The age of the smartphone and tablet has brought about a rather sharp and unexpected revival of companies that, since the dot-com bubble, have generally been considered unremarkable by investors. Firms that were either left-for-dead or simply ignored by the investment community have now essentially become superstars in this "new" era of technology investment. This means high multiples, strong media hype, and a jarring disconnect between investor-perception and the underlying financial fundamentals. This type of hype creates some of the most lucrative short-selling opportunities on the market.
The latest and greatest example of such financial market folly lies in shares of ARM Holdings (ARMH), the U.K.-based developer of processor IP for low-power embedded systems such as micro-controllers, smartphones, and tablets. The company exhibits the classic signs of a bubble, similar in nature and sentiment to previous pie-in-the-sky stocks such as Netflix (NFLX), Groupon (GRPN), and Zynga (ZNGA). These companies each saw a sharp drop in share price and valuation once it became clear that these business models could not hope to generate the kinds of returns that the hordes of sell-side analysts as well as giddy retail investors had previously expected.
ARM Holdings is one of the most expensive semiconductors in the world, trading at 68x trailing-twelve-month earnings (against an industry average of 24.08x), 18.61x sales (against an industry average of 3.66x), and PEG ratio of 3.68 against an industry mean of 1.99. More importantly, ARM generated $386.31M in cash from operations, which values the company at 42.5x free-cash-flow. Compare this to its peers (data sourced from Yahoo Finance)
- Qualcomm (QCOM): 23.4x FCF
- Intel (INTC): 18.42x FCF
- Nvidia (NVDA): 17.75x FCF
- Texas Instruments (TXN): 11.6x FCF
- Broadcom (BRCM): 14.36x FCF
One would suspect that for this premium, a potential investor would be buying fairly explosive growth. Not so with ARM Holdings. According to the 11 sell-side analysts that cover the company, ARM is expected to post 12.8% sales growth in the current fiscal year. While nothing to sneeze at, it certainly does not support the firm's current valuation. However, this but scratches the surface of the absurdity of ARM's current price, as we will detail here.
Understanding The Company's Technology
ARM Holdings is a U.K.-based semiconductor firm that focuses on developing low-power embedded processor IP for use in a broad range of applications from the now in-vogue smartphones and tablets to more mundane applications such as microcontrollers (which comprise the majority of ARM's business). In particular, there seems to be much confusion as to what ARM actually produces, so it is worth a few paragraphs to go into a little more detail.
There are two critical -- but distinguishable -- aspects to microprocessor design worth mentioning here (there are others). First, we have what is known as "instruction set architecture," which essentially defines what commands that processor understands. These commands -- known as the processor's assembly language -- are what define a particular processor from a programmer/compiler writer's perspective. This is why a program written for an Intel-based machine (which runs a different instruction set) will not run on an ARM-based machine.
The next aspect of microprocessor design is called the micro-architecture. This is the actual design of hardware that can execute the instructions defined by the instruction set architecture. The quality of the micro-architecture as well as the particular design target for that particular micro-architecture are generally what determine performance and power efficiency. For example, one could design a very low power 2W system-on-chip for a tablet or a 150W power-hungry high-performance processor by implementing the same instruction set with wildly different microarchitectures, fabrication processes, and so forth.
ARM Holdings designs a base instruction set as well as a handful of micro-architectures for use by product licensees. Customers can choose either to simply license the right to design a CPU that implements the ARM instruction set (and thus gain software compatibility with all pre-existing ARM code), or it can simply use ARM's off-the-shelf designs and then build a system-on-chip around them.
Separating The Hype From The Business Model
There is no doubt that ARM's technology has come to the forefront of the technology stock universe as the media and the sell-side analysts champion ARM Holdings' positioning as the core and/or instruction-set supplier for the majority of the world's smartphones. The company collects a small royalty for every ARM-compatible chip that is shipped, and despite 2.2B ARM-compatible products shipped in Q2 (but license revenue recognized in Q3), ARM only saw $107M in profits before tax.
To make it perfectly clear, despite having nearly 100% of the smartphone and tablet market, as well as nearly the entirety of the microcontroller and embedded space, ARM fails to generate net income that comes out to even a fraction of that of its licensees and peers in the semiconductor industry.
The company's initial success in the mobile computing markets has been primarily due to the firm's long-standing focus on the low power/embedded space, which allowed its products to evolve, due primarily to process technology advancements, to become fast enough to enable smartphones and tablets. The low power, low ASP application processor segment was largely overlooked by larger competitors early on, which gave ARM and its licensees the spotlight in the mobile processor world.
However, going forward, it is clear that the actual ARM core designs will become less important, and that ARM's primary role in the mobile world will be as an instruction set licensee, given that there is now a substantial software base for the firm's instruction set in place. No other company could reasonably hope to compete in designing its own instruction set (other than Intel and its already established X86 ISA) since the software support and toolset -- usually the product of many years of development -- would be an almost intractable barrier to overcome.
In short, ARM's business model will generate consistent, but low amounts of revenue and net income compared to its peers, which means that it is very unlikely that it will ever grow the top and bottom lines enough to warrant its valuation.
However, the stock has seen an extraordinary run post-earnings and even more recently over the last couple of days. The earnings report was nothing to write home about -- it beat analyst revenue forecasts by $2M (for a total quarterly sales figure of $227M) , and as a result, the company has gained over $4B in market capitalization.
But now there's a new catalyst propelling ARM to even higher, more shortable, Internet-bubble like valuations: a buyout rumor.
Intel Buying ARM Makes No Sense
The rumor that began spreading was that ARM was in "off-the-record" talks with Intel about a potential acquisition. The source is questionable, as vague notions of "dealers" hearing "whispers" certainly does not lend itself to credibility. However, shady sources are not the problem here. The problem here is that this makes zero sense.
First, there is a misconception that Intel cannot design low power chips to fit into smartphones. This is patently false, and has been disproven by highly competitive shipping products from Intel today. For instance, Intel is now shipping its Atom Z2460 mobile processor for smartphones, and in every relevant metric -- performance, power consumption, and compatibility -- it is as good or better than leading edge ARM processors. In fact, Anandtech -- a leading technology review website -- compared the Motorola Droid Razr i (powered by the aforementioned Intel chip) to the leading smartphones with the highest-end ARM processors in them, including Apple's (AAPL) iPhone. The site had this to say:
At least based on this data, it looks like Intel is the closest to offering a real competitor to Apple's own platform from a power efficiency standpoint.
Further, performance is not a concern (the "rumor" suggests Intel's chips are slower than the various designs from the ARM camp) as evidenced in the very same review. Not only is the Intel chip a formidable competitor, it is also the fastest processor available:
Intel's Atom core remains very competitive with the best of the ARM world. A single core Atom still ends up being the only CPU that can regularly outperform Apple's Swift
So right off the bat, the rumor is predicated on a very serious and egregious error. Keep in mind that Intel's processor is using what is essentially a design that has been unchanged since 2008, only fit into a lower power envelope. A next generation chip, built from the ground up, should only further Intel's lead.
In terms of chip design, Intel is fighting the best of the best in the ARM ecosystem. With a 5-year-old design.
Compatibility - Still No Reason To Go ARM
Since ARM processors were in mobile devices before Intel's, a substantial software base for ARM's processors was built up. However, due to the way that the Google (GOOG) Android operating system and its programs are designed, instruction set is not relevant to the majority of applications since almost every Android application is written in Java. Java programs are written in a high level language and then compiled down into an intermediate form called bytecode. This bytecode is then, at execution time, translated into the actual machine instructions by a virtual machine. This means that the only barrier to compatibility is having a virtual machine built into Android that can translate Java bytecode into native instructions. Intel and Google have achieved this, which is why software compatibility on Android is essentially a non-issue. This is also why we have shipping Android phones with Intel processors in them today.
So the main barrier to entry for a new instruction set -- X86 in the mobile world -- is a non-issue on the target platform that Intel is aiming for in phones. It also helps that Android's global market share is 75%. That means that Intel's chips could power the majority of the world's smartphones without any technical limitations. The remaining 25% -- which consists of iOS, Blackberry, and Windows Phone -- is locked up mostly by Apple, which designs its own processors. There should be very little in the way of a barrier to entry for an Intel-powered Windows Phone, as Microsoft's applications are built in a very similar "virtual-machine" based way. This is how Windows RT and Windows 8 will be able to run the same, "modern UI" applications despite running on different instruction sets.
These arguments all apply to tablets, with an additional edge for Intel, as Windows 8 tablets are natively X86 compatible.
Even If Intel Went ARM, No Need To Buy The Company
ARM doesn't build chips. It licenses core designs and instruction sets. Intel, if it were to go ARM, would not use off-the-shelf ARM cores, but would likely design its own ARM compatible chips. There is, of course, no technical reason to do this (and certainly no financial, as it would need to pay for a license and royalty fees -- not to mention that it loses all ability to modify/enhance the instruction set). But if it were to do so, it would simply buy an ARM license and pay the 1.8% royalty to build and sell the chips. Why pay $17B (plus whatever extra fee) when Intel could accomplish the same goal with less risk, and for substantially cheaper?
Keep in mind that Intel already builds many ARM chips for things such as network processors and SSD controllers. In fact, Intel is one of the largest ARM licensees!
Also, as discussed above, ARM trades at ~43x free-cash-flow and 70x earnings at current levels. A buyout premium would likely increase that significantly. This makes no sense when an instruction set license is cheap.
Insider Sales Confirm Bubble
While ARM is busy telling the investment public that it is going to take over the world with its licensing business model, the insiders are using these inflated prices as an opportunity to get out. ARM is not a U.S.- based company, so it does not need to register with the SEC. That means it is exceptionally difficult for U.S.- based investors (the ones generating all the media hype) to watch insider sales. Luckily, I have done the work for you:
November 7th Selling:
November 9th Selling:
There is a lot of insider selling, and no insider buying. A coincidence at near all-time highs? I think not.
MIPS Buyout - Did Nobody Notice?
ARM's chief competitor in the low power processor licensing space -- MIPS technologies (MIPS) -- sold its CPU assets to Imagination Technologies for $60M. Now keep in mind as ARM is just transitioning to a 64-bit design and generating huge amounts of fanfare, MIPS has had a 64-bit instruction set and accompanying designs for many, many years.
Further, MIPS processors are now supported on Android. With Imagination's wide technological lead in mobile graphics, it now stands to reason that Imagination is positioning itself to become a direct competitor to ARM, eating into its high gross margins.
This is just an additional threat, on top of the Intel threat.
In short, ARM is now exhibiting the classic signs of a bubble. In addition to a sky-high valuation, it is now subject to a horde of nonsensical buyout rumors and speculation. At near all-time highs, this is an obvious short candidate for those who can also see the obvious folly in the price of the firm's shares.