The settlement between Qualcomm and Nokia, whatever the terms, opens the door for the market leader in handset baseband chips to do business with the market leader in cellphones, a combination that will strike fear into the hearts of every other mobile chipmaker. At least, until now, the Qualcomm-Nokia enmity ‘ long predating the most recent licensing battles ‘ has kept the CDMA giant out of the most valuable customer available, somewhat limiting its potential impact has it has moved out of its CDMA corner into head-to-head competition with the W-CDMA chip vendors. So, faced with pressurized markets and a new competitive threat, how are Nokia’s other chip suppliers shaping up, and does Intel have a chance of joining the line-up?
Intel last week announced its entry into the market for system-on-chip (SoC) integrated circuits, as part of its stepped-up assault on the mobile devices sector. Logically, this assault will have to involve bidding for Nokia business in future, beyond just the WiMAX internet tablet that is based on Intel silicon. There are many potential mutual advantages to a Nokia-Intel rapprochement, particularly the chance for the two main formats for mobile internet products ‘ the PC-based MID and the cellphone-oriented smartphone or tablet ‘ to converge to drive standards and economies of scale. However, previous attempts to work together, such as an abortive joint venture for laptop 3G, have foundered on conflicts of interest, as both giants look to take the dominant position in shaping the future of the mobile device. And the Qualcomm-Nokia deal may make Intel less politically useful to the Finnish company ‘ Qualcomm is making its own new moves, into Intel’s core wireless notebook territory, and may now appear a viable alternative partner for Nokia in this territory.
Intel, then, like the more established mobile chip companies, will now see Qualcomm as an increased threat, because of the ability for the CDMA leader to penetrate Nokia. This will prompt it to accelerate its own mobile strategy, centered on future iterations of the Atom processor. So far this strategy has revolved around chips for laptops, PDAs and what Intel calls mobile internet devices, which are the heirs to the PC concept. But to maintain its position in the overall semiconductor market in the next generation of client devices, Intel will need to make a serious play for smaller, more power constrained mobile devices, including handsets and media players that will support the mobile web in a manner that stems from the smartphone heritage. That means lower power consumption, highly integrated and compact chips and support for 3G and 4G radios as well as Wi-Fi and WiMAX. In these areas, Intel has little track record (and its previous attempts have generally ended in failure, notably when it exited handsets last time around, selling its XScale processor architecture to Marvell).
But it has been working hard to drive down the power requirements of its x86 architecture ‘ though not yet to the levels of the ARM processor most of its rivals use ‘ and plans a full mobile version of Atom, its key microprocessor for ultramobile PCs, next year. Now its latest step is to enter SoCs, most of them likely to be based on the low power Atom. Intel currently has 15 SoC projects in its pipeline, all tailored for new growth markets, according to Gadi Singer, general manager of the SoC enabling group, talking to EETimes. Of these 15 projects, eight will be rolled out as products soon, he said, stepping up the challenge to ARM in the mobile and embedded/industrial markets (and Mips in embedded processors).
Intel’s first SoC, the EP80579, is about to enter production, combining a Pentium M-based processor design with memory controller, I/O controller and a set of integrated application specific accelerators on a single chip. According to the company, the device will feature 45% smaller footprint and 34% lower power compared to separate implementations. In most subsequent versions, Atom will replace Pentium M and there are also multicore models on the roadmap.
Looking over both shoulders at Intel and Qualcomm are Nokia’s established suppliers, which with the exception of long time partner Texas Instruments were only admitted to the Finn’s magic circle last year, and have yet to generate significant revenue from the giant phonemaker. Two of its chosen vendors, STMicro and Infineon, are enmeshed in a complex reworking of the European semiconductor industry, and a wave of merger or joint venture speculation as ST puts its wireless activities into a joint venture with NXP of the Netherlands.
Infineon has scored several key design wins in recent years, notably the 2G and 3G iPhone and Nokia’s low power GSM business. However, it was excluded from the ST-NXP Wireless venture, despite widespread talk that these three European giants would create a super-company to challenge TI and Qualcomm, and unless its wireless activities are spun off (or merged, as other analysts speculate, with Freescale) they will continue to be dragged down by the losses of the German firms’ memory chip unit Qimonda. Posting a widened third quarter net loss last Friday, Infineon said it would cut 3,000 jobs (of 43,000 worldwide) as part of a restructuring program, to save €200m ($314m) annually, and offset the effects of the weak dollar.
Infineon’s loss was €592m, compared with a year-earlier loss of €197m. They included a writedown of €411m related to Infineon’s 77.5% stake in Qimonda. Infineon already wrote down €1bn on its Qimonda stake in the second quarter. The company wants to cut its Qimonda stake to below 50% by early 2009. Revenue rose 1.8% to €1.03bn from €1.01bn. Despite recent strong progress, the wireless unit remained unprofitable, though the automotive, industrial and multimarket division reported a profit, despite weakness in the auto sector.
Over at STMicro, which is working with Nokia on 3G and 3G+ designs, the Swiss-based company plans further divestments of product divisions, following the deconsolidation of its flash memory business and the formation of the wireless venture with NXP. Three to five product families could go in the next few months, according to CEO Carlo Bozotti, through sale, JV or closure. Partner NXP saw flat sales in the second quarter, accompanied by a significant decline in operating profit and a ‘cautious at best’ outlook for the quarter ahead. The company reported sales of $1.524bn, a slight decline of .just under 1% against the same period a year. The EBITDA earnings figure declined by about 40% year-on-yearn to $114m.
After its own lacklustre quarter, Texas Instruments is seeking to raise its game in areas where it believes it has, or can grab, advantage over Qualcomm, notably analog markets and very low power DSPs. It said last week that it plans to overhaul nearly all its DSP lines with 15 new chips that reduce power consumption up to 20 times. The line-up will include TI’s first chip to combine an ARM core with a floating point DSP and will be aimed at attracting additional Nokia business in low power devices as well as advanced mediaphones, trying to fend off possible incursions by Qualcomm.
Finally, Nokia’s EDGE chip second source, Broadcom, demonstrated its legendary resilience and keen sense of timing by defying the market downturn and reporting quarterly profit that nnearly quadrupled, and better than expected revenue. Profit rose to $134.8m, or 25 cents a share, from $34.3m, or 6 cents a share, in the year-ago quarter. Revenue increased to $1.2bn from $897.9m. On average analysts had expected revenue of $1.102bn.