this question all the time, how did we end up in this big-time shortage situation, and then how do you feel that TI was prepared for that?

CP said...

Texas Instruments Inc. authorized $15 billion in share repurchases and boosted its quarterly dividend by 8% to $1.24 a share, rewarding investors after a difficult year for chip stocks.