During this past year, the big winner in mobile computing has been Micron (MU) as profits rebounded and the stock has soared from $5 to over $21. With their acquisition of Elpida a vastly narrowed field from 20 years ago, when we had true Peak DRAM has presented itself. The increased financial and IP barriers appear to make Micron a safe harbor in a sea of mobile turbulence. It has been a long slog, however, it can no longer be ignored, that in both mobile and servers, memory is becoming a greater force in terms of value and potential performance/watt upside. Can we expect Intel (INTC), the platform king to downplay memory much longer? Intel's analyst meeting may have offered some clues that were hidden in their presentation.
In reviewing Intel's CFO Stacy Smith's presentation one more time, there is a sense that the company is making a couple of strategic shifts. The first and most significant is the plans to reduce the average die size of their x86 client processors by 29% over the next two years. This has several meanings and ramifications. First, Intel believes that the war with AMD (AMD) and Nvidia (NVDA) in the PC space (outside Gamer PC boxes) is effectively over. They have won based on the optimal mix of performance, power and space. They are communicating that they have the luxury of downsizing the die size at a faster rate than their pricing and therefore margins should expand on the largest piece of their business, the $33B x86 client group. This is significant, as I will explain soon.
As per expectations, the Data Center business appears to be able to clock in 12-15% CAGR, which will double Intel's most profitable business in the next 5 to 6 years. At nearly 50% operating margin, the effect is to be accretive on average margins.
The combination of the two strongest business units delivering a combined $44B in revenue, $16.8B in Operating margins and Gross Margins that I estimate is roughly 70% and growing means that Stacy Smith's corporate Gross Margin target of 60% will be easily exceeded if everything remains equal. But everything will not remain equal. Intel is communicating that it will ramp lower margin Atoms and possibly NAND in order to increase their presence and overall share.
Intel's success in the past has always been based on the strength of expanding its presence beyond the microprocessor. In the 1990s, in order to build a castle wall around the x86, they took control of the chipset that served as the conduit between the processor and memory, graphics, disk drives and I/O. The major achievement under the Otellini regime was to bring the graphics chip onboard and with the ultrabook platform, the story is now complete. Except when it comes to memory, which they should be eyeing closely because the ultrabook platform is still shrinking. Smaller, should play to Intel's integration and packaging strengths.
The tragedy of mobile for Intel these past few years is that the company first tried to slot in a device that had a performance and power and space profile for netbooks using cheap HDDs. If Intel wants to own a platform, they have to have the high dollar silicon. Application CPUs are commoditizing quickly as per the rumor of $10 mobile Atoms and ARM processors in Taiwan. The significant dollars are in the memory (DRAM and NAND) and the wireless (Baseband and Wifi). Intel's roadmap shows significant average die size reductions with the integration of baseband over the next couple years. This confirms the apps processor is low man on the totem pole - quite a change from the value Intel gets from x86 in the server and client markets. To win mobile, then means to own memory, which is lower margin but delivers revenue and in high volume. Stacy Smith showed only one slide on NAND Flash in his presentation. It's what Intel doesn't say or ignores that always intrigues me and leads me to what if?
If I were to guess, I would expect that major initiatives are underway to closely couple NAND and DRAM in all platforms (server, client, mobile) underneath the benefits of better performance/watt, smaller packaging and smartly managed thermals that are difficult for competitors to match. This would seem to suggest a closer collaboration with Micron with some possible expansion of NAND production in Intel's Joint Venture. To win mobile the next couple years is no different from winning the PC in the 1990s, one has to control the major pieces that force customers to launch their products on your introduction timeline and not that of another semiconductor vendor. Intel has that with Server and Client, but lags Qualcomm (QCOM) in that regard on mobile and must know Samsung (OTC:SSNLF) is headed in the same direction. Perhaps Apple (AAPL) as well.