I've been a pretty vocal ARM (ARMH) bear, but as my readers know, making money on this trade hasn't been easy. While the short position had been marginally profitable following some serious active trading, I didn't get my "big break" until the "Silvermont" launch, the Galaxy Tab 3 win, and the Computex trade show (catalysts that I said would drive the next leg down) lining up the stars for my puts and short common stock position to pay off handsomely. I am currently out of the position, but still believe that ARM - despite being a good company - has significant long-term downside with immediate catalysts. I plan to enter again, but will be very strategic about it. I outline my assessment of the situation and my strategy here.
ARM Is Overvalued And Fraught With Risk...And People See It
40x EV/EBITDA, EPS estimates that - even in the most optimistic of scenarios - aren't likely to be hit as high end smartphones/tablets saturate and prices at the low end are driven down significantly. It also doesn't help that the economics of the fabless ecosystem (20nm from the foundries won't bring a reduction in cost/transistor, and "16nm" - which is really just 20nm with FinFETs given a smaller name to signify a new process generation - will only make things worse) are about to collapse at the high end, which sets the stage beautifully for Intel (INTC), an IDM without any royalties to pay and fabs that are paid for by non-mobile devices to come in and take share from ARM's licensees.
Further, if anybody cares to peruse the sell-side literature claiming price targets of ~$44 - $50, you will find gems such as the following,
Given the back-end loading of ARM's valuation (we estimate c.80% of the EV is driven by forecasting beyond 2020), it is understandable that the stock (-30% off its peak) is prone to respond to the vagaries of news-flow which potentially affects the company's long term earnings potential.
So, let's think about this for a minute...if 80% of a $44.25/share valuation is derived from looking out beyond 2020, then what the hell is this thing worth over the next 5 years, which represents a much more reasonable time-frame given how technological disruptions are bound to happen? 20% of $44.25 is $8.25. Are you really willing to pay $38 for a stock that could be worth $44.25 if everything goes according to plan beyond 2020? Yeah, I know, sounds silly to me too.
The Problems With Shorting Now...
The problem that I see with taking a short position now, right ahead of the firm's earnings report, is that the company is - as it usually does - likely to hit/beat estimates. While the multiple has taken a good, solid whack since the Intel wake-up call and the MIPS alarm bell sounded (did you see those next generation, 64-bit multithreaded cores that they announced while ARM still works to get its very first 64 bit design out?), I don't think you're actually going to see any impact from these events on the financials for quite some time, which could make this earnings report painful to be short during.
More importantly though, ARM recognizes royalty revenues 1 quarter ahead of its partners' sales numbers. This means that right now, ARM based devices are still selling in droves, TSMC's (TSM) fabs - likely filled to the brim with ARM designs - are booked solid through Q3, and the next couple of quarters are probably going to be pretty damn good.
However, it's not all bad news for bears with bigger cajones, as ARM's rich multiple means that it is - as the analyst that I quoted above - very, very sensitive to sentiment and the deeper the fears of future market share loss, the lower that huge multiple (P/E, EV/EBITDA, whatever you want to use) goes.
Intel: Major Android Design Wins And/Or Windows 8 Tablet Pose Significant Risk To Multiple
My readers know that I think Windows RT (that is, Windows tablets that run only the Windows RT OS) is damned to burn in the fiery pits of software hell. It literally offers no advantages over an Intel powered Windows 8 tablet; not in battery life, not in performance, and certainly not in flexibility. The only "enterprise" users of the ill-fated Surface RT are the poor souls at ARM itself, and the rest of the world will be enjoying full Windows 8.1, "Bay Trail" tablets/cheap notebooks by the end of the year. In short, if Windows tablets become popular, then this means a broad tablet market share shift in favor of either Intel or AMD (AMD). Not so great for ARM's multiple.
Further, Intel will also be playing in the Android tablet space with its Bay Trail part, although I suspect that ARM's licensee Qualcomm (QCOM) will do its damnest to keep Intel Outside of the Android tablet turf that it worked so diligently to steal from Nvidia (NVDA). The Nexus 7 goes to Qualcomm, but there's a chance the next gen Nexus 10 goes to Intel (although I wouldn't bet my money on that one this round). Each design that Intel wins here further damages ARM's multiple, but we'll see just how popular Bay Trail is among the Android device vendors this Christmas. I would feel queasy, though, if I were long ARM to see Intel talk about sub $199 Bay Trail based tablets, as mentioned here.
Servers Serving Up A Disappointment
The most recent run up in ARM's share price has been driven in no small part by hyperbole surrounding its server efforts. While ARM's extension of its instruction set to 64 bits gets ARM to the starting line, it is by no means enough to simply "win" this market, no matter how many small startups loudly chant the "party line" of "ARM magic" leading to lower TCO. Please. If these companies were really serious about attacking the micro-server space with a 64 bit architecture, and if that's all it took to beat big-bad Intel, then they'd have been smart about it and gone at it with a MIPS64 based design (and there have been some seriously beastly MIPS64 based chips built - just ask Broadcom (BRCM) about its new XLP900 NPU).
Quite frankly, what's going on here is that you have startups that are looking to get acquired by developing a unique piece of technology that might be valuable to a larger player trying to go after this space. Applied Micro (AMCC), Calxeda, and such are so obviously looking to be bought out by the likes of Samsung (SSNLF.PK), Qualcomm, or perhaps even a server vendor looking to vertically integrate. They may succeed, but only time will tell whether Intel's dominance in this space can be challenged. But if the last decade has been a guide, the server market has been all about putting everyone else out of business or relegating big players like IBM (IBM) and Oracle (ORCL) to niches.
More troublingly, though, is that ARM gained nearly $10B in market capitalization for an opportunity that is worth $4.5B...to the rest of the industry. This is where ARM's ~2-3% royalty rate starts making things look significantly less attractive. If the entire micro-server space is set to be ~$4.5B by 2016, then even at a healthy 3% royalty and an even healthier 100% market share, this is a mere $135M incremental opportunity. Seems a bit overdone, no?
What Am I Doing?
Quite frankly, while I do believe ARM will "beat", I wouldn't be surprised if this ends up being a heck of a lot like what happened to Apple (AAPL) - any sign of competitive weakness is going to drive continued selling action with knife-catchers bragging about how they're getting a "steal" driving the next legs down as they panic stop out of their positions. And believe me, if Apple could see its multiple compress at 12x EPS and significantly less than that in terms of EV/EBITDA, you can be that things could get really ugly for ARM.
Two major catalysts that I have my sights set on for ARM are the following,
- September launch of Intel's "Bay Trail" which, if accompanies with some significant design wins, could cause ARM's shares to crater
- The launch of Intel's "Avoton" and "Rangeley" micro-server and network appliance SoCs, respectively. The first should shatter the assumptions that ARM's partners will win this space, and the second should represent an aggressive expansion into territory owned by ARM/MIPS and knock the micro-server wind out of ARM's multiple. This will likely be in late July/early August.
My strategy is to buy 1/4th or so of a January put position ahead of the earnings report and following any pop add to the position. Now, I expect that the upcoming earnings report will drive shares to the upside, but there's always a risk that the market does to ARM what it did to Apple - look for any/all chinks in the armor. If royalty rates don't rise as expected, if licensing revenue is down a bit, if the company even so hints at seeing competitive pressure from Intel, or even if the Street doesn't like the new CEO all that much, then these could be enough to take the shares down despite a "headline beat". That being said, ARM management is exceptionally skilled at chest-beating at these reports, and I think Warren East will try his best go out with a "bang".
Look, short selling isn't for everybody - it's dangerous and can be incredibly frustrating. But if you want something with actual, clear catalysts (seriously, I wouldn't want to touch Amazon (AMZN) or Netflix (NFLX) with a 10-foot pole), then this is probably one of the better shorts on the market at the moment. ARM is priced to perfection, but I have some real conviction that Mr. East's statements that Intel could never catch up to ARM in terms of a power efficient design will haunt ARM shareholders this Christmas - the makings of a real good short.
Additional disclosure: I may go long IBM and/or go short ARMH at any time.