Portable Devices: Parts is Parts
SEE LAST PAGE OF THIS REPORT Paul Sagawa / Artur Pylak
FOR IMPORTANT DISCLOSURES 203.901.1633 /.1634
sagawa@ / email@example.com
July 2, 2012
Portable Devices: Parts is Parts
- Driven by the ongoing paradigm shift from traditional PCs, feature phones and cable boxes, we expect the global market for portable devices, including smartphones, tablets and ultrabooks, to grow at a better than 22.9% CAGR through 2016. While this is an obvious boon to Apple, Google, and Microsoft, it is also a huge opportunity for component suppliers. Innovations in these technologies will drive the functionality and performance of devices, usually through incremental improvement, but occasionally via wholly new approaches that have, in the past, triggered significant shifts in the device market. We have divided the parts of future portable devices into 8 categories –processors, communications, displays, memory, sensors, power, materials, and miscellaneous. Of these, displays, sensors, processors, and communications are expected to see significant innovation and to grow as a % of the typical bill of materials, while memory prices are expected to decline much faster than device requirements are expected to rise, sharply dropping the % of the BOM.
- Smartphones, tablets and ultrabooks will grow at more than 22%/year, adding new innovations and addressing broader user needs. The TMT sector is in the midst of a once-a-generation paradigm shift that is remaking the way people access and use information. Portable devices, based on operating platforms from Apple, Google and Microsoft, tie user experiences to cloud-based resources across fast wireless networks and integrate them across devices and venues. We expect these devices to obsolete old paradigm platforms like PCs, feature phones and cable boxes, extending current functionality and adding richness, convenience, connectivity, ubiquity, and speed along the way. Support for workplace apps, commercial transactions, location based services, cross platform integration, media consumption, and unthought-of new uses will advance both incrementally and in leaps, with disparate market tiers, form factors and niches developing in what is currently a fairly homogenous market.
- Booming device demand will drive considerable opportunity for component suppliers. As the Internet continues to annex value from the rest of the economy, we expect platform owners to take a disproportionate share as they integrate the functionality of key apps into their platforms and exert control over 3rd party apps. However, these platforms rely on innovation in hardware technologies, creating a range of opportunities in 8 categories:
- Processor – ARM has become the de facto standard for its power efficiency, speed and wide chip vendor support. Intel faces a considerable uphill struggle for its alternative design, despite its process leadership, as device makers have embedded investment in the ARM architecture. Future processors will deliver more performance, in less space and with less power, likely integrating further communications and power management functions in the process. We expect the category to remain stable as a % of BOM.
- Communications – LTE is being slotted in many new frequencies. LTE Adv, a significant speed upgrade to 4G, will be available before mid-decade. NFC and WiFi Direct will enhance device-to-device connections. All of this will require more complex RF, more powerful baseband processors and more intricate antenna systems, although rising spending in these areas may be offset by integration of baseband functionality into processor-driven system-on-a-chip packages.
- Display – Today, LEDs offer better pixel density and reliability, but OLEDs are closing the gap and offer intriguing future possibilities like flexible displays, two-sided displays, exceptional thinness, and superior power efficiency. We expect the range of sizes, resolutions and price points to broaden to match developments in the devices themselves. The importance of the display in driving the user experience suggests spending in the category to outpace the total BOM.
- Memory – Old news: portable devices use FLASH, not hard disks, for storage. While on-board memory and FLASH have risen with successive device generations, increasing reliance on cloud-based processing and storage is likely to see RAM and FLASH capacities plateau or even decline going forward. Given price declines of more than 28% per year per GB, the revolution may pass by memory makers, with a long term downward trend as a % of BOM.
- Sensors – Top end smartphones come with GPS, 2 cameras, a microphone, an accelerometer, a gyroscope, a compass and a capacitive touch screen. Better image sensors will drive fine photography. Low latency, pressure sensitivity, haptic feedback and even keypad “buttons” that rise from flat glass as needed are around the corner for touch screens, which may get pressure from no-touch gesture and voice controls. New sensors measuring things like ambient light, proximity, barometric pressure, altitude, and heart rate could find their way into niche applications if not generally available devices. Trends suggest sensors will be the fastest growing part of the BOM.
- Power – Batteries aren’t getting any better, pressuring all other components to reduce their power needs. Some help can come from improvements in power management ICs, particularly as they are integrated into System-on-a-chip solutions. As such, power is unlikely to yield much savings for future device BOMs.
- Materials – Whether it’s Gorilla Glass, liquid metal or VaporMg, portable device manufacturers seek unusual casing materials to differentiate their products both aesthetically and functionally. The role of materials in BOM will vary wildly by product category and tier.
- Miscellaneous – Peripherals, such as keyboards, headsets, docking stations, charging solutions etc., will remain important, particularly for niche markets and applications. Mechanical items – hinges, buttons, etc. – tend to be idiosyncratic to individual models and a bigger piece of BOM for more complex devices like ultrabooks.
- These component opportunities are being addressed by established players and small entrants alike. Well positioned companies include: ARM, Qualcomm, and Nvidia in processors, Samsung, AU Optronics, LG Displays, Chimei Innolux and Toshiba in displays, Qualcomm, Avago, Skyworks, TriQuint, RFMD, TI and Broadcom in communications, Atmel, Cypress, TPK, Wintek, Qualcomm, Samsung, Sharp, Sony and others in sensors, Qualcomm, TI and Maxim in power management, Corning in materials, and Skullcandy in peripherals.
The massive paradigm shift that we believe is remaking the entire TMT landscape is driving a major shift from an older generation of devices, like PCs, cell phones, and cable boxes, toward portable devices supporting integrated user platforms that bridge across devices to tap resources resident on the Internet. The new generation devices, smartphones and tablets today, perhaps ultrabooks and televisions tomorrow, have sold in the hundreds of millions of units since their symbolic birth with the first iPhones in 2007, and are maintaining a strong growth trajectory that we expect to continue for many years (Exhibit 1). These future devices will improve upon the performance and functionality of today’s products and introduce new ones as they become viable. Stronger integration across devices including the living room television, better support for enterprise applications, secure mobile commerce, augmented reality to enrich location aware services, artificial intelligence to make user interfaces even easier and more intuitive – these and other advancements are on the near-term development roadmap.
Exh 1: Global Smartphone Mobile Device Production, 2012-2016
Portable devices will be a windfall for the companies that control the user platforms, which, unlike PC browsers, do not offer a neutral window to the web-based resources needed to address new applications, but rather dictate many consumer choices through functions integrated directly into the platform, and exert power full influence over most others via default apps shipped with devices and control of the app store. Through these mechanisms, the platform owners – Apple, Google, Microsoft and maybe, Amazon – are positioned to take the most valuable opportunities for themselves and to take tribute on apps offered by 3rd parties. While well differentiated web-based businesses with sustainable barriers to competition will share in the spoils, many software and web-services companies will suffer from the power influence of the platform owners.
To that end, we see hardware as a more fruitful arena for would-be platform ecosystem partners. While Apple remains tightly integrated to the hardware, and Microsoft and Google have been making moves toward vertical integration that leave device OEMs nervous, the component technologies that drive the performance, portability, connectivity, interactivity and, even, the aesthetic beauty of portable devices portend to profound growth and value creation looking forward. We have broken these component technologies into eight rough categories: Processor, Display, Communications, Memory, Sensors, Power, Materials, and Miscellaneous. In each category, there is both a trajectory of incremental improvement for existing component architectures, and in most cases, emerging alternative technologies likely to come into play as current architectures mature. Relative to the average bill of materials, we expect displays and sensors to show the greatest growth (Exhibit 2). Processors and communications will also likely grow as a percentage of BOM, but will see increasing integration of communications functions into the same package as the processor as part of cost and space saving System-on-a-chip solutions. Memory is likely to drop significantly as a part of total costs, as flash prices are projected to drop more than 28% a year going forward and the ability to leverage storage in the cloud is reducing the need for on board storage.
Exh 2: Component Composition of Typical Smartphone
Many companies are positioned against the component opportunities being created. Amongst large cap companies: ARM, Qualcomm and Nvidia lead in processors, Samsung, AU Optronics, Sharp and LG Display are ahead in displays; Qualcomm, Broadcom, Avago, and Skyworks provide important communications semiconductors; Samsung, Micron and SanDisk supply flash memory; Samsung, Sharp and Sony are leading sensor suppliers; Corning’s Gorilla Glass is a near standard for high end portable devices. Smaller public companies, like Universal Display, eMagin, TriQuint, RFMD, and OmniVision are vying with dozens of private companies for new seats at the table.
I Hear the Train a Comin’
We’ve made no secret of our perspective that the entire TMT landscape is in the midst of a comprehensive paradigm shift, with powerful integrated user platforms bridging a new generation of devices to resources in cloud-based distributed data centers via wireless broadband networks. The implications of this shift are profound for almost all corners of the technology industry, but nowhere more so than in devices. Not long ago, cell phones were just cell phones, cameras were just cameras, GPS was just being introduced into cars, the Internet required a PC, and television required a cable box. Arguably, not long ago means before 2007, when Apple introduced the iPhone and completely changed the way people thought about devices. In iOS and rival Android, all of these functions and many more were absorbed into an integrated smartphone package. Two years ago, the iPad extended the smartphone metaphor into an Internet and media centered tablet form factor, with the Android and Microsoft ecosystems following suit with a panoply of devices directed at both general and niche market applications.
Exh 3: U.S. Smartphone Market Share Trends – March 2009 – April 2012
Hundreds of millions of portable devices based on iOS and Android have been sold, yet we believe that it is the tip of the iceberg. Smartphones based on iOS, Android and Windows Phone are still only 28% of the global mobile phone market – we expect that to go much higher, particularly as Android (and perhaps Windows Phone) designs press the $100 unit price point to hit a global mass market that cannot afford an iPhone (Exhibit 3-4). In just two years on the market, the iPad is the fastest growing Apple product ever, hitting 67 million cumulative units sold in April, crushing consumer PC sales in the process. Microsoft’s recently announced Surface tablet appears squarely aimed at enterprises and their many millions of laptop toting employees. Amazon had a Yuletide hit with its media-focused and e-commerce subsidized Kindle Fire tablet, with rumors of a broader range of products built on the highly customized Android-based platform. Android partner Samsung is blurring the boundaries between Smartphone and Tablet with their popular 5” screen Galaxy Note. Google’s just announced Nexus 7 tablet puts media consumption front and center with high end specs in a 7 inch form factor at a $199 price point 60% lower than the cheapest iPad.
All told, we believe Gartner’s projections for 19.7% annual unit growth in smartphones, tablets and ultrabooks is conservative, with low-end penetration and PC displacement a much bigger factor than assumed. This suggests that the total annual sales of new paradigm portable devices could top 1 billion by 2014, and that the total installed base could touch a fifth of the world’s population in the same time frame. This is a huge market.
Exh 4: 2006 vs. Q1 2012 Mobile Device Market Share
Future devices will not just proliferate, but they will get better and cover broader ground. Processors will be faster, displays will be sharper, batteries will last longer, and connectivity will offer more throughput. Form factors will range from wearable devices, like the eyeglass-like Google Glass, through smartphones of various ilk and size, through tablets big and small, and on to keyboard equipped ultra-books. Target markets will dictate design and application emphasis, with general purpose devices sharing the market with units focused on specific purposes – e-readers, video-centric media, mobile workers, point-of-sale terminals, etc. Emerging applications will also drive device innovation. For example, near field communication (NFC) chips are becoming common with the availability of mobile wallet payment options at retail. Similarly, we expect futures like “augmented reality”, artificially intelligent user agents, cross platform media integration, and enterprise application support to exert meaningful demands on device functionality.
Who has the Hot Hand?
Bottom line: There is a lot of innovation going on in portable devices and we expect strong growth to continue as far as we can see. Of course, much of the benefit will accrue to the platform developers – Apple, Google, and Microsoft – who use the “app” model of internet access as means to influence users toward their own applications and services, or those of their favored (and typically, paying) partners. Platform owners are integrating new functions directly into their operating systems with each release, and spotting default apps onto prime screen real estate where they are very unlikely to be swapped out by many users. Even the stores through which users can add new apps to their smartphones and tablets are controlled by the platform owner – Apple even demands a triple tithe of all revenues generated via apps, including advertising, and restricts their ability to link directly to generic web sites. Given this looming platform domination of all things Internet, we see the best opportunities for others on the hardware side.
The phone makers themselves would SEEM a good place to start. Obviously Apple, which has fully integrated its devices with its software platform, generates enormous margins. Of the Google and Microsoft licensees, Samsung is the only company generating consistently attractive margins on its devices. While it is tempting to anoint them as fundamentally advantaged by scale and their leading portfolio of component businesses, we note that the history of cell phones is rife with companies that grabbed the market lead and generated outsized profitability, only to stumble badly when blindsided by design innovation by competitors. The progression is eerie – Ericsson’s mid-90’s success was eclipsed by Nokia and its small, reliable candy bar phones. Nokia’s 24% operating margins at the beginning of the millennium were crushed by the faddish Motorola RAZR, which accounted for 22% of all phones sold on earth in mid-decade (Exhibit 4). RAZR-mania was over by 2007, when the first iPhone was viewed as a niche product for fanboys and RIMM Blackberries were de rigeur for the serious professional. Five years later, RIMM is circling the drain, and Samsung is now viewed as a permanent number two to Apple’s permanent number one. While Apple’s broader position offering an integrated platform yields protection from competition on a purely hardware design angle, Samsung has no such shield. We suspect that share shifts amongst device makers will be no less a rollercoaster ride than it has been in the past.
Exh 5: Component manufacturers found in teardowns of major devices
Taking it Down a Level
While divining a long term winner at the device level is difficult, the circumstances for providers of component technology to the field are more consistently beneficial (Exhibit 5). Unit volumes will rise and the range of device categories will broaden. Importantly, current component architectures common in mobile devices will progress, but will meet challenges over time from alternative technologies offering step function improvement, entirely new capabilities and moving on a steeper cost improvement trajectories. Successful component suppliers to the portable device industry will lead on incremental improvement to existing solutions, but also participate in the development of the alternatives. To parse out the component opportunities in mobile devices, we have defined eight categories: processors, communications, displays, memory, power, sensors, materials, and miscellaneous parts. In each category, we assess the trajectory of current component solutions and identify the candidate technologies that could add to or displace those solutions.
Before the iPhone, cell phones and PCs were almost completely distinct in both use and architecture. From their inceptions, PCs were built on Intel’s x86 complex instruction set (CISC) processor, which, according to Moore’s law, doubled in speed and capacity with each generation, fueling ever more complicated software (Exhibit 6). Cell phones, which were intended to be pocket-portable and to run entirely on batteries, did not have the luxury of the power-hungry CISC approach, standardizing instead, upon a reduced instruction set (RISC) architecture designed by ARM holdings, and licensed for chips produced by a wide range of suppliers. By relying on a reduced set of computing actions, RISC processors can be implemented with far fewer transistors, taking much less physical space on a chip, and requiring much less power. While CISC architectures are much faster for very complex calculations, these are a very small part of the typical activity of a portable device and thus, the performance benefit was not deemed sufficient to justify the substantial penalty in on-board real estate and battery draw.
Exh 6: Moore’s Law – Processor Transistor Counts Over Time
Today’s portable devices, which at the high end compete directly with PC architecture notebook computers, have pushed the performance envelop of ARM’s RISC architecture very far forward, relying on lightning fast clock speeds and multiple cores to deliver serious computing horsepower despite the reduced computing instruction set. These devices are beginning to crowd out PC sales to consumers, and perhaps, will threaten the enterprise PC hegemony as well. Indeed, long-time Intel partner in crime Microsoft is supporting the ARM architecture with its latest version of Windows and sophisticated cloud-based businesses like Facebook and Google are experimenting with ARM based servers for their data centers.
Of course, Intel is fighting back. Perhaps the most sophisticated manufacturer of semiconductors in the world, Intel has a roughly 18-month lead over the Asian contract semiconductor fabricators that manufacture almost all of the ARM-based processors in moving to chip resolutions below 28 nanometers (Exhibit 7). This is a considerable advantage – shrinking chip resolutions are the basis for Moore’s law, allowing more transistors to be packed into the same real estate, drawing less power in the process. The move below 28nm is a big one, requiring a fundamentally different transistor design to avoid the electrical interference that occurs at the molecular level as the gates of a semiconductor get closer together. Intel has solved the problem and is moving to 22nm and on to 14nm, while the ARM world is, at least temporarily, stuck at 28nm. Because of this, Intel will be able to make CISC processors that draw dramatically less power than their previous generation solutions, giving them near term advantage vs. rivals.
Exh 7: Mobile Processor Roadmap Summaries by Vendor
The reality is that Intel has a long row to hoe. Of the hundreds of smartphones available on the global market, only a small handful of them use Intel chips and those products commanded an infinitesimal share of the 144.4M smartphones shipped to end users in 1Q2012 (Exhibit 8). While Google has pledged to support Intel architecture for its Android OS, the reality is that Google’s biggest device partners have customized Android with references to the hardware below, as have many of the best selling apps for the platform. Changing processor architecture would render all of this software incompatible, a serious hassle that leaves most OEMs inclined to wait for ARM manufacturers to catch up with Intel’s process advantage.
Moreover, moving to sub-18nm is not the only way to get to better performance and lower power draw. ARM has introduced a multi-core design called “big.LITTLE” that combines fast, powerful and more power hungry processor cores with small, slow and extremely power efficient partners (Exhibit 9). This approach allows simple tasks to be executed by the smaller cores, while the big brothers remain asleep. ARM estimates smartphone power savings of up to 70%, and even if this is a substantial overestimation for real world conditions, it goes a long way toward eliminating the disadvantage of remaining at 28nm. Nvidia has already introduced an ARM-based portable device processor with its own, similar, asymmetric multi-core design that has seen wide adoption in the latest generation of tablets, including Microsoft’s recently announced Surface and Google’s recently announced Nexus7.
Exh 8: Global Smartphone and Tablet Processor Revenue, 2010-2015
Exh 9: ARM big.LITTLE Dynamic Voltage and Frequency Scaling (DVFS)
As the rest of the world’s semiconductor manufacturers eventually make Intel’s leap below 28nm, the processing power available to mobile devices will continue ever upward, powering new applications, such as augmented reality, artificially intelligent user interfaces, facial recognition, and gesture controls. Communications functions, such as baseband signal processing, could be brought on board for cost, real estate and power friendly system-on-a-chip solutions. Projecting an increasingly ARM world going forward, we see companies like Qualcomm, Nvidia and Samsung as substantial beneficiaries.
Devices today connect using an alphabet soup of wireless standards – e.g. GSM/CDMA/UMTS/EV-DO/HSPA+/LTE cellular networks, WiFi a/b/g/n, Bluetooth, GPS, and NFC – each requiring fast coding and decoding from a baseband digital signal processor. These networks may fall across a wide array of radio frequency bands, all requiring antennas and radio transceivers attuned to specific frequencies to capture the signals and convert them into the digital formats expected by the baseband chip. The needs of portable devices may be simple, like a WiFi only tablet, or complicated, like a worldphone with support for multiple cellular standards in multiple frequency bands, along with WiFi, Bluetooth, GPS and NFC. These more complicated needs will drive the innovation in the market.
Exh 10: Roadmap for Communications components
Smartphone makers want better, faster, smaller and cheaper, impelling suppliers to integrate multiple communications components into system-on-a-chip packages (Exhibit 10). Where elements are fairly standard and stable across geographies, such as WiFi, Bluetooth, GPS or NFC, functions will likely be integrated right up into the processor. However, cellular baseband processors may remain discrete, to accommodate the distinctly different frequency and wireless standard needs of individual wireless carriers. Apple’s iPhone, for instance, has distinctly different versions with specific baseband solutions for AT&T, Verizon, and Sprint in the US. This fragmentation also affects RF transceivers and antennas, which must also match the frequency and standard requirements of the end device. For large carriers, scale may allow component makers to design customized solutions integrating RF and baseband to a single package or even a single silicon die, but these components would not be appropriate for carriers with different spectrum assignments or standards.
The next round of network deployments will muddy the waters further. 4G LTE licenses have been issued for many new frequencies around the world, complicating the task for RF and antenna design. The next generation LTE Advanced standard was ratified a year ago and is expected for commercial deployment before mid-decade. This new standard incorporates MIMO, a technique that uses multiple antennas within both device and base station to multiply the amount of information that can be sent over a specific frequency (Exhibit 11). This places significant pressure on antenna design, as the multiple antennas must be separated from one another while still adhering to antenna length imperatives dictated by the laws of physics. It also pressures baseband processors to manage the process by which the signals captured by the separate antennas are recombined and decoded. This complexity equates to opportunity for the suppliers of communications components. Semiconductor players positioned for this and for the ongoing growth in wirelessly connected portable devices include Qualcomm, Broadcom, MediaTek, TriQuint, Skyworks, and RF Micro. Thus far, antenna suppliers have been fairly fragmented and either private or tiny parts of much larger firms, but the sea change from LTE Advanced may bring new players to the fore.
Exh 11: MIMO- Multiple Input Multiple Output Defined
Device displays are the most obvious point of differentiation between portable devices for most users – speed, connectivity, storage, and other elements are much harder to judge in the store. Today, the display battle has two distinct sides: LCDs and AMOLEDs (Exhibit 12-13). The LCD (Liquid Crystal display) camp is led by Apple, with its stunning Retina displays, built to their specifications by Samsung and LG Display. LCDs are a fairly mature technology, but allow for very small pixels that can be very densely packed. This pixel density is behind the trademark clarity of the Apple displays, with potential that future versions might accomplish even more dots of light per square inch, although it is entirely possible that we have reached a point of seriously diminishing returns to further pixel density. There are, however, downsides to the LCD approach. First, LCD displays are inherently two layers thick, with a uniform white LED backlight necessary to illuminate the colored crystals in the front. This makes the display thicker, uses more battery power, limits the angle of viewing, and increases the effect of external glare. Because of the backlight, LCDs are also not really capable of truly inky blacks, reducing contrast in images.
Exh 12: OLED versus LCD Display Characteristics
The primary alternative, AMOLED (Active Matrix Organic Light Emitting Diode), is championed by Samsung, amongst others. In AMOLED displays, each pixel is comprised of sub-pixels which emit their own colored light, enabling extremely vibrant colors and true blacks with glare-resistant viewing from extremely wide angles. Without the need for backlighting, AMOLEDs are also inherently thinner and theoretically more power efficient. However, AMOLED is a less mature technology than LCD – its power efficiency and pixel density are competitive but far from the eventual maximum and AMOLED displays are still less reliable and more expensive. Longer term, as efficiency and density continue to improve, it seems inevitable that AMOLED technology will be the overwhelming choice for portable devices, given its inherent power, thinness and color richness advantages. We also note that AMOLEDs can be made in curved or even flexible shapes, allowing for future design creativity – rolling or folding displays, etc.
Exh 13: Mobile Phone Display Revenue Forecast
There are some further technologies worth mentioning. Inexpensive, very low power display architectures, such as Qualcomm’s Mirasol and Pixtronix’s DMS (both based on reflective MEMS technology), or E-Ink’s electronic paper displays, are finding play in specialized products – such as e-readers – and could improve to someday challenge AMOLEDs at higher quality levels. Texas Instrument has promoted pico-projectors based on its DLP technology that enable portable devices to project their display images to an external screen. These products have found modest support for enterprise applications.
We believe that AMOLED will become the primary display technology within the next several years, not just for portable devices but for larger format monitors and televisions. LCD technology will remain a viable alternative, but leading edge device makers, including Apple, will need to make the leap as AMOLED pixel density improves. Samsung is clear the leader in the technology, holding more than 90% global market share, with its well admired Super AMOLED Plus screens setting a standard in the current generation of portable devices (Exhibit 14). Korean rival LG Display is the second biggest current supplier of AMOLED displays, but just 1/10th the size of Samsung in the market. Taiwan’s AU Optronics and Chimei Innolux, and Japan’s Toshiba Mobile Display have also seen substantial growth in AMOLED display sales. We also note that there are several companies holding patents and/or supplying diodes and polymers to the AMOLED display makers that should benefit, including Universal Display, Dupont and others.
Exh 14: Status of Major Vendors of OLEDs
In 2005, Apple canceled the wildly popular hard-disk based iPod Mini in favor of the new Flash memory based iPod Nano, effectively marking the beginning of the end for hard disks in devices. With the rise of cloud-based applications and fast wireless networks, the need for carrying a couple of hundred gigabytes of information came into serious question, and with that, came the tablets and ultra-books, sans hard drives and sporting a healthy chunk of Flash. Thus far, each generation of devices has upped the ante on Flash, with music, pictures, videos, and games demanding space, but there is reason to believe that this trend will peak, and perhaps even decline, within a few years Portable device applications can tap functionally limitless storage resources on the Internet, with streaming music and video positioned to displace the storage thirsty download and play model, and image galleries automatically uploaded to the cloud. For many users, extra flash will prove to be an unnecessary extravagance. System memory needs may also plateau. Operating software for mobile devices is fairly lean and hasn’t shown the generation-to-generation bloat that pushed PC memory needs inexorably higher, and while device processors are getting faster and faster, their RAM needs are not following in lock step.
Despite strong growth in device demand, the math does not look promising for memory makers. The amount of storage that can be put on a single chip is moving inexorably higher, bringing the price per GB inexorably lower at a better than 28% annual pace (Exhibit 15). Combining this with flattening per device capacity needs is bad news for the likes of Samsung, Hynix and Micron.
Exh 15: Memory Consumption, Wireless Communication Devices
Portable devices pack a lot of different sensors. Typically, there is at least one camera, with a semiconductor image sensor, and often, a second for video calling purposes. There is also a capacitive touch screen controller to capture screen input. All smartphones and many tablets include microphones to pick up speech and other sounds. At the high end, there are compasses, digital accelerometers and gyroscopes to track the motion and physical disposition of the device. Recent innovations include ambient light and proximity sensors that cue changes to the touch screen to conserve power and guard against registering inadvertent touches, and even digital barometers for weather alerts. Future sensors could monitor heartbeats, detect smoke, rate air quality, provide biometric user authentication, or interpret facial expressions and physical gestures. Most of these sensors are inexpensive and discrete, but a few represent significant opportunities.
The touch screen controller is a fairly high cost item in the bill of materials, running between $12-20 depending on sizes and volumes (Exhibit 16). In portable devices, these controllers work on capacitive principles relying on a finger or other electrically resistive materials to disturb an electrostatic field and triangulating on its location. Future touchscreens will likely recognize the pressure used to give more nuance to touch gestures and reduce the number of inputs to accomplish actions. A proposed next-generation technology will measure the proximity of a finger, useful for “rolling over” links for a preview before clicking or other compound control gestures. NEC and Tactus have each developed ways to introduce changing texture to a touch screen, with Tactus proposing a “deformable tactile surface that creates dynamic physical buttons” that rise from the screen as needed and disappear when they are not. The leaders in the current generation of touch screen controllers are Atmel and Cypress, although would be challengers abound and could use innovation to quickly seize market share. In particular, non-touch gesture control, analogous to the Microsoft Kinect system, may appear in commercial smartphones as an adjunct to touch screens by 2013. One of the many start-ups pushing this innovation, Gesture Tek, was snapped up by Qualcomm in 2011, which has become an aggressive advocate for the technology.
Exh 16: Bill of materials breakdown of major devices
Cameras are the other big ticket item amongst sensors. The iPhone 4S’s two cameras – 8 megapixel on the back and 1.2 megapixel on the front – account for almost 10% of the total bill of materials according to iSuppli. Digital image sensor quality took a step function leap forward with the commercial introduction of back-illuminated CMOS sensors, which moved circuitry away from the light sensing elements, allowing a 50% increase in light captured (Exhibit 17). This technique was pioneered by Sony, followed closely by OmniVision, and these two companies remain the leaders in supplying high end image sensors for portable devices. Recently, Nokia introduced its PureView Pro imaging technology, implementing an unusually large 41Mpix image sensor with an integrated image processor. While the 1” footprint of this sensor requires modest compromises to the size of the device itself, the large format sensor allows the camera to over sample to support exceptional image quality, including logical zoom and low light performance, and the integrated processor manages image noise and speeds the process of picture taking. This innovation appears a bold step toward bringing the quality of integrated smartphone cameras into the realm of stand-alone products.
Exh 17: Portable Device Image Sensor Roadmap
Compared to most other portable device component categories, innovations in power management and batteries arrive at a snail’s pace (Exhibit 18). Lithium ion batteries, common to nearly all mobile devices, have been around since the early 1990’s with only modest, linear improvements to their ability to store energy. Lithium Polymer, available since 1995, added the option to mold batteries into customized shapes, an attribute that Apple has used to wedge ever larger batteries into its products. In Apple’s case, this is a blessing and a curse as it allows it to minimize physical size of each device, but makes it impractical to allow users to replace the batteries themselves should they want to install a freshly charged spare.
The glacial pace of improvement in the efficiency of batteries contrasts directly with the progress of bigger and brighter displays, faster processors and more sophisticated communications options, all of which would benefit from a bit more power. Unfortunately, there are no Star Trek “di-lithium crystals” here – all of the new power storage technologies of any promise appear to be more than a decade away from commercialization.
The result is that device makers have to make power tradeoffs. Using Apple as an example again, the new 2012 iPad has power hungry display using 4x as many pixels as the previous generation with faster processor and new 4G wireless technology. Including that display, the 4G connectivity and not wanting to compromise on battery life required finding room for a battery with 70% more capacity. As a result, the New iPad is thicker and heavier than its predecessor, an unusual step backward in the world of portability, but a necessary compromise to the reality of battery technology. Teardown analysis haven’t revealed battery vendors, but they are likely to be larger Japanese players including Sony, Toshiba, and Panasonic as well as lesser known Taiwanese companies such as Dynapack and Celxpert.
Exh 18: Major developments in rechargeable battery storage
Device makers can also leverage power management integrated circuits (PMICs) and software to control power consumption. PMICs tend to be custom made and incorporate many separate voltage regulators onto a single chip. Today however, most smartphones and tablets feature separate PMICs for applications and baseband processors. These are likely to converge onto a single chip and perhaps eventually even onto a single SoC (system on chip) design. Gartner estimates that PMIC revenue will increase from $1.1B in 2011 to $2.5B in 2016 and breaks out PMIC for both baseband and apps processors. We’re inclined to believe that further integration of functionality will lead to incremental energy savings based on current power constraints. Major PMIC players include App Processor makers such as Qualcomm and Texas Instruments as well as Intel which entered the space buying Infineon’s wireless assets. Dialog Semi out of Germany and Maxim also make PMICs.
From the interchangeable color plastic shells launched by Nokia in 1999, to the titanium casing around the iconic Motorola RAZR, to all-glass face of the iPhone, and the extraordinary stiffness of the thin magnesium skin of the recently announced Microsoft Surface, materials have played a major role in defining the image of portable devices. Apple’s yen for materials advantage has led them into arcane paths of metallurgy, with a recent exclusive intellectual property agreement with tiny Liquidmetal Technologies sending that stock to nearly 100 times its trailing 12 months sales. Unfortunately for investors, these choices seem to be idiosyncratic, often without obvious public company beneficiaries.
The one exception to this has been Corning with its unusually strong, light and scratch resistant Gorilla Glass (Exhibit 19). Long the leader in glass for large format televisions and monitors, Gorilla Glass has become near standard for higher end smartphones and tablets. With smartphone screen sizes growing and supporting bezels shrinking, we see no reason for Corning’s success in the portable device category to flag. Longer term, the potential for AMOLEDs to support flexible displays and the possibility of non-touch gesture controls could become threats.
Exh 19: Corning Specialty Materials Sales
The eight previous categories cover most of the value of the large majority of portable devices. The hodgepodge of boards, connectors, clips and wires that make up the rest of the average bill of materials is not particularly interesting to anyone and particularly investors. There are, however, peripheral and specialty components that could represent opportunities. For example, Square’s magnetic card reader turns portable devices into point-of-sale terminals and has quickly grown into a near standard for small merchants. Skullcandy and Beatz have demonstrated the market for high end headphones. A search on Amazon for iPad accessories yields 143,037 results – cases are the most popular category, but connectivity kits, card readers, external keyboards, charging solutions, headsets, and other electronic peripherals are well represented.
Exh 20: The Winners and Losers
Winners and Losers
Viewed at the component level, and excluding the impact on older generation devices, like PCs, feature phones and cable boxes, the explosive growth of portable devices is creating huge opportunities. Breaking it down into eight categories, we see strong upside for component players in processors, communications, displays, sensors, power management, materials, and peripherals. We are concerned that price declines in flash and RAM memory, combined with cloud-based storage and processing could yield declining value for memory products despite the unit growth. Companies that are well positioned include: ARM, Qualcomm, and Nvidia in processors, Samsung, AU Optronics, LG Displays, Chimei Innolux and Toshiba in displays, Qualcomm, Avago, Skyworks, TriQuint, RFMD, TI and Broadcom in communications, Atmel, Cypress, Qualcomm, Samsung, Sharp, Sony and others in sensors, Qualcomm, TI and Maxim in power management, Corning in materials, and Skullcandy in peripherals.