Summary: LONG.  Intel’s dominant share in markets which face reasonable long term growth prospects reflect historical competitive advantages which show no sign of fading, producing above normal profits and allowing significant reinvestment in sustaining these competitive advantages through R&D.  For this business of above-average quality, the consensus market valuation implies very low expectations that the business will be able to sustain recent earnings.  With net cash on balance sheet, a history of decent capital allocation, and an extremely liquid stock, this is relatively safe and cheap.

 

 

Note: standard aspects such as business model discussion, presentation of segment profitability and history are omitted since best covered elsewhere.

 

 

Introduction

 

Two well-established super-trends, worldwide internet adoption and the rise of living standards for large population regions, have proven much more difficult to make money from than early dreamers had hoped.  Many benefits have accrued to consumers rather than capital-owners, because companies that intended to exploit these two big growth trends have either not been able to keep barriers to entry high enough to earn above-average returns on capital, or their stock price valuations already reflected their competitive advantages so investors’ actual returns have merely met or disappointed original lofty expectations.

 

One or two decades into these super-trends that seem likely to continue for a while yet despite some potential macroeconomic speed-bumps, Intel appears to be an opportunity because it simultaneously enjoys sustainable competitive advantages and is cheaply priced.  As the internet continues its global penetration, expansion and revolution of so many industries, and as the forces of globalization continue to lift many millions more out of poverty into a self-sustaining positive cycle of economic development, Intel’s best days could well be ahead, not behind, and yet the stock valuation does not suggest this.

 

 

Investment thesis summary

 

1. High and stable market shares with consistently above average returns on capital are evidence that a competitive advantage has existed in the past.
2. The ability to do what few or no-one else can do is evidence that a competitive advantage can be sustained into the future. 
3. When a high market share meets an upward demand shock, it creates a rare opportunity to simultaneously make a lot of money and reinvest in the competitive advantage. This opportunity is not commonly encountered, and even less commonly exploited.
4. Technology investing presents fatter left tails (high uncertainty) but also longer right hand tails (successful outcomes can be extremely positive). 

 

Why does this opportunity exist?

 

The long-term, structural reason is that occasionally good investments are neglected due to the “missing the forest for the trees” problem. 

 

Parts of this investment thesis are so clichéd that they are hard to include in what attempts to be a serious write-up: an upward demand shock has happened and remains ongoing as the internet grows in its impact, and as computing devices become affordable to large populations.  There, I said it, I have a great idea: long INTC due to globalization and this thing called the internet, which I have it on good authority, changes everything. 

 

Now that most readers have left in disgust, I can feel a little less self-conscious to you select few who have continued reading (hi Mom).  There has never been a write-up of Intel on VIC since inception.  This despite it qualifying as a Top 30 Magic Formula stock for the universe all the way down to $4bn market cap.  (Meantime there have been 4 bullish write-ups of AMD, by people far more knowledgeable than me.)  Why is that?  Maybe it reflects an investing landscape where fees no longer go to people who naively present stocks that are simultaneously cheap, with high returns on capital, and deeply liquid.  Maybe the company is inherently flawed to anyone with technological know-how.  Maybe it will never re-rate to even an average valuation since there is no “catalyst”; (the list of Intel’s major shareholders is dominated by passive investing ETFs, indicating either avoidance or apathy from active investors).  Maybe the whole concept of pitching Intel is, like, so 1996.

 

Most generalists understand the Intel business model, and little of what I write will be news to anyone here.  Most will also be frequently reminded through personal technology experiences that PC demand is doubly challenged by weak consumption in mature economies (previously on VIC: “I'd be tempted to write up Intel for VIC but I guess it's hard to make a case for why people will keep buying PCs when the one I'm typing on right now is fine, and it is over 5 years old”), and also by the adoption of tablets and smart phones in both developed and emerging economies.  For the past 9 years the stock price has ranged $12-29, confirming to generalists that there is nothing going on with this story, it’s just dead money.

 

Specialist readers, apart from being completely horrified at my over simplistic analysis of what is undoubtedly a complicated business, also often seem intellectually unfulfilled and bored when discussing such a slow-moving monopoly as Intel’s.  Glamorous and new it is not.

 

Astute observers of technology, who do not happen to also be financial analysts, frequently neglect to even mention Intel’s role in the internet.  Andrew Blum, author of the recent book “Tubes – a Journey to the Center of the Internet”, spent many months researching and visiting locations all over the world where the internet physically exists.  His well-written book describes data centers and undersea cables in entertaining detail.  However Intel is not mentioned once, despite its 90%+ data center market share.  The well-paid plumber is not always obvious.

 

Quick quiz: which company makes the most money out of the internet?

 

If I take a broad definition, that the company has to be considerably helped in its current product sales by the internet (today very few buyers of PCs or iTunes would buy without the internet, hence both Intel and Apple qualify for my list), or was likely to be significantly impacted by internet growth (hence telecoms are included) I would say that Intel is currently the 4^th most profitable internet company (after Apple, Microsoft, and IBM, but ahead of Oracle, Google, eBay, Amazon, Tencent, Facebook and Baidu).

 

Why does this matter?  Two reasons.  I suspect that many people not familiar with current absolute EBIT numbers for all these companies would not instantly associate Intel with reaping so much of the internet harvest.  And more importantly, when you go back in time, few companies have persisted in this list during the massive changes that have been wrought since the first website was posted in 1991.  (My list would have put Intel in 3^rd place in that year behind IBM and HP; 3^rd in 1995 behind AT&T and IBM; 4^th in 2000 behind SBC Communications, Microsoft and IBM; 4^th in 2005 behind IBM, Microsoft and Verizon). 

 

Thus as the internet has swept across industries and geographies during the first 21 years of its life, Intel has shown both resilience and an ability to get paid, while not being the internet’s most frequently discussed or obvious beneficiary.  The forest has been missed for the trees.

 

The short term reason that this opportunity exists is probably because when Intel’s two biggest customers (HP and Dell) report weak sales as they recently did, it is hard for the market not to lower Intel’s valuation, despite Intel clearly occupying a very different position in the PC value chain.  That, and the curiously tasty, triple combo sundae “death of PCs”, “death of equities” and “death of China” seems rather popular at the moment.

 

 

High and stable market shares with consistently above average returns on capital are evidence that a competitive advantage has existed in the past.

 

Intel’s ROE has averaged 20.6% over the past 25 years, ROA averaged 15.7%.  While some technology companies can generate higher returns, few can sustain them for such long periods against the onslaught of competition where product life cycles are often measured in months not years.  These above average returns on capital, the product of dominant shares in several markets, are the result of significant barriers to entry, either legitimate or illegal due to antitrust reasons, as noted by the Federal Trade Commission’s 2010 complaint:

 

“Over the last decade, Intel’s share of the overall x86 CPU market (desktop, notebook and server) has consistently exceeded 65 percent; its share of the x86 CPU desktop market has consistently exceeded 70 percent; and its share of the x86 CPU notebook market has consistently exceeded 80 percent.  Intel’s monopoly position in these markets is partially protected by significant barriers to entry, including reputation, scale economies, intellectual property rights, costs associated with building and operating large manufacturing facilities, and research and development costs.  These legitimate barriers to entry make vigorous enforcement of the competition laws all the more important.”

www.ftc.gov/os/fedreg/2010/august/100810intelconsent.pdf

 

 

Legitimate barriers to entry are hard to find.

 

 

The ability to do what few or no-one else can do is evidence that a competitive advantage can be sustained into the future. 

 

Many people can serve a Coke cold; not many (legally) know the recipe to make one.  Therefore restaurants and Coca Cola Inc. earn vastly different returns on capital.  Likewise many people can build a PC; not many can turn compressed sand into a 14 nano-meters wide microprocessor.  Therefore PC makers and Intel should earn vastly different returns on capital.

 

As a company that has cumulatively spent ¾ of its current market cap, some $90 billion, on Research and Development activity since its 1971 IPO, Intel knows how to do lots of things which few others know.  In addition, and unusually, this technical know-how can also be translated into a non-commoditized manufacturing ability, enlarging the amount of activity that can be rationally expected to generate above average returns on capital.  Therefore the cumulative amount spent on capex, roughly similar to the amount spent on R&D or another $90 billion, has also been a good use of shareholder capital.  The complexity and scale required to manufacture such intricate products pushes up the returns on capital for what is evidently not just plain vanilla manufacturing.  Even if economies of scale in R&D can be scaled by competitors, given the incumbent’s market share few companies or allied groups can generate sufficient sales or profits to justify the capital outlay required to construct one new leading edge fab (fabrication plant), which is set to reach $10 billion for 450mm diameter wafers.  This means that as the industry and technology develops over time, the number of competing manufacturers should decline, not increase.

 

Gordon Moore, who observed that the density of transistors, or computing complexity, should roughly double every two years (Moore’s Law), neatly summarized why this technology could naturally lead to few dominant suppliers:

 

“Semiconductor technology has the peculiar characteristic that the next generation always makes this higher performance and cheaper – both.  So if you are a generation behind the leading edge technology, you have both a cost disadvantage and a performance disadvantage, so it’s a very non-competitive situation.  So companies all recognize that they need to stay on this curve or get a little ahead of it.”

http://www.charlierose.com/view/interview/657   (2005)

 

It therefore pays off for Intel, having acquired a leading edge on the development curve, to keep reinvesting in both R&D and manufacturing capacity, since this will extend and sustain its competitive advantage.  This is precisely what current management is doing.  Intel can do two big things that few others can replicate: how to solve the physical constraints to designing ever smaller, cheaper, more powerful chips, and how to efficiently manufacture these designs with high yields.  This is evidence that at least some of its competitive advantage should be sustained into the future.

 

 

When a high market share meets an upward demand shock, it creates a rare opportunity to simultaneously make a lot of money and reinvest in the competitive advantage.  This opportunity is not commonly encountered, and even less commonly exploited.

 

It might sound odd to describe the increasing aggregate demand for personal computing devices, mainly due to the spread of the internet and globalization, as “shocks”.  These trends have been written about for years and ad nauseam.  Furthermore I use the term as a permanent shift right in the demand curve.  Aggregate demand for the company’s products has risen beyond the mere proliferation of microprocessors per device: the number of devices, and the servers that connect them, is dramatically higher than it was just a few years ago, and is likely to remain elevated for the next several years.

 

This can be seen in both narrow and broad definitions of Intel’s total addressable market.  Despite a slowing in global PC shipments (with headline grabbing decreases in developed markets where many investors reside), aggregate unit growth is still positive.  Total PC shipments are now 360 million units annually, vs. 271 million five years ago and 132 million five years before that.  Meanwhile the total number of personal computing devices is growing tenfold every decade. (IDC: 1995: 100 million, 2005: 1 billion, 2015 forecast: 10 billion).

 

My insight into, and predictions about, future aggregate demand are laughably incomplete.  I do remember a great deal of angst about future tech spending in the US after the boost of once-off Y2K expenditures and the subsequent tech crash.  Yet despite a couple of slow years, total PC shipments climbed again, as laggards bought and early adopters refreshed.  I think that will happen again, but on a much bigger scale (the US now accounts for just 18% of global PC shipments vs. 35% ten years ago), almost irrespective of what happens to the global economy over the near term.  As an example of why I think the demand curve has shifted permanently to the right can be seen in Brazil, which in 2011 became the 3^rd largest PC market globally with a household penetration rate of 75%; this served as a catalyst for the government to make mandatory online tax filing for individuals; even if Chinese real estate crashes and Brazil's iron ore exports plummet, many Brazilian taxpayers will still need to keep a functioning reliable computer for this annual chore – hence a complete reversal to PC demand levels of ten years ago appears quite unlikely. 

 

Regardless of slight demand fluctuations over the next year or two, Intel’s total addressable market is now much bigger than it was, even if you just consider PCs, but especially if you include the servers required to power this ever-expanding network.  (If you are an optimist and think that Intel stands a reasonable chance of successfully entering the tablet and smart phone markets in a similar fashion to its 1980’s, seven year sprint from zero to 80% PC market share, then I guess you could justify a growth valuation).  For a company with dominant market shares, an upward demand shock is clearly a good thing.  For a company that takes this windfall and reinvests in extending its technological lead and competitive advantage, it is a very good thing.

 

Sure enough, while it must have been tempting for management to let windfall sales growth drop down the income statement to generate a record high operating margin, instead they elected to spend a record amount of sales on R&D, depressing current reported profitability commensurately.  Whereas on average over the last two decades the company had spent 13.0% of sales on R&D, last year the company upped that to a record 15.5% and in the last two quarters they have spent at a nosebleed 18.6%.  To give some relative context, Samsung Electronics is currently committing 7.4% of sales to R&D.  In absolute terms, in order to dimension the intellectual property and technical advancement that Intel’s current $9.4 billion annual R&D spend might be acquiring for them compared to potential competitors, Samsung Electronic’s R&D budget is the industry’s next largest, at $8.7 billion (TTM), but semiconductors represent just 25% of Samsung’s total sales.  This suggests that the biggest piece of Intel's R&D spend has zero overlap with Samsung, or in other words Samsung's direct R&D assault on Intel is currently outgunned about 4:1.  Other R&D budgets look puny in comparison: totals for Taiwan Semiconductor Manufacture were $1.1 billion (2011), AMD at $1.4 billion (TTM).  Intel spends more on R&D in a fortnight than ARM does all year ($255 million TTM).  This is not a company that is obviously resting on its laurels and squandering its hard-won, historical advantage. 

 

Similarly on capex investment in new fabs and manufacturing capacity, reinvestment is significantly outstripping merely maintaining the status quo (capex exceeded D&A by 78% last year and by 55% during the past 2 quarters).  Despite all this costly investment in the company’s future at the expense of current profitability, returns on capital have reached the upper end of historical ranges (ROE 26.2%, ROA 17.6%).

 

Such incremental reinvestment might therefore be masking just how good current earnings are.  Therefore it seems odd that the dominant theme in the consensus market’s assessment of this above-average business at its current below-average valuation seems to be the fear of an approaching cliff drop in profitability.  Is this valuation justified simply because of a "technology-is-hard-to-predict" discount?

 

 

Technology investing presents fatter left tails (high uncertainty) but also longer right hand tails (successful outcomes can be extremely positive). 

 

In addition to the standard risks of any investment, it could be argued that technology companies are harder to predict for anyone other than genuine specialists, or that the threat of technological obsolescence greatly increases the downside risks.  Specifically for Intel I would list the following investment risks:

 

Antitrust.  Numerous complaints have been settled and paid for in recent years, including those from the FTC, European Commission, NY Attorney General, Korean Fair Trade Commission, AMD, and NVIDIA.  Cumulative payments and changes in business practice to date may prove insufficient, especially in emerging economies, where the company is rapidly growing its sales but has yet to encounter significant antitrust friction.  It seems unlikely that these countries, with typical incentives to fund regulatory agencies by aggressively pursuing antitrust windfalls, will be restrained in their demands for penalties as they watch Intel dominate growing local industries where no national champions can economically compete.

 

Co-operation not competition, or the revenge of the ecosystem.  The process of scientific discovery is always unpredictable, but especially so if participants conclude that one party has become overly extractive in nature.

 

A step change in the underlying technology.  Most technology companies will face their Kodak moment sooner or later, and large successful companies can be very poorly prepared. 

 

Wasted time.  Bad execution or strategy that narrows the technological gap to the competition might be fatal, not merely inconvenient, since the whole gap to the competition can be expressed in time.  A one or two year lead can seem very significant when enjoyed, but can slip by very quickly during errors.  “It’s a very competitive industry.  There are lot of opportunities to make a mistake with something that turns over as fast as the products do in our business.  I think we have a good team in place and a good plan, but you always have to be concerned about the future.” 

Gordon Moore 2005 interview, http://www.charlierose.com/view/interview/657

 

Very cheap energy.  The benefits of Moore’s Law have been reaped in three dimensions to date: higher computing power, lower cost, and lower energy usage.  Any product that can be designed around three variables has a lot of utility, due to the number of combinations that can be designed for.  If energy ceases to be such a design constraint (datacenters alone account for 2% of total US electricity usage, meaning that server chip designs for lower energy use have been very valuable in recent years), then the opportunity to design for, and charge for, a variety of customer preferences will be significant reduced.

 

Apple.  The mixture of ambition and cash equivalent to Intel‘s market cap has to cause a few sleepless nights.

 

Capital allocation.  Consistency of mission has historically driven a very tight ship for capital allocation, with sensible reinvestments and large amounts being returned to shareholders by a 3.6% dividend yield and intelligent buybacks (“We took advantage of record low interest rates and borrowed $5 billion, primarily for the purpose of repurchasing stock. This was one of those rare opportunities where the weighted average cost after-tax cost of the debt was approximately half that of our dividend yield on the date of issuance.”  CFO Stacy Smith, Q3 2011 earnings call.)  However recent strong earnings have also increased management’s acquisition appetite, with notable investments in ASML and McAfee in the past couple of years.

 

 

Potentially offsetting all these downside risks, however is Moore’s law.  Management joke that when talking to investors, “when you run out of things to say, just talk about Moore’s law.”  It is easy to see why.  Big numbers impress financial types, and the long-term outcome of Moore’s law is that computing complexity increases by 1,000x in 20 years, or 1,000,000x in 40 years.  (Alternatively, what the last 40 years has looked like for Intel has been: 4,000x faster, 5,000x less energy per transistor, and 50,000x cheaper per transistor.)  It is simply very hard to predict what markets Intel might dominate in 40 years’ time when computing power has increased one million fold from today.  Or in other words, don’t sell low delta, long dated calls.

 

 

 

 

 

Additional materials

 

Emerging Market affordability – pages 8-14

2/27/12 Morgan Stanley 2012 Technology, Media & Telecom Conference

http://www.intc.com/events.cfm

 

History of market entry and market shares, PCs, Data Centers, HPCs - page 28

11/29/11 Credit Suisse Annual Technology Conference

http://www.intc.com/events.cfm

 

Investment needed for one leading edge fab – page 35

and

Delivering on Moore’s Law for 40 years – page 30

05/10/12 Investor Meeting 2012, CEO Paul Otellini

http://intelstudios.edgesuite.net/im/2012/pdf/2012_Intel_Investor_Meeting_Otellini.pdf

 

sources: IDC, Gartner, company, Bloomberg.

 

This is not investment advice and is not intended to be distributed in any jurisdiction where it would contravene local laws.  Header forecasts are consensus; diluted share count; deferred income is subtracted from cash.