There hasn't been much action or conversation on Intel (INTC) lately. Baird downgraded them today due to their incompetence in analysis.
Intel bought the Fujitsu 4G RF transceiver business. Now, that is a big deal because apparently Infineon isn't getting the job done on LTE RF and baseband products. It seems like a year ago that Paul Otellini said, to the effect, "The Germans seem to take longer that we would like, but when the products are done, they are very good." C'mon Paul, you were beating those guys with a stick and still nothing got done.
Quiet times with large companies are interesting times. The last time there was silence surrounding Intel was back in 2008-2009, some of which was in the depths of the Great Recession. Intel must have found the silence invigorating since in 2010 sales grew (exploded) by 24%, followed by another 24% in 2011. In two years, Intel added nearly $19 billion in sales for a 54% two year growth. Who says this elephant can't dance? That $53+ billion level of business has held for 2012 and 2013. I wonder if anyone out there understands how difficult that is. $19 billion is as big as Qualcomm (QCOM) is today; it is as big as TSMC (TSM) is today. It is almost five times the TOTAL FPGA business as represented by the combined business of Altera (ALTR) and Xilinx (XLNX). Intel's growth for those two years was about four times the total of AMD (AMD) or NVidia (NVDA). The growth was 1.5 Texas Instruments' (TXN). OK, OK, you get the idea -- 2010-2011 was a nuclear explosion of growth for Intel and they held it. Remarkable, simply remarkable. The analysts didn't see it coming and were caught flat-footed when it did happen. The stock price increased 50% while earnings tripled. So much for the "Efficient Market" idea. One of the things that was going on during the quiet period was the creation of a $10 billion data center business.
From 2010, Intel has spun two complete conversions from 45nm to32nm to 22nm throughout this huge enterprise. I might point out that Intel grew 54% with existing brick and mortar capacity; such is the power of moving only one technology node (45nm-32nm) during that time. Intel's capital spending averaged about $5 billion per year. Since then Intel has spent the $5 billion per year plus another $20 billion over three years for new ground-up facilities. I have spent a lot of time writing and thinking about this capacity, which should all be in place by the end of 2013. Adjusted for the 22nm shrink, I figure that Intel has 2-4 times the effective capacity that they ended 2011 with.
So, where are things sitting today?
The Latest processor, Haswell, was released in early June. The conversion from Ivy Bridge has continued so that Haswells in notebooks are finally available, and more are coming fast. Of course, that won't help because the pundits tell us that the PC is dead; much like the US was dead in the Great Recession. The Haswell will be in full production (no more Ivy Bridge) by the end of the year. Intel has recently commissioned a survey of PC usage that shows that, among a lot of other interesting information, 41% of US PC users intend to upgrade their computer during the next 12 months.
So Haswell is ramping aggressively and the factories are preparing for the 14nm Haswell, which will be called Broadwell, in 2014.
The first version of the new Atom architecture code named Silvermont has been announced. Now, announcement and release are two entirely different things. "Announce" is conversation, "Release" is real availability of parts. The first version of Silvermont, called Bay Trail, is due to be released in the third calendar quarter -- which is soon. These parts will be available in volume from the release date. The Bay Trail is a ground-up new design SoC with four cores and targeted for tablet computer format. The Bay Trail is targeted at tablets because tablets don't need an LTE device in many cases. That fits since the Germans are apparently unusually slow on delivering an LTE chip. Bay Trail is very likely to dominate the tablet market since the device is able to run both Android or Windows with no compromises. The Bay Trail is in pilot production with enough parts to get into customer hands for new design evaluation. These parts are usually provided under severe non-disclosure agreements that snuffs any reliable performance conversations. When the Bay Trail is released, there will be a flood of every kind of performance reports available overnight.
The second version of Silvermont will be the dual core Merrifield, which will be the Atom SoC targeted for smartphones. Merrifield will definitely need an integrated LTE since a smartphone isn't much good without the LTE function. We don't hear much encouragement on the LTE front, so the Germans are probably still having problems.
An interesting tidbit that I received from an investment site today listed all the SoCs used in the new high end smartphones. With the exception of Apple (AAPL), which uses their home designed application processor and a QCOM discrete LTE chip, all other smartphones use a QCOM SoCs, many with an integrated LTE function. All of 'em. No NVidia design wins in the high end.
That's good and bad news for both QCOM and Intel. Good news for QCOM because they have 90% of the business. Bad news for QCOM because they become a huge target on their back for Intel to aim at when they are ready with a complete solution for smartphones. It is bad for Intel because it shines a bright light on how powerful the LTE function is in delivering the design wins and business. The good news for Intel is that the LTE, integrated and discrete, will be here sooner or later. If they have to, they will stack money on top of that square peg until it fits into the round hole. When that becomes true, Intel will have a two node technology lead+ Trigate on QCOM and will be able to sell a complete solution for any price it takes to get the business, while maintaining high margins. More good news is the "big target" thing. Intel will be in a mud-wrestling contest with only one other contestant and that contestant doesn't have a fab.
Analysts have had a great deal of fun at Intel's expense on the McAfee acquisition. The point of that deal was to bring some resolution to the malware and virus issues by allowing some functions of the security solution to run in the CPU itself, thus speeding things up. When they bought McAfee, the Sandy Bridge was just coming available. They needed to design a CPU that had the "hooks" in it to run antivirus very efficiently. With the well-established Intel Tick-Tock methodology for designing new processors, Intel was two generations away from the Haswell, which of course has the necessary hardware. Associated with that, Intel/McAfee has announced LiveSafe which is a comprehensive security solution that sells for $79.99 per year. LiveSafe represents a high multi-billion dollar business for Intel once fully developed.
If you have been reading my conspiracy theory articles, you know that I think there are two things that keep Apple and Intel from fully embracing each other. First is the "memory trap" that Apple found themselves in with the impending bankruptcy and liquidation of Elpida. Without Elpida, Apple was totally dependent on their worst competitor for a source of mobile DRAM chips. Since the closing of the Micron (MU) acquisition of Elpida on July 31, 2103, that problem goes away forever.
The other problem is the dependence on Qualcomm for a discrete LTE chip to mate up with the Apple Ax chip. That problem will go away when the Germans finally get the LTE chip done, which they will.
Carl Icahn notwithstanding, the share price of Apple has been almost straight up since late July, when the closing of the Micron/Elpida deal was certain. Maybe that memory trap was more important than most think.
So, the new architecture CPU is here and shipping and ramping into Ultrabooks, the data center business is intact and likely to remain that way, the new Atom chips are being designed in and moving to production for tablets, smartphones, servers, and a sneaky little industrial/automotive version.
Intel has the capacity to double in semiconductors.
All the planets are coming into alignment for another $20-30 billion spurt of growth for Intel.
Will you be included in the run to $50 per share?