There is a battle going on in the mobile space. The battle is over nothing less than domination of the mobile business. The contest can be boiled down to Samsung vs. Intel.
This article is about how the pieces of the mobile puzzle fit together and how Intel has alternate paths to ultimate victory and how this will become visible by mid-year.
The players in this struggle are:
ARM Holdings (ARMH)
What is at stake:
Including personal hardware, service providers, data center operators, internet commerce, advertising, and all other things adjacent to or dependent upon the mobile network, we are looking at a multi-trillion dollars business worldwide.
Voice and/or data terminals involved:
Convertible notebook PCs
Large format tablet computers
Small format tablet computers
The most valuable piece of these terminals will be silicon-based components for application processing, graphics processing, camera and image processing, GPS, Bluetooth, WiFi, communications modem, audio codec, NFC (Near field communications), program memory (NAND), and DRAM.
Either now or in the near future virtually all of these functions except memory can be built on a comprehensive "Smartphone on a chip" or as the industry calls them SoCs (System on a Chip). The same kind of device will serve for tablet computers. The PC and convertible PC segment are likely to use very powerful CPU/GPU chips from Intel supported by a SoC chip that contains the functions not included in the CPU/GPU.
The PC segment is mature with a near monopoly on the major components by Intel. With that in mind, the battle will be around tablets and smartphones.
The silicon content will become about 50% of the total cost of the bill of material for these mobile devices, split between logic and memory, so we should look at the semiconductor manufacturers to determine who is in a position to win that 50% of the value of mobile devices.
At this time, Samsung is the world's largest supplier of cell phones, smart and otherwise, and the only vertically integrated manufacturer of mobile devices. Samsung uses their own processors in most of the devices they make. In addition to the Samsung captive business, Samsung provides the semiconductor foundry product for the Apple "A" chips. In addition to this, Samsung is the largest memory manufacturer in the world. Combined, Apple and internal consumption makes Samsung the undisputed world leader in the manufacture of smartphone application processors.
Once the immense capability of Samsung is understood, it is a simple step to appreciate that Samsung is a life/death competitor to every other name on the above list.
Apple has moved most of their component business away from Samsung in what can only be called an act of self-preservation. The only business left at Samsung is the Application Processor. Apple is involved in several lawsuits with Samsung, thus the sense of urgency in replacing Samsung as a supplier to Apple.
Google has developed the open source Android mobile operating system. This system is available free to smartphone manufacturers and is the most widely used mobile operating system in use today. Google and Apple have a significant conflict underway with Apple claiming the Android OS is a rip-off of the Apple iOS operating system.
Android based smartphones outsell Apple iPhones by a 3:1 ratio, driven primarily by Samsung. After using the free Android OS, Samsung is threatening to become a thorn in Google's side on advertising and other for-pay or licensed services.
Apple is also challenging Google in access to Google applications and the revenue sharing on internet search clicks.
In response to these actions, Google is frantically positioning themselves in the mobile device business through the Nexus family of devices and Motorola in the smartphone space. Google could become a formidable competitor in the mobile business because their need for a normal profit is not important next to the challenge to their search business and suite of mobile services. Google has an 18-month-old agreement with Intel to optimize the Android OS to work optimally with the Intel x86 processor architecture.
Qualcomm is the leading Application Processor and communications processor supplier to the non-Apple and non-Samsung mobile customers. From the Qualcomm viewpoint Samsung is a competitor in the sense that every Samsung phone shipped with a Samsung AP is revenue that Qualcomm will never see.
ARM Holdings is the prime intellectual property supplier to the mobile business. ARM supplies a low power microprocessor core (basically a design file in a CAD system) to the fabless designers such as Qualcomm and Samsung for use by the foundries like TSMC. That's it; that's all they do. The ARMH business model is an IP licensing and royalty model.
Intel is the world's largest semiconductor manufacturer by dollar volume. They are by far the largest supplier to the PC and server segment and are threatening to become a major supplier in the mobile business. Intel has the advantage that the PC CPU business has driven Intel's manufacturing technology to be the envy of all other semiconductor companies. Intel's technology represents the single threat to domination of the mobile business by Samsung.
At this point, Intel has not been successful in embedding this advanced technology in the mobile device business. Qualcomm is succeeding there due to their lead in communications technology. The hole in Qualcomm's armor is manufacturing. Qualcomm uses TSMC as a semiconductor foundry to build their products. The TSMC manufacturing technology is on a par with that of Samsung, but both Samsung and TSMC are two steps (or nodes) behind the leading edge Intel technology.
By the end of this year, Intel will have spent $35 billion on new manufacturing facilities to produce these leading edge technologies. The amount of Intel capacity is thought to be 2-3 times that needed to maintain their monopoly position in PC CPU chips. This means that half to two-thirds of the Intel capacity is idle at this time with no obvious short-term revenue stream to fill those factories.
Intel has a weakness in the communication part of the smartphone/tablet market and as such cannot supply processors into any smartphone that needs 4G LTE technology. The largest requirement for 4G LTE is for phones headed for the US market. Qualcomm has the communications technology but is hamstrung in manufacturing capability.
Intel will design the communication for 4G LTE and include it into their X86 based application processor. The problem is that by the time this happens, Samsung may have secured a sufficiently large market share in mobile as to freeze everyone else out of the business.
TSMC doesn't have a workable technology that is within a node of Intel. With TSMC looking at enough empty capacity at Intel to satisfy the entire needs of the mobile business, they might think twice about spending the $10-20 billion to construct the needed factories even if they had the required technology in hand.
The best case for Intel is to be able to demonstrate, short term, an x86 Application Processor and communications SoC product that is so compelling in speed and low power performance that all major smartphone and tablet manufacturers will abandon the ARM-based products and design the Intel SoC into all new products.
A short-term clue to the ability of Intel to supply such a device will come this Thursday when Xolo unveils a new Intel-based smartphone that is represented as the "fastest smartphone ever." If this device (Clover Trail+) is a true run-away on performance and battery life, the using community will begin to project performance of the upcoming 22nm Redhookbay chip coming from Intel later this year. I find it amusing that this announcement by Xolo is only hours before the scheduled unveiling of the new Samsung Galaxy S4.
Since nearly all existing AP suppliers, including Samsung and Qualcomm, use ARM cores their products would become obsolete overnight. Apple's A chips would also become obsolete overnight.
Intel has been working with Google for nearly two years to refine and tune the Android operation system to work more efficiently with the Intel x86 based products. Some think that the delay in Intel's introduction of mobile devices built on the leading 22nm Trigate process might be due to incorporating hardware optimization for Android onto the Intel future products as well as tweaking Android for x86.
If Intel goes it alone, and is successful, it means that Qualcomm's (and others) Application Processor business will decline precipitously. It means that Apple will have no choice but to use Intel designed x86 based application processors or they will not be competitive in the market. It also means that Samsung will have to abandon their in-house designs and use the Intel devices or they will not be competitive in the market. It will also mean that the ARMH stock price will return to $10.
All of these mobile devices will need various volume of DRAM and NAND memory. Since an Intel success of this nature is tantamount to nuclear war with Samsung, the latter would not be a very cooperative memory collaborator (in such things as stacked memory, etc.). Fortunately Micron will shortly be a similar size to Samsung in memory products of all types and Intel and Micron have a comprehensive joint venture on memory. Micron will be assured of a great deal of Intel-generated business for the foreseeable future.
In the event that Intel is not successful in their lone ranger approach, they still have $50 billion in annual capacity of the finest leading edge semiconductor technology in the world to be used as a massive foundry for Apple, Qualcomm, and even Samsung if they can't get beyond their present 32nm technology.
There are companies' futures and literally hundreds of billions of dollars involved with the final shakeout of the mobile supplier base. We shouldn't expect that anyone will telegraph their moves in advance. However, the plays here are so large that we should be able to discern a few hints.
The sheer magnitude of the Intel over-capacity sends a message that the company is very, very confident that they will have a use for all that output. With the Haswell shrink coming, they certainly don't need it for PC CPU chips.
TSMC talking about moving to a 20nm planar process gives a little hint that while not stopping process development altogether, they are taking a half step that is of little value. A company might be forgiven for that kind of conservatism when staring at $50 billion worth of unused Intel output that could be put to work to TSMC's distinct disadvantage.
I don't see any "tells" out of Qualcomm, but then the AP side of their business is not a big profit percentage relative to the royalty side, so maybe they are just letting things play out.
Nvidia has, out of frustration with TSMC, publicly called for Intel to act as a foundry for the industry. I guess that makes sense for him to say.
The conclusion is that Intel has a couple of ways into the mobile business. 1. The traditional approach using x86 processors in Intel proprietary mobile "smartphone on a chip" products that are so advanced as to ensure success. 2. Intel can turn the excess capacity into a giant semiconductor foundry available to all.
The logic of the situation is that if Intel were to offer foundry services to current competitors, there is no turning back. If they use the Lone Range approach and fail, the foundry option is still available.
My opinion is that Intel will go it alone with collaboration with Micron for the memory that is needed for all of these devices.
From an investment point of view, it seems that Intel has a high probability of success regardless of the path taken. In the same sense, Micron wins in either case.