In the past 18 months, mobile phone manufacturers in Asia and elsewhere have introduced over a dozen handset models based on Linux, and before the end of 2005, you can expect to see a dozen more smart and feature phones announced and shipping. While it's easy to gush over this emerging trend, and to wax eloquent over the technical particulars of these intelligent mobile devices, it may be more interesting to examine the drivers behind this wave of adoption.

This article examines the economics of building next-generation mobile handsets with Linux. In particular, it explains how phone makers have come to view Linux and Open Source not just as software components in their wares, but as a strategic platform that streamlines their business, improves their margins, and protects their brand equity. It will illustrate how Linux helps phone OEMs manage ever-increasing software content and complexity in mobile devices, meet emerging technical requirements from carriers and consumers, and helps them deliver their wares through fiercely competitive channels.

Facing Phone Fragmentation
Mobile phone manufacturers - from recognizable top branded players to small, unbranded suppliers in emerging markets - face a common set of product line challenges. They must develop, deploy, and support a range of handheld products that target the requirements and market niches they serve. At the least, they must offer entry-level and full-featured devices; to differentiate or to get higher margins, they often must also offer a high-end deluxe model. They may have to serve diverse geographical markets with different languages or wireless network standards. Their product lines may include inherited and acquired technologies. In extreme cases, each device they make may employ a unique processor, different wireless and multimedia chipsets, and incompatible operating systems, middleware and applications. Developing and deploying such heterogeneous product lines can entail hiring, training, and maintaining separate engineering and support teams for each line. And, in a market faced with slim margins, lack of commonality puts further strain on profitability by denying mobile OEMs economies of scale in their supply chains and production lines.

Under pressure to simplify and unify, why haven't mobile device OEMs already found a golden mean, a common platform, even in their own organizations? To answer that question, let's examine how each tier of handheld devices is built and goes to market:

• Entry-level - So-called "throw-away phones" dominate the market in volume. They typically offer monochrome or low color-depth displays and adequate, if unexciting, feature sets. Some phone models arrive as entry-level devices by design; others end up here at end-of-life from higher tiers. Traditionally, entry-level phones use low-end ARM processors (MMU-less ARM7 or older CPUs) running in-house OSes and stacks, or COTS RTOSes like Mentor Graphics' Nucleus, or legacy kernels like VRTX or pSOS. A phone maker once explained that these device designs target a BOM (Bill of Materials) of $10-$20. They end up in consumer hands for under $100, or even for free at end-of-life.
• Full-featured - These "feature phones" have increased in share and volume by offering more on-line services (e-mail, web, and messaging), an enhanced multimedia experience with medium-resolution color graphics and polyphonic ring tones, and in the last 18 months, cameras of ever increasing resolution. These devices differentiate themselves in terms of battery life, signal and voice quality, and software look-and-feel. The majority of these devices are built on ARM CPUs as well and run a mix of in-house software and commercial OSes like SymbianOS. Target BOM is US$40-$60 and MSRP ranges from US$100-$250.
• Deluxe "Smart Phones" - There are the must-haves not just among executives, but among the status-conscious consumers across Asia. These phones are like "little desktop" computers fitted with rich productivity software stacks and offering a high-resolution multimedia experience. They go beyond feature phones in style and performance and integrate ARM9 CPU cores together with dedicated multimedia chips, DPSs, and encapsulated wireless modems for GPRS and CDMA, BlueTooth, and Wi-Fi. Their software stacks are rich and deep and build on platforms including PalmOS, SymbianOS, WindowsCE, and increasingly Linux. A target BOM of $100-$150 reaches consumers in the $300-$400 range.

Why are mobile handset OEMs choosing Linux over SymbianOS, WindowsCE, and PalmOS? Multiple motives are driving them to Linux and Open Source:

Total Costs
In mobile communications as elsewhere there is a tendency to over-emphasize the benefit of "free software" in the choice of Linux. Cost certainly is compelling, especially when OEMs anticipate deployment volumes in millions and profit margins in pennies. Many phones actually ship with negative margin, achieving profitability by amortizing device cost over multi-year service commitments. OEMs turn to Linux and to Open Source to improve their margins by reducing or even eliminating the BOM burden of royalties. Cost is also key when proprietary platforms like SymbianOS cause device OEMs to incur steep start-up costs, upwards of US$250,000 for licensing, tools and source code. Instead of eating up large portions of R&D budget before a phone is even reaches prototype stage, OEMs turn to Linux and Open Source.

Brand Equity
From a purely technical perspective, phone-targeted versions of Microsoft Windows present the most complete hand-held platforms, replete with underlying OS, tools, middleware, GUI, multimedia CODECs, power management, and other features. Then why have so few mobile device OEMs gone to market with WindowsCE family platforms? Because any phone deployed with a Windows platform becomes a Microsoft phone! According to published studies from Interbrands, Redmond enjoys global brand equity only second to Coca Cola, and a brand position severalfold stronger than companies like Nokia and Motorola. The Microsoft brand and Microsoft branding requirements eclipse the name and equity of even top-tier handset OEMs, let alone emerging Chinese brands like Datang, Huawei, and ZTE. Linux, by contrast, while a strong brand, is non-threatening (IDC figures for all Linux-related revenues put the Linux brand in the top 20). Mobile device OEMs can leverage the Linux brand or obscure it at will, and still enjoy Linux technical and economic benefits.