Summary: New analysis suggests that only only three or four mobile handset software platforms will remain by 2012.
This is a Guest Briefing from Arete Research, a Telco 2.0™ partner specialising in investment analysis.
The views in this article are not intended to constitute investment advice from Telco 2.0™ or STL Partners. We are reprinting Arete’s Analysis to give our customers some additional insight into how some Investors see the Telecoms Market.
Mobile Software Home Truths
Wireless Devices
Amidst all the swirl of excitement around mobile software, some dull realities are setting in. As the barn gets crowded with ever more exotic breeds (in alphabetical order: Android, Apple OSX, Blackberry, LiMo, Maemo, Moblin, Symbian, WebOS, WindowsMobile), there is a growing risk of fragmentation and consumer confusion. We see some unglamorous “home truths” about mobile software getting lost in the weeds.
Few, if any, vendors make money from mobile software. Microsoft makes $160 of gross profit per PC while mobile software is moving royalty-free. The few pure plays (like Opera) rely on sales of services around their software. Mobile software only gets leverage from related services (often a single one). These must be tightly linked to devices, e.g., e-mail (Blackberry), e-books (Kindle), music (iTunes) or gaming (XBoxLive), with resulting communities controlled by their choice of software; few services work equally well on all devices (e.g., search, YouTube).
AppStores are not (yet) content stores. OEMs must link themselves with cloud services (like Motorola’s new BLUR platform) or offer their own (e.g., ITunes, Ovi, etc.). Individual developers find it hard to make money through AppStores: if even one were making $10m in sales, it would be widely publicised. Exclusive or “sponsored” applications like navigation or content-like games should fare much better.
We see room for only three to four platforms by 2012. The pace of innovation, R&D cost, and need for customisation (for hardware, operators and languages) invites consolidation. Supporting OEMs and reaching out to developers is costly and labour-intensive; only over time might HTML5 browsers supplant device-specific applications. No platform is so productised as to simply hand over to licensees (be they OEMs or operators).
Every smartphone will support one (or more) AppStores. We do not know how many services or what content AppStores 2.0 might offer, or how they will be made relevant to consumers. The most popular applications should work on every smartphone, even as some devices (like INQ) are optimised for versions of Facebook, Amazon, Twitter, Skype and other popular digital brands and services. AppStores may help OEMs build relationships with users of those services, though both vendors and operators will try to control billing.
All phones are becoming smart. So-called smartphones get attention as a growth segment in a declining handset market, but “dumbphones” (using proprietary software like Nokia’s S40, Samsung SHP/TouchWiz or LG’s S-Class) are getting more sophisticated. The costs of the two are converging. Featurephones will soon also support AppStores and Internet services.
Table 1: Platform Penetration
|
|
’09E |
’11E |
|
|
Symb. v9.3+/S60 |
~110m |
~240m |
S60 goes mid-range |
|
Apple OSX |
~30m |
~120m |
Incl. iPod Touch |
|
B’berry OS 4.5+ |
~40m |
~80m |
Doubling OS base |
|
Android |
<10m |
~80m |
From >10 OEMs |
|
WinMo 6+ |
~10m |
~50m |
Transition to Win7 |
|
Palm WebOS |
<5m |
~15m |
Limits w/o licensing |
|
LiMo |
<5m |
~20m |
Platform for LCHs |
Hard Graft
The costs for developing and maintaining complex software platforms are increasing. There are no shortcuts to the sheer volume of work, especially in building on legacy code bases and supporting operator requirements, or developing language packs. Every platform faces significant roadmap issues. Some handset OEMs are building adaptation layers to port a range of applications to their own branded UIs. Just supporting multi-core chipsets for handling streaming or managing financial transactions needs additional processing power to deal with security and viruses. Yet it requires software re-writes and poses power management challenges (i.e., tripling or quadrupling processing will drain batteries faster).
We long predicted video would become as ubiquitous as voice, i.e., with devices designed around handling video traffic. There are a wide range of solutions to cope with streaming video, including in software (i.e., Flash or Silverlight) rather than via hardware optimisations. Apple patented technology around adaptive bit rate codecs to handle streaming in its forthcoming iPhones. All platforms need to support over the air (OTA) updates, embrace graphics-rich applications, handle HD content, and comply with an array of USB drivers and accessories.
It is also not clear whether application downloads are a novelty or a mass market phenomenon. Discovery and recommendation engines need to be improved on most platforms, and marketing must focus on what applications offer. The gap between legacy platforms and an over-the-air customisable user experience is a wide one, and will not be resolved by AppStores, fresh UIs, or moves to go open source. Widget and webkit technologies could bring similar UXs across multiple devices. Most developers will not need access to lower layers or optimise applications for specific hardware. Over time, HTML5 browsers could supplant device-specific applications (e.g., GMail runs on an iPhone as a web application, as does WebOutlook on Android), but OEMs are unlikely to embrace this approach. This also does nothing to extend billing or allow for collection of detailed customer analytics.
At the same time, operators’ selection criteria are moving from form factors to user experiences. Operator UX teams now number in the 100s of staff, even if they fake a fragmented approach: Vodafone-subsidised devices currently support Android Market, Blackberry AppsWorld, OviStore and iPhone AppStore, and runs its own developer programme (Betavine). Few telcos develop native applications, but mostly use ones that run in Java, Webkit, Widgets, etc. Only a few (e.g., Verizon Wireless) offer customised UI.
While Apple and Google get the most attention (as pioneers of the AppStore concept, and for providing a shop-front for the open source community), Nokia and Microsoft have pivotal roles to play. Both offer unprecedented scale (in handsets and computing software), even if both are fast followers. We do not see Nokia’s commitment to Ovi or Symbian wavering. Though Microsoft’s successive versions of WindowsMobile failed to get traction beyond 10-15m units p.a., we expect a renewed push around Windows7 in 2H10. The MSFT/Yahoo search deal could be a blueprint for closer collaboration with Nokia. With its resources (a $9.5bn R&D budget) and assets (enterprise installed base, XBox, HotMail, and Bing), Microsoft could offer handset OEMs revenue share deals. LGE already committed to ship 50+ Windows models by 2012.
Figure 1: Product Differentiation?
Source: Arete Research.
Content, Not Applications
An AppStore is not a content store, yet. The next battle will be to add intelligence and filtering to AppStores, and tightly integrate content with platforms (as with iTunes, Kindle, Zune HD, or Comes With Music). There are limits to how many applications consumers are likely to use, whereas there is a wide range of content to access via mobile devices. To handle this, mobile devices also need integration with home CE/PC products. Samsung, for one, aims to provide “three screen” offerings spanning TVs, PCs, cameras, and handsets. There will be efforts by Sony, Apple, Samsung and others to make a single harmonised software platform that spans a wide range of video-capable devices.
Figure 2: Putting Software at the Centre of a CE “User Experience”
|
Source: Arete Research. |
With multi-radio (e.g., 3G, WiFi and Bluetooth) integration and voice recognition, mobile devices could become a control point to reach “virtualised” content. This is a longer-term “cloud computing” angle to mobile software, handling access to and storage of personal content. OEMs will need to offer tight integration with cloud services, or offer their own “stores of content.”
Apple and Google designed platforms with PCs in mind, and drew developers from the vastly larger desktop world. They benefit from programming in AJAX, whereas Symbian uses a range of older object-oriented languages. Yet in both handset and PC worlds, OEMs, not developers, create devices. They are the gatekeepers for software and AppStores, managing the flow of any OTA updates that might alter the UX. Adobe has provided a good model, with regular updates of its popular Flash and Acrobat software. Yet user expectations of handset stability will get re-set if devices regularly need updates like PCs do.
Too Much Choice?
The number of companies vying to become the platform of choice is staggering, and itself a problem. Beyond the ones we discuss below, we can add Intel (with its Moblin effort), Palm’s WebOS (which remains device-specific) and the range of Linux variants (like the Nokia-sponsored Maemo, LiMO, and components developed under the OMTP). The latter shows how limited group initiatives have been: OMTP involves VOD, TMOB, TI, TEF, AT&T, and others, but all of these compete for exclusivity with operator-subsidised devices that will never be OMTP-compliant. None of the above options are yet mass market (i.e., likely to top 10m+ units in ’10). Just to confuse matters further, there are other applications environments (e.g., BREW) as well as “component” vendors like Opera, Access, and Adobe. We look at leading platforms below:
Apple’s OSX
Apple excelled at innovating around the UX and using animation to mask some of the iPhone’s early weaknesses (lack of multi-threading, slow image processing). Apple’s marketing anticipated the market’s direction with its focus on applications, and Apple’s PA Semi unit will help it be first to market with multi-core processing (supporting streaming video). Apple is still attracting developers with the clarity and simplicity of its SDK, and by testing and proving in each layer of stack via PC products. We expect OSX to be extended to CE products, and also for Apple to bring AppStores to the PC.
Google’s Android
For a two-year-old platform, Android got ample OEM support, following up its G1 (a.k.a. the Android Developer Phone) with subsequent releases Cupcake/Android v1.1, with the Éclair release being v2.0. Android aims to be binary forward compatible, i.e., existing applications written for G1s will run on new devices without modifications. Developers create Android Virtual Devices with the SDK to run applications for a range of devices. Development and emulator debug time is far shorter in Android compared with Symbian.
Despite OEM support, Android’s governance remains fuzzy. Android is open-sourced licensed, but not an open source project: a small (~300 staff) team controls the developer ecosystem and Android Market distribution. It has not productised source code or offered post-sales software management tools, and has limited support for operator-compliant packs, libraries of hardware drivers, and language variants. Some developers say Android is slow to respond to change requests and to accept code modifications. In exchange for access to Android Market, Google requires OEMs to bundle Google Apps and supply usage analytics from devices. One key commercialisation partner, WindRiver, was bought by Intel, while another, Teleca, started an Android Feature Club to resolve common integration issues. Android’s end-game is unclear: is it a hedge against Microsoft or Apple controlling end-devices? A Trojan Horse for Google services? Or will it become an independent company with license fees? If operators don’t need devices “with Google,” then Android may fragment into many custom UIs.
Nokia’s Symbian
After a decade under a shifting set of parents, the rump of Symbian was bought by Nokia and made an open source project, including Nokia’s own S60 UI. Symbian/S60 was initially developed for phone functions, and saw limited traction for downloads under cumbersome tree and branch menu structures. Many developers feel Nokia/Symbian offers too many choices (native Symbian code, J2ME, FlashLite, Web runtime and Python), each with limitations and compatibility issues. The S60 browser is based on webkit, but lacks HTML5 support. Nokia’s decision to open source Symbian/S60 has stalled its development, as Symbian re-writes and tests third-party software in its 40m line code base. It will be difficult to make major improvements to Symbian (i.e., to support multi-core processors) during this process.
When Nokia ships Direct UI in mid ’10, Symbian will effectively break its backwards compatibility. Whether it also moves to a completely new release (v.10 from v.9.6) is still open. This may alienate developers that have to re-develop for a new platform and comply with Nokia’s new Direct UI (based on QT). They also must resolve whether Symbian horizon is sufficient as a publishing tool, or if Nokia can get other OEMs to use OviStore, which still lags rivals on many fronts. Nokia hopes Symbian will present a credible alternative to Android in mid-2010 when it is fully open source/EPL licensed, with Nokia assuring a large market.
Microsoft’s Windows
Windows Mobile 6.5 traced a long evolution from the Pocket PC OS, but still uses an older WinCE 5 kernel. Microsoft recognised its failings by bringing in new management for Mobile, acquiring Danger (designers of the Sidekick device), and engaging LG as a mass market OEM alongside long-term supporter HTC. We see 6.5 as simply a stopgap solution until Microsoft brings the innovation seen with its ZuneHD UI and leaner Win7 platforms to mobile. Microsoft is also offering its software in a reference design called Pink, and may tweak its long-held license fee model with PC-like terms (rebates, discounts and marketing support). This may gain traction among Chinese OEMs, after Taiwanese and US OEMs failed to ramp WinMo to volume. It is too early to rule out a now-dormant Microsoft, given its scale in computing and revival with Win7.
RIM’s Blackberry OS
In a world moving more “open,” RIM keeps its OS development in-house, stressing the need for security and compression. Yet RIM must evolve the BlackBerry’s UI and bring more developers to its AppsWorld platform, as well as open up its charging model beyond PayPal to embrace operator billing. BlackBerry’s application environment works on a J2ME framework with proprietary extensions, which adds fragmentation and compatibility issues. However, the security and bandwidth compression so valued by enterprises may limit performance for consumers, as applications traverse its NOCs via RIM’s proprietary browser. RIM’s premium pricing still relies on its messaging franchise, which faces challenges from ActiveSync and efforts to bring push e-mail to mass market price levels. Rivals may not match Blackberry’s UX, but some segments may be less sensitive to RIM’s security and delivery than the price of handsets. While RIM stresses incremental upgrades for its AppsWorld, we hear they are undertaking an extensive OS re-write to support new multi-core chipsets.
Down to Earth
This space gets too much attention for the revenue it directly generates. Mobile software is a means to an end, and the end is selling devices and Internet services. The cost of development will narrow the number of platforms by 2013, but not before the sheer number of options bewilder consumers who know about them and frustrate others wishing to get simple access to specific content. Given how rapidly key hardware costs are falling, and how sophisticated mid-range software platforms are becoming, all phones will become smartphones of some sort. Who wants to own a dumbphone?
AppStores will evolve to offer a range of content and services, with a major battle brewing over billing and data on consumer usage. Every device will support some AppStores and work with a range of Internet brands and services. Some content will be packaged and tightly linked to specific devices. The Holy Grail in all this mobile software will be its extension to ranges of other CE products. There is ample reason to scoff at the hype around mobile software — for its marginal economics and inevitable fragmentation — but no doubt as to its future role as a control point for more valuable content and Internet-based services and brands.
IMPORTANT DISCLOSURES
For important disclosure information regarding the companies in this report, please call +44 (0)207 959 1300, or send an email to michael.pizzi@arete.net.
This publication was produced by Arete Research Services LLP (“Arete”) and is distributed in the US by Arete Research, LLC (“Arete LLC”).
Arete’s Rating System. Long (L), Neutral (N), Short (S). Analysts recommend stocks as Long or Short for inclusion in Arete Best Ideas, a monthly publication consisting of the firm’s highest conviction recommendations. Being assigned a Long or Short rating is determined by a stock’s absolute return potential and other factors, which may include share liquidity, debt refinancing, estimate risk, economic outlook of principal countries of operation, or other company or industry considerations. Any stock not assigned a Long or Short rating for inclusion in Arete Best Ideas is deemed to be Neutral. A stock’s return potential represents the difference between the current stock price and the target price.
Arete’s Recommendation Distribution. As of 30 June 2009, research analysts at Arete have recommended 16.9% of issuers covered with Long (Buy) ratings, 21.1% with Short (Sell) ratings, with the remaining 62.0% (which are not included in Arete Best Ideas) deemed Neutral. A list of all stocks in each coverage group can be found at www.arete.net.
Required Disclosures. Analyst Certification: the research analyst(s) whose name(s) appear(s) on the front cover of this report certify that: all of the views expressed in this report accurately reflect their personal views about the subject company or companies and its or their securities, and that no part of their compensation was, is, or will be, directly or indirectly, related to the specific recommendations or views expressed in this report.
Research Disclosures. Arete Research Services LLP (“Arete”) provides investment advice for eligible counterparties and professional clients. Arete receives no compensation from the companies its analysts cover, does no investment banking, market making, money management or proprietary trading, derives no compensation from these activities and will not engage in these activities or receive compensation for these activities in the future. Arete’s analysts are based in London, authorized and regulated by the UK’s Financial Services Authority (“FSA”); they are not registered as research analysts with FINRA. Additionally, Arete’s analysts are not associated persons and therefore are not subject to Rule 2711 restrictions on communications with a subject company, public appearances and trading securities held by a research analyst account. Arete restricts the distribution of its research services to approved persons only.
Reports are prepared for non-private customers using sources believed to be wholly reliable and accurate but which cannot be warranted as to accuracy or completeness. Opinions held are subject to change without prior notice. No Arete director, employee or representative accepts liability for any loss arising from the use of any advice provided. Please see www.arete.net for details of any interests held by Arete representatives in securities discussed and for our conflicts of interest policy.
© Arete Research Services LLP 2009. All rights reserved. No part of this report may be reproduced or distributed in any manner without Arete’s written permission. Arete specifically prohibits the re-distribution of this report and accepts no liability for the actions of third parties in this respect.
Arete Research Services LLP, 27 St John’s Lane, London, EC1M 4BU, Tel: +44 (0)20 7959 1300
Registered in England: Number OC303210
Registered Office: Fairfax House, 15 Fulwood Place, London WC1V 6AY
Arete Research Services LLP is authorized and regulated by the Financial Services Authority
US Distribution Disclosures. Distribution in the United States is through Arete Research, LLC (“Arete LLC”), a wholly owned subsidiary of Arete, registered as a broker-dealer with the Financial Industry Regulatory Authority (FINRA). Arete LLC is registered for the purpose of distributing third-party research. It employs no analysts and conducts no equity research. Additionally, Arete LLC conducts no investment banking, market making, money management or proprietary trading, derives no compensation from these activities and will not engage in these activities or receive compensation for these activities in the future. Arete LLC accepts responsibility for the content of this report.
Section 28(e) Safe Harbor. Arete LLC has entered into commission sharing agreements with a number of broker-dealers pursuant to which Arete LLC is involved in “effecting” trades on behalf of its clients by agreeing with the other broker-dealer that Arete LLC will monitor and respond to customer comments concerning the trading process, which is one of the four minimum functions listed by the Securities and Exchange Commission in its latest guidance on client commission practices under Section 28(e). Arete LLC encourages its clients to contact Anthony W. Graziano, III (+1 617 357 4800 or anthony.graziano@arete.net) with any comments or concerns they may have concerning the trading process.
Arete Research LLC, 3 Post Office Square, 7th Floor, Boston, MA 02109, Tel: +1 617 357 4800