This month at Computex in Taiwan, Qualcomm announced that it has begun sampling its first Snapdragon chipsets, which integrate two Scorpion application processor cores (based on ARM's Cortex-A8 ISA) along with embedded modem cores. The application processors, fabricated in 45nm CMOS, are capable of running to at least 1.2 GHz and are targeted at high-end smartphones. Qualcomm’s first-generation (1 GHz) Snapdragon chipsets are currently shipping in smartphones by Toshiba (TG01), HTC (Evo), and Google (Nexus One, also made by HTC). Two Snapdragon versions will be available, differing only in their embedded modems. The MSM8260 is for HSPA+ networks and the MSM8660 is for multimode HSPA+/CDMA 1xEV-DO Rev B. networks. Both versions boast more powerful GPUs with 3D acceleration engines for Open GLES 2.0 and Open VG 1.1, 1080p video encode/decode, and support for 24-bit WXGA 1280x800 resolution displays. They also feature integrated GPS and a low-power audio engine (read: a separate DSP core for MP3, etc.). It should be noted that Texas Instruments’ OMAP 4 application processor, announced in February, will feature dual-core ARM Cortex-A9s (and coupled with modems by others). Do I detect a dual-core apps processor trend here?

MIPS targets LTE cell phones

Also at Computex, MIPS Technologies made an announcement with SySDSoft about great benchmarking results they have achieved with MIPS processors employing SySDSoft’s LTE UE (L2-L3-NAS) software stack for Category 4 performance on Android and Linux platforms. SySDSoft claims to already have seven licensees for its LTE stack, lending credibility to MIPS claims of wireless capability. MIPS’ history in portability has been largely through NetLogic/RMI’s Alchemy low-power media processor, which is popular in personal media players (developed by some of the same talent that earlier developed the StrongARM processor for DEC, a device that morphed into Intel’s XScale). Now, MIPS’ challenge will be to demonstrate power efficiency in wireless applications to make headway in the 4G handset market.

White LEDs boost cellular RF capabilities

No, we're not talking about a new form of alchemy, but follow the logic. White LEDs are fabricated on sapphire substrates. The fast-growing white LED market for backlighting TV and PC displays has the effect of lowering the cost of sapphire crystals. Chip fabrication on sapphire wafers has been around for years as Silicon on Sapphire (SoS), originally as expensive high-performance, radiation-hardened Metal-Oxide-Semiconductor (MOS) devices for military and space electronics. But, with the lower cost of sapphire, SoS is proving cost-effective for commercial chip implementation.

Peregrine Semiconductor uses SoS for RF switches, displacing GaAs and Micro-ElectroMechanical Systems (MEMS) competitors in many cell phones. Peregrine’s switches, employing SoS-based UltraCMOS technology, are currently shipping in selected cell phones from companies like LG, NEC, Samsung, Sharp, and Sony Ericsson. The chip vendor claims smaller size, lower insertion loss, and better isolation than competitors’ switches. Peregrine recently added IBM as a new foundry partner, further improving its production capability. And Peregrine plans to expand further in the RF chain with high-power switches, tunable components, and multi-mode power amplifiers with its unique technology. So, Peregrine will be an interesting company to watch...especially as it plans an IPO later this year.

3D video coming to cell phones

It’s already invading the living room through new TVs that enable three-dimensional programs like Avatar. Next year, you'll be able to capture 3D video on a new-generation smartphone. Movidius (Dublin, Ireland) is busy perfecting its Myriad video processing platform that will enable 3D capture, and a few display companies are already demonstrating 3D LCD products that enable stereoscopic viewing without special glasses. Movidius recently announced that it has received a series B investment of $7.5 million, which should get the chip maker to initial production in early 2011. Of course, not all smartphone makers will be rolling out 3D handsets next year, but with so many Android-based devices now hitting the market, cell phone vendors will become desperate for product differentiation. Clearly, this could be a big differentiation from the rest of the mob.

Intel anxious to return to cellular business

Remember that Alexander the Great is said to have cried when there were no new worlds to conquer. While Intel may not be crying, the company realizes that the only world that can rival that of PCs is wireless. And Intel, now under different management, is searching for ways to get back into the cell phone chip market that it once abandoned. It had reasonable success with its XScale-based application processors, but had a tough time merging its multilevel flash memory on the same die with its baseband “Frio” DSP core. Intel sold both properties to Marvell, which has since revitalized and expanded on them. In January, Intel loudly touted the first cell phone customer for its new Atom chip, as an application processor for LG Electronics’ GW990 smartphone. However, by April, LG admitted that it was just a concept product and will not be released. So now, Intel is betting heavily on the latest Moorestown implementation, specifically the new Z600 family of Atom processor chipset to address the huge cell phone application processor market, now dominated by companies like Texas Instruments, Samsung, Qualcomm, and, yes, Marvell.

In mid May the Financial Times Deutschland published an article stating that Intel is interested in acquiring the cellular chip properties of Infineon Technologies AG (which provides the cellular modem and RF transceivers for all iPhones and the iPad). Such an acquisition would, of course, put Intel head-to-head with Qualcomm (the world's largest fabless chip house and Number 1 cellular chip vendor). But, Infineon's CEO explained that his cellular chip business is doing quite well and is not for sale. That would leave Intel with the option of bypassing management and making an offer that stockholders wouldn't refuse...or consider another acquisition that would provide it with a less-complete cell phone solution. The other desirable acquisitions (from a complete product portfolio standpoint) would be much more difficult, namely, Renesas’ cellular operation (I sense a culture clash there) or ST-Ericsson (which would have to be approved by three national governments and the EC...not likely by a long shot).

Nokia announced (in February, 2009) that it was making its 3G IP available to Intel. As we pointed out in an earlier newsletter, that IP currently runs on Texas Instruments DSP chips. If Intel doesn’t find a cellular chip acquisition, it’ll have to find a suitable DSP core (maybe even from TI?) for porting that IP, and begin investing heavily in LTE development. We’re still watching.

What of Icera?

If Intel were to succeed in acquiring Infineon’s cellular business, that would make Icera “the prettiest girl at the party.”  Icera, after much sweat and VC money, has its own HSPA and LTE chips and stacks (albeit only for data dongles at this writing, but with plans to be in 3G cell phones this year).  And with the stock market recovering this could be Icera’s year to float an IPO. Some pundits have suggested an Icera IPO could raise from $600 million to $1 billion. Hey, Intel paid about $2 billion for DSP Communications Inc., for which it wrote off $600 million and later sold the product line for another $600 million to Marvell. But, if Infineon is not available, maybe Intel should be looking at acquiring Icera before the IPO hits?

LTE chip startups abound...but

Any future LTE cell phone chip plans must have a 3G solution in place, too. There are over a half-dozen chip startups addressing LTE terminal chips, but most are addressing LTE-only capability for USB data dongles (which don’t need screens and application processors). For LTE cell phones, there must be a 3G voice and fallback capability (to 2G, too). 3G/UMTS software stacks are extremely complex and are available for licensing only from Infineon’s Comneon subsidiary or InterDigital Communications. All other sources are either in-house (like Qualcomm’s, Motorola’s, Nokia’s or ST-E’s) or developed from earlier-licensed TTPCom stacks...that are no longer licensed since Motorola bought the operation and shut it down. At least one of the new LTE chip wannabees understands the cell phone 3G situation. WiMAX chip supplier Beceem Communications has licensed InterDigital's SlimChip 3G/2G chip design to go with their LTE solution now under development. Intel Capital is a major investor in Beceem, so maybe there’s more to this?

Shameless plug

Our newest market study, “Global Cellphone Market & Subscriber/Operator Ecosystem” provides strategies and insight for the cell phone market and the dynamics of subscribers, operators, and Internet penetration by country. The study covers more than 50 cell phone vendors and their market shares. It also provides dozens of detailed forecasts of the cell phone market by global region and air interface technology within each region through 2014. Subscriber makeup by population (POPs), air technology (GSM, UMTS, HSPA, CDMA-1x, CDMA EV-DO), mobile Internet subscribers, and 2009 subscriber growth are provided for each of 188 individual countries. Cellular service operators in each global region are profiled by the technologies they employ, their subscribers for each, and 2009 subscriber growth. Details are at: www.fwdconcepts.com/globalcell.htm

My next speaking engagement

I’ll be presenting at the next Wireless 2.0 conference in Bristol, U.K. on July 9th covering technology and strategy challenges of multimedia, and HD video in the handset. Details are at www.siliconsouthwest.co.uk I look forward to seeing many of you there.

As always, I invite your comments.

Will Strauss

President & Principal Analyst

Forward Concepts
www.fwdconcepts.com
wis@fwdconcepts.com