Intel reported its best quarter ever (Q3 2011) on October 18 2011 and projected a further 3% revenue upside for Q4 2011. This announcement came just three days after WD’s Hard Disk Drive factory was flooded and thus the extent of any impact was probably not understood. However, because Intel is a heavy CapEx, fast depreciating driven company, it is key for them to assess quickly how external events affect market demand and drives fab loading. In the past they have been known to quickly shift gears to get to where the market is going at the cost of short-term revenue. And so as they admitted in their January 2012 conference call they converted some of their production from 32nm to 22nm. We will never know how much was converted, but I speculate it was significant. Wall St. was initially not pleased and AMD benefited from Intel’s pull back. But now the benefits are starting to accrue as Qualcomm, AMD and nVidia scramble for 28nm production that targets the Mobile Tsunami marketplace. Only Apple and Intel are prepared.
Apple’s announcement on its refresh of the MacBook Air and MacBook Pro line this week shows that Intel has gained additional sockets with its Ivy Bridge with integrated graphics at the expense of AMD and nVidia. Whereas two years ago, an external graphics chip was used across the Apple product line, now it is relegated to the $1799 and higher MacBook Pro notebooks. Expect Intel to make similar inroads with the Ultrabook market.
Question: How much of the 28nm TSMC shortage has forced nVidia to focus its limited supply on desktop add-in cards instead of mobiles? It’s a Major Retreat that is likely unrecoverable.
Now consider Qualcomm and their 28nm ramp for the upcoming Apple iPhone5 and the Snapdragon all-in-one chip going into a wide range of non-Apple smartphones and tablets. Two months ago, it was revealed that they would not be able to meet demand until Q4. Now they won’t meet demand until the end of Q4 2012. As I speculated in a previous blog, I believe Apple has worked Qualcomm into a position that it has its supply needs met at the expense of its competitors and thus setting themselves up for a strong second half ramp with reduced competition from Samsung, HTC and the rest of the mobile players.
All of the above however doesn’t encompass the full extent of what is transpiring in the mobile semiconductor market and how Intel is prepared to increase its presence. Intel’s Medfield processor is late to the smartphone market but still competitive. Built in written off 32nm fabs, I presume Intel is cutting some very nice deals in order to ramp the product in volume and create havoc for nVidia, Mediatek, TI, Broadcom and Marvell. Qualcomm’s dominance at the high end with Snapdragon combined with Medfield at the low end will cause a market squeeze with several of the above exiting in the next twelve months.
Looking ahead to 2013, I expect Intel to enter the market with a 4G LTE solution and the mobile market merchant suppliers to narrow down to Qualcomm and Intel with Apple and Samsung continuing to develop for their own needs. If Apple gains significant market share in the Smartphone market over the next 9 months due to the 4G LTE supply situation, then I expect Intel will make inroads at Samsung with x86 Atoms. Samsung will have to do some serious soul searching as to whether it makes sense to design ARM based processors when Intel can offer solutions that are lower in cost and power and higher in performance – all due to Intel’s three to four year lead in process technology. If Samsung balks, look for Intel to leverage another player like HTC, Motorola to undercut the market.
Apple, unlike Samsung, can get Intel to build an ARM based processor in its fabs and likely has silicon running in its 22nm process at this moment. The dilemma they face is that going to Intel offers all the upsides mentioned above but the risk of losing other alternative suppliers. In addition, for Apple to get the best deal, they have to get to Intel before Samsung signs up. It is a true high stakes negotiation as Intel will try to land both Samsung and Apple at the exclusion of none.
Looking back over the past 18 months, one can see a definite trend that likely drove Paul Otellini’s decision to ramp 22nm hard last November. In my experience, the PC market has always had a need for processors built on the most leading edge process technology. Qualcomm is experiencing this same phenomenon today in the smartphone market with their 4G LTE chips. Intel, with its process lead, is making a bet that they will win the entire mobile market by 2014 when the successors to Medfield will integrate all functions, including baseband and be significantly cheaper, higher performance and lower power than anything its competitors can delivery. The Black Swan of 2011 likely gave Otellini the excuse he needed to break free from Wall Street’s quarterly demands in order to embrace the Mobile Tsunami Leading Edge Process End Game much sooner.
FULL DISCLOSURE: I am Long AAPL, INTC, ALTR, QCOM