Take a look at which chips power the latest smartphones. If you ask the average investor, he or she would probably say, "ARM (ARMH)", but they would be dead wrong. Sure, the processors that power the majority of smartphones are indeed ARM-compatible, meaning that the processors all understand the same language, but the majority of them aren't actually designed by ARM; rather, they are designed by Qualcomm (QCOM). So, when people talk about Intel's (INTC) competitive position in the smartphone market and they say "Intel vs. ARM", what they actually mean to say is "Intel vs. Qualcomm". Of all of the ARM licensees left playing in the high end game (and on the merchant vendor side, there's only a few), Qualcomm is the fiercest.
Intel's problem isn't ARM; it's Qualcomm, and in the mobile space, it is going to be quite a challenge for Intel to threaten Qualcomm's dominance.
Qualcomm Has The Upper Hand
What you need to understand is that very similarly to Intel's dominant position in PCs and servers, where Intel has all of the key OEM relationships, the scale, and on top of all of that the best parts from a performance/watt perspective, Qualcomm holds a similar position in smartphones, and if what I'm seeing of the Snapdragon 800 in the media represents performance in production performance, then Qualcomm is well on its way to establishing itself as a major force in Android tablets. Even Microsoft (MSFT) seems to be in bed with Qualcomm as well, as evidenced by the recent flow of PR from Qualcomm regarding Windows RT.
So, Intel has a few problems that it needs to overcome in phones (tablets are an area where Intel is rapidly gaining traction, so I'm not too concerned about those; particularly as Intel still has a home-field advantage on Windows 8 and is throwing all of its resources at Android):
- Having a highly integrated solution
- Having the best solution from a performance/watt perspective in common smartphone tasks
- Establishing leadership in cellular modems
- Developing/acquiring the rest of the connectivity suite
Further, even if Intel were to pull ahead on the technology side of things, it would still have to work hard to displace the perception among the phone vendors that Qualcomm is the superior solution. Take, for example, this recent slide that Acer showed to its partners (image courtesy of DailyTech):
Notice that Qualcomm is viewed as providing the "best in class technology and experience", while Intel is listed as a "strategic play only". Now, for 2013, this makes perfect sense; Snapdragon 800 is simply a better platform than Intel's current Atom Z2580 as it has four cores (and can be marketed as such), LTE-Advanced, and substantially faster graphics. Now, while it remains to be seen how production devices perform under battery life/power constraints, the point is that Qualcomm's solution is the créme de la créme today, and Intel's next chance to really break in will be in 1Q 2014 with "Merrifield". According to slides presented at the launch of the "Silvermont" micro-architecture, it should finally have a significant performance per watt advantage - at least on the CPU side of things - over the Qualcomm and any other competition:
But that's not going to be enough.
Intel Needs Leadership In Modems And Integration
Tablets aren't so dependent on cellular connectivity, which is why you are seeing Intel push hard there first, where the computer performance and power consumption of the system on chip is king. However, in smartphones, the trend is towards integration of everything possible onto a single chip. This does a couple of major things:
- Mo' Integration, Mo' Money - the more that a system on chip vendor can integrate onto the chip, the more that the vendor can charge for the chip, as it will have fundamentally more value/content. This is why Qualcomm has been aggressively trying to integrate the cellular modem, WiFi, bluetooth, GPS, and FM onto a single chip. Once Intel has all of the right IP in place (which means getting its cellular modems caught up to speed with the latest Qualcomm parts and developing its own in-house low power WiFi), then it will be in a much better position than it is in today and would be poised for leadership.
- More Power Efficient/Smaller Footprint - the more you have on a single chip, the smaller the board footprint will be, and the lower power your system will be as any power overhead involved in connecting more chips disappears. For very low power devices in which battery life is key, it is generally much more efficient to have a single chip than multiple chips.
Intel has most of the puzzle pieces; the firm's acquisition of Infineon, coupled with organic expansion efforts (Intel has a modem development team in San Diego...right across the way from Qualcomm) shows that it is going to seriously fight on the modem/RF side of things (not to say that this will be an easy fight). Further, I believe that Intel is developing its own low power WiFi (its WiFi chips are for higher power settings, so it currently uses chips from Texas Instruments (TXN) as a stopgap), and the recent acquisition of ST-Ericsson's GPS business put it in a good position for future chips.
But the point here is simple: Intel can't just show up with the fastest, most power efficient CPU and call it a day, and the company knows that. Qualcomm and Broadcom (BRCM) both have extensive experience in such integration without too much concern about tweaking for maximum frequency/performance, while Intel's heritage is very high speed, laid out by hand processors that are incredibly fast, so there's a bit of a learning curve for the company (although I think after the Medfield/Clover Trail experiments, the worst of it is over) to get into that "SoC methodology" mindset.
Can Intel Do It?
I believe so. Anybody following the company knows that the firm has been on a hiring and acquisition spree for the last several years. Intel has been quietly assembling the pieces that it needs to really take the fight to Qualcomm. That's not to say that I think that gaining any sizable market segment share is going to be easy; it's not, and it's going to take a combination of world-class products, the development of key partnerships (Qualcomm's in bed with HTC and seems to be penetrating further into the Samsung lineup), and a clear technological lead in order to gain real design win momentum. "Clover Trail+" and "Medfield" were nice and all, but it's going to take "Bay Trail" and "Merrifield" - two parts designed with leadership in mind and not simply as a proof-of-concept - to really get a sense of what Intel can do here.
My guess is that leadership in tablets comes this year with "Bay Trail", but smartphone leadership probably won't be up for discussion in terms of the full system (I expect CPU leadership) until the 14 nanometer generation, and even then it depends on how aggressively Intel's modem folks can stay on the cutting edge of mobile standards and integration.
Long term, I expect the high end smartphone chip market to be a fight between Intel and Qualcomm, although it's too early to tell what kind of market share balance to expect over the next decade. In any case, Intel's problem ain't ARM, it's Qualcomm.
Additional disclosure: I will go short ARMH when the opportunity arises.