Last week we published our smartphone applications processor market share estimates for 2010. By our estimates, the smartphone applications processor market showed an impressive 77 year-over-year revenue growth in 2010 to reach $4.52 billion. Baseband-integrated applications processors accounted for 74 percent of total apps processor revenue in 2010 and we think it will remain relatively stable in the next few years.

Here are some highlights of our smartphone applications processor market share tracker 2010 report:

• Qualcomm led the smartphone applications processor market in 2010 with 41 percent revenue share on the strength of its strong participation in the multiple software ecosystems and its broad range of integrated processors. Qualcomm’s Snapdragon processors MSM8x50 and MSM7x30 gained good traction in the second half of 2010 and featured in multiple popular smartphones. We note that Qualcomm had over 50 percent revenue share lead over its nearest competitor Texas Instruments (TI). We believe that Qualcomm has a two-to-three year advantage over most of its competitors in terms of integration and is well positioned to move further ahead of its competitors with its upcoming MSM8960 LTE apps processor.

• TI ranked number two with 27 percent revenue share in 2010 despite topped the unit shipment rankings. TI’s stand-alone apps processor unit shipment share declined to 20 percent in 2010, by our estimates. We think TI's unit shipment leadership will be challenged by Qualcomm and Samsung in 2010.

• Samsung finished with 19 percent revenue share in 2010 and its unit shipments doubled on a year-over-year basis. Samsung's 1GHz Cortex-A8 Hummingbird apps processors featured in high-volume products such as Samsung Galaxy S, Wave and Wave II.

•  Marvell ranked number four in revenue terms but the company's over-dependency on a single customer is a cause for concern.

With over ten vendors already announced 1GHz dual-core Cortex-A9 processors, we expect 2011 to be highly competitive and ARM-based applications processor vendors should build competitive barriers by investing in graphics processing technologies, integration, micro architecture, multi-core processors, semiconductor process technologies and software expertise as the entry barrier is relatively low for new vendors to enter into the ARM-based ecosystem compared to the x86 chip ecosystem.

Client reading: Smartphone Apps Processor Market Share: Qualcomm’s Revenue Share Reached 41 Percent in 2010

Multi-core processors are the name of the game in high-end smartphones and tablets in 2011 and are increasingly becoming an essential marketing and technology strategy for every handset vendor and silicon vendor. We estimate multi-core processor penetration in smartphones will rise from 15 percent in 2011 to 45 percent in 2015, and that ARM's Cortex-A9, Cortex-A5 and Cortex-A15 and Intel's x86 architectures will fuel the multi-core processor growth in smartphones over the next few years. 

In 2010, over 45 percent of Android smartphones shipped with a 1GHz or higher processor, hitting the ceiling on what is possible with a single core processor. Adding a second core is the logical next step to boosting the speed without raising the power consumption in handheld devices such as smartphones.

Samsung, Qualcomm, NVIDIA, ST-Ericsson, Texas Instruments, Marvell, Renesas and Broadcom are well positioned to capture share in the smartphone multi-core processor market. Freescale recently announced its i.MX 6 processor family but the company has less visibility when it comes to smartphone design-wins.

Having largely won the single core smartphone apps processor market, baseband-integrated vendors are now faced with a multi-core processor challenge. We believe the smartphone multi-core processor market will be dominated by stand-alone apps processor vendors in 2011. By our estimates, baseband-integrated multi-core apps processor will account for about one third of total smartphone multi-core apps processor shipments in 2011 with the potential for higher share in future.

Sravan Kundojjala

Microsoft announced at CES-2011 that its next generation Windows PC platform will run on ARM architectures. NVIDIA, Qualcomm and Texas Instruments were named as key ARM silicon partners. We think this is good news for stand-alone ARM applications processor vendors such as Texas Instruments (TI) and throws the company a lifeline at a time when the market is moving towards integrated solutions.

By our estimates, the market for stand-alone applications processors in smartphones contracted from 70 percent in 2007 to just under 30 percent in 2009. Vendors such as Qualcomm are driving the baseband-integrated applications processor market and these integrated processors are increasingly gaining acceptance in the smartphone market. The integrated approach is currently popular in the low-to-mid range smartphones but is gaining traction in the high-end as well lately.

Microsoft's decision to port the Windows PC platform to ARM architectures changes the market dynamics dramatically. However, it's too early to tell what impact it will have on Intel as Microsoft hasn't revealed any timeline for these new products. ARM hasn't yet revealed its 64-bit architecture and current generation ARM processors also cannot support graphics computing APIs such as OpenCL. It will take at least another 18-24 months for Microsoft to reengineer its platform for ARM architectures and to build a software ecosystem around it. This should give enough time for Intel to bring power-efficient processors to the market. Intel is expected to feature in Nokia's MeeGo smartphones and tablets in 2012 which we think would be a great psychological boost to counter ARM-based processor vendors.

The likely specs for Windows on ARM architectures are:

• CPU: Dual-core / Quad-core ARM Cortex-A9; GPU: ARM Mali-400 / Third-party GPU

• CPU: Multi-core ARM Cortex-A15; GPU: ARM Mali T-604 / Third-party GPU (Most likely configuration and leading chip vendors such as TI, NVIDIA and Samsung were already announced as licensees of Cortex-A15 cores)

• Custom ARM processor designed by Microsoft and its hardware partners

This move is a good news for NVIDIA, Qualcomm and TI as it opens up new opportunities. Qualcomm already has customer relationships with top PC vendors and is expected to leverage them to expand into the Windows PC ecosystem. This move equally helps the stand-alone applications processor vendors to revitalize their businesses and to compete for higher volume. Broadcom, ST-Ericsson, Samsung, Renesas, Marvell and Freescale could also potentially feature as future Microsoft's partners.

We believe diversified device and OS landscape favors chip vendors as they can avoid the risk of being sidelined by powerful vendors. With Microsoft's support, ARM chipset vendors can now look forward to new markets that were unavailable to them previously.

Consolidation in the baseband market continued with Intel acquiring Infineon's wireless business on Monday for $1.4 billion. Infineon's takeover is the fourth high-profile announcement in the last two years. Previously NXP, Freescale and Texas Instruments announced their exits from the merchant baseband market. This signals that the entry barrier is now too high for new entrants and raises pressure on smaller players such as Marvell, Spreadtrum and Icera. Intel's acquisition of Infineon's wireless business leaves Qualcomm, MediaTek, ST-Ericsson, Intel, Broadcom and Renesas as long-term players in the cellular baseband market.

We analyzed this back in May when the original rumour of the Intel/Infineon wireless deal surfaced. Infineon's wireless business is definitely an attractive target for Intel as it gives immediate access to valuable IP and an established customer base. Infineon is a rising star in the W-CDMA baseband market and ranked number two in the W-CDMA baseband market behind Qualcomm (excluding Texas Instruments’ custom baseband business with Nokia). Infineon has so far focused on slim modems and has stayed away from developing ARM-based applications processors. So far Infineon’s niche strategy has paid off, but the time has come for the company to expand its portfolio by offering a wider range of processors to address smartphones, tablets and other connected consumer devices. Although 3G and smartphones are of most interest to Intel, the company is likely to keep Infineon’s 2G business as it provides scale which is crucial to play in the cellular baseband segment.

It will be another year or two before Intel's applications processors are smartphone-ready. Meanwhile, in the next 12 - 18 months Intel can bundle Infineon's 3G platform with all of its notebook and netbook chipsets and gain significant 3G share to undercut Qualcomm and ST-Ericsson's non-handset offerings. But both Qualcomm and ST-Ericsson appear to have a two year technology lead over Infineon in chipsets for HSPA+ and LTE; Infineon is expected to introduce its first HSPA+ chipset in 2011 and an LTE chipset in 2012. MediaTek’s grey handset chip business may come under pressure from Infineon in the future if Intel expands its field application engineering force in China. Likewise, Broadcom is expected to come under pressure from this deal but should survive with help from its other businesses such as connectivity and digital home. This acquisition will also likely affect stand-alone application processor vendors such as Texas Instruments and NVIDIA that currently lack their own 3G / 4G modems. But we believe that it will take another 4-5 years for Intel to integrate basebands into its Atom CPU, if at all.

Infineon's acquisition still doesn’t guarantee Intel's success in the wireless handset market and the company's ultimate success in this market hinges on how soon Intel can come up with power-efficient processors that can stand-up and compete against a host of ARM-based processor vendors. On a positive note, Intel appears to be moving ahead of its other chip competitors in the software area which positions it well against ARM-based processor vendors. The company’s strong software expertise could help it to gain some OEM mindshare. Intel’s uninspiring track record in the wireless market during the last decade suggests that the company has to execute flawlessly this time to make it successful and keep Infineon’s wireless business as independent as possible. Clients can read our eight page analysis of the Intel/Infineon deal here.

Intel is said to be in talks with Infineon to buy latter's wireless chip business unit. Of course this not the first time we heard about Infineon's wireless chip division rumors. Previously Samsung, NXP and ST Micro were also reported as potential buyers of Infineon's wireless chip business.

Previously Intel invested in the wireless handset chip business but divested it to Marvell. That time Intel was more focused on the application processor and was less focused on the baseband processor. So what's changed? Strategy Analytics believes that the growth in the mobile broadband is a key factor and Intel definitely needs a modem know-how to participate in this market. Intel is a key player in the WiMax market but lacks 3G and LTE baseband technology. Currently the company ships 3rd party modems in its Atom chipsets but in the long-term it would be beneficial to own modem technology to integrate it in its future Atom CPUs. Last year Intel acquired Freescale's France-based wireless operations and also licensed 3G / HSPA modem technology from Nokia.

Strategy Analytics believes that Infineon could be a good partner to companies looking for RF expertise and baseband customer relationships. Infineon's wireless business looks well settled after the company divested its Qimonda stake.

Here are some quick facts about Infineon's wireless business based on Strategy Analytics data.

* Infineon's wireless chip revenues represented close to 30% of its total revenues in calendar year 2009 and 25 percent in Q1-2010.

* Infineon's wireless chip products include basebands, RF transceivers, power management ICs, connectivity ICs (Bluetooth, GPS, and WLAN) and platform solutions.

* Infineon's baseband revenues represented close to 53% of its wireless chip revenues in 2009. * The company has baseband / RF customer relationships with almost all of the top-10 handset OEMs.

* Infineon ranked fourth with 11% unit shipment share in the $11.0 billion cellular baseband market in 2009.

* Infineon's wireless business has been profitable for the last four consecutive quarters. The company had just 2.5% operating margin in its wireless business in 2009.

* Despite its impressive baseband and RF product portfolio, the company currently lacks stand-alone application processors. It remains to be seen whether Infineon will go it alone or partner with Intel. We believe that it would cost $3-$5 billion for Intel to acquire Infineon's wireless business.

- Sravan Kundojjala

The handset chip market has gone through some rapid changes over the past 3 years and only a handful of the remaining players have long-term viability and sustainable profitability. We estimate that the total cellular baseband revenues in 2008 were $11.15 billion and $11.04 billion in 2009 (provisional).

Some of the highlights of Strategy Analytics' Quarterly Baseband Share Tracker for Q3 2009,

* Qualcomm was the market leader with 38 % revenue share and 19 % unit share;

* MediaTek was second only to Qualcomm in terms of unit and revenue share;

* Broadcom and Infineon are improving market share and expanding at their respective key customers. We estimate Broadcom's cellular baseband revenues grew 350 % in 2009. However, Broadcom is still a minor player in the cellular baseband market but is expected to be an important player in 2010;

* ST-Ericsson ranked fourth with close to 10 % baseband revenue market share;

* Texas Instruments' baseband revenues were down almost 33 % on a Y-o-Y basis in Q3 2009.

With Texas Instruments and Freescale exiting from the cellular baseband market, Strategy Analytics believes that Qualcomm, ST-Ericsson, MediaTek, Infineon and Broadcom are well positioned to remain long-term players. However, new entrants are likely to appear as demand increases for LTE and TD-SCDMA chips.

- Sravan Kundojjala

Strategy Analytics feels Marvell’s lack of exposure to high volume handset makers, lack of a visible long-term baseband product roadmap and missing technology pieces like RF transceivers will force the company to reconsider its position as a cellular baseband processor supplier. Marvell is a prominent player in application processor and connectivity market but a minor player in the cellular baseband market. Since acquiring Intel’s cellular business in 2006, Marvell has made limited progress in the cellular baseband processor market where as competitors Infineon, Qualcomm, and ST-Ericsson built scale during the same period. The company's total cellular revenues (including basebands, application processors, connectivity) are still less than one billion dollar.

Research in Motion (RIM) is Marvell’s biggest cellular baseband customer but Marvell is expected to lose to share to Qualcomm in the long-term. Qualcomm already supplies basebands to over 25 percent of the RIM’s handsets (mostly CDMA). Freescale also recently scored some baseband design-wins at RIM at the expense of Marvell although Freescale recently gave up its baseband efforts. Marvell’s other cellular baseband customers include Palm, Asustek, and some Asian ODMs. Most recently the company scored Ophone TD-SCDMA Android smartphone design-wins at China Mobile and we believe that most of the TD-SCDMA baseband IP belongs to China Mobile and affiliates. Marvell's competitors MediaTek, ST-Ericsson are also investing in TD-SCDMA and most recently Qualcomm also expressed its intention to compete for TD-SCDMA design-wins in China next year. We don't see how a niche market strategy can be successful for this R&D intensive technology.

The company's lack of RF transceiver products is likely to have an impact on its penetration in entry level and mid market handset segments where transceiver-baseband approach is appropriate. The lack of visible LTE roadmap could put Marvell behind its competitors Qualcomm, ST-Ericsson and Infineon. We believe that growing R&D costs to develop advanced communication IP could put pressure on Marvell to reconsider its long-term position as a baseband vendor.

We feel that Marvell could perhaps partner with companies like MediaTek or Infineon who lack high-end application processor technology. Marvell's application processors are already popular in the Chinese smartphone market and could be a good fit for MediaTek. However it is not clear whether these companies are interested in Marvell's XScale technology.

Sravan Kundojjala

Client reading: Baseband Processor Profile: Time for Marvell to Reconsider Its Baseband Position