ARM Holdings (ARMH) is the world's premier microprocessor intellectual property company. The business model is unfamiliar to many in that the company sells licenses and rights to use its proprietary microprocessor designs on a royalty basis. These microprocessors exist as a design file in a computer-aided design system.
ARM is about a billion-dollar company with an astounding 94% gross margin. The costs of the company come lower in the income statement as R&D (29%) and SG&A (30%), Net margin, after taxes, has ranged from 21% to a high of 28% last year. The three year compound growth rate is 21%.
Given these profit and growth parameters, ARM could support a valuation of 8.4 times revenue (P/S=8.4). The current P/S is 22.4! If that doesn't give a shareholder a bad case of vertigo, you might consider employment as a high iron worker on skyscrapers. The fair value of ARM is 8.4/22.4X47.20 = $17.70. The ugly fact is that even that share price is predicated on ARM holding the growth and margin numbers into the future. In my opinion, both of those numbers are unsustainable for a number of reasons, which means that even the $17.70 is risky high.
The product history of ARMH is interesting. The company receives royalties on literally billions of instances of its various microprocessors being used in a wide variety of products. By far the largest volume comes from relatively dumb cores used in products such as automobile keyless entry systems, garage door openers, security systems, toys, and many other ultra-low power, but very simple applications.
On the other hand, by far the largest source of revenue comes from the multi-year explosive growth in smartphones and tablet computers. The revenue per use of the first category, above, of processors might be measured in pennies or a piece of a penny each, but in huge aggregate volumes.
The revenue per smartphone or tablet for the more complex mobile cores could amount to $.50-$1.00 per appliance. ARM cores are now used in up to eight instances in the mobile Application Processor and are also used in most baseband processors. The mobile business could easily represent half or more of ARM's recurring revenue.
The continued success of ARM is tightly coupled to continued success in the mobile space since any reduction in adoption rate will prove disastrous to ARM revenue and earnings. ARM's continued success is also dependent on ARM's fabless customers' ability to secure state of the art manufacturing capacity from third party silicon foundries.
What are the threats to ARM's continued success?
On May 6, 2013, Intel (INTC) will introduce its new 22nm out-of-order Atom core. It will first appear in quad core trim as the Bay Trail SoC targeted specifically for tablet computers. Bay Trail is the only tablet chip that will be able to run both full Window 8 and Android operating systems.
Right behind Bay Trail will be the dual core version codenamed Merryfield, which will be targeted at smartphone applications with higher performance and lower power than the, already very good, Clovertrail+. The Merryfield will come with the XMM7160 baseband chip to enable worldwide compatibility.
Bay Trail and Merryfield, built on Intel's 22nm Trigate technology, represent clear and present danger to ARM based tablet and smartphone designs.
On the process side, there is no evidence that third party silicon foundries such as TSMC (TSM) can ever close the technology gap with Intel. No one has seen any kind of working or demonstration finfet device from TSMC. If TSMC could show such a process today, it would take it two years to ship real world products using it. By that time, Intel will be ending 22nm and beginning 14nm Trigate production.
The problem with the ARM business model is that applications are either won or lost. There is no in-between as there might be in the case of commodity products. In multi-sourced products you can get 100% of the business or 0% of the business, or somewhere in between. In the case of ARM, a lost design takes revenue to zero very quickly. If ARM begins to lose designs, it will spread and revenue will begin to decline.
ARMH stock valuation reminds me of the bubble valuation of Intel back in 2000-2001. The stock was selling for $70 and 70+P/E. It was worth $17/share, and I told all my friends and family who held the stock to sell it. Very few did, of course. Intel has since grown into about a $50 valuation.
In summary, there is nothing wrong with ARM. ARM has enabled more dumb, smart, and intelligence products than perhaps any other company I can think of.
There are, however, terrible things wrong with the ARMH stock price. To me, a 35-year veteran of the semiconductor business, the ARMH share price smells like the 2001 tech wreck bubble writ small.
If you own ARMH, protect yourself. Sell the shares; you can always buy them back. Alternatively, move to in-the-money call options and harvest most of your gains. I know not many will take this advice, just as not many took my advice to sell Intel at $70.
Tomorrow I will initiate a short (put) position on ARMH just ahead of the Monday webcast.