The mobile chip market has been a scene of dramatic upheaval for the past few years, as smartphones gain PC-class computing tasks and processors to match, drawing a new set of contenders into the space. The entry of Nvidia, Intel and others has put pressure on mobile incumbents to up their apps processor game and understand the needs of intensive computing and web activity. It has also increased their focus on their own differentiators, their expertise in modems and in integrating multiple radios with the processors. Integration of more and more components in low cost, low power SoCs (systems on chip) have opened up another battlefield, one that is increasingly vital as smartphones enter the mass market and become price sensitive. Here, traditional all-in-one suppliers like Qualcomm and ST-Ericsson are kings, and Broadcom has obvious credentials, but the newcomers are stepping up their game too, with Intel acquiring Infineon Wireless and Nvidia buying Icera, among other transactions aimed at the fully integrated smartphone SoC.
This is reducing the competitive advantage of the well-established majors, especially as two of them, Texas Instruments and Freescale, have virtually exited the handset baseband sector, possibly with poor timing, as they are now excluded from the all-in-one segment. Qualcomm is showing few signs of weakness though, and has managed to leap ahead of its rivals both on the modem side, with early HSPA+ and LTE launches, and on the applications front, where its close alliances with software giants are paying dividends, as seen in its monopoly (for now) of WP7 smartphones.
That does not mean Qualcomm can rest on its laurels, and the company’s story will be one of transition over the coming few years ‘ so far, it has avoided dramatic changes of direction because it has not been forced to react defensively to a major threat, but the first half of this decade will nevertheless mark a huge shift in the giant’s business priorities as it adapts to the slow decline of CDMA-based technologies and to a new IPR landscape in LTE (though this is far less transformed than Qualcomm’s enemies would have wanted).
For now though, this is Qualcomm’s market to lose, and it is expanding into the digital home and PC-class devices like tablets too. According to Strategy Analytics, Qualcomm continued to dominate the smartphone application processor and baseband markets in the first quarter, but in the app processor alone, Nvidia was looking threatening, breaking into the top five for the first time in terms of revenue share. The analysts estimate that the global smartphone apps processor market grew by 108% year-on-year in the quarter, to reach $1.68 billion in value. Qualcomm, TI, Samsung, Marvell and Nvidia formed the top five, and Qualcomm saw its own revenue rise by 46% year-on-year to $3.88 billion. Strategy Analytics believes that the company’s Snapdragon app processors accounted for 22% of the total and helped keep average selling prices from slipping.
Jim McGregor, an analyst at another chip research firm, In-Stat, said both TI and Qualcomm will need to be agile to stay ahead of new competitors but is sanguine on both. TI is ‘obviously facing more competition now than in the past, but that hasn’t stopped them getting designs,” he said. He also thinks Nvidia will rise further up the league table though, and that it will take on Qualcomm head-to-head with an integrated processor/baseband offering. The Strategy Analytics survey found that the cellular baseband market grew by 20% in the first quarter to reach $3.5 billion. Broadcom, Intel and Spreadtrum all showed good growth, but MediaTek and STE had negative revenue growth in this sector. According to Forward Concepts: “Qualcomm retains the leadership position in cellphone chips of all types, followed by Texas Instruments, Intel (after acquiring Infineon), ST-Ericsson, MediaTek and Broadcom. TD-SCDMA handset growth is strong now in China with ST-Ericsson, Spreadtrum and MediaTek as the lead baseband vendors, and Qualcomm is expected to soon follow.”
But very different patterns are emerging in LTE, and in fluid new markets danger lies for incumbents, and chances for new entrants. We have seen Broadcom buy 4G modem maker Beceem, as well as Nvidia’s move on Icera, while other 4G specialists remain available (such as recently IPO’d Sequans). Already, the LTE market is looking different to that of 3G, with Samsung in first place ahead of Qualcomm and LG. The LG Revolution LTE handset uses GCT’s 4G modem with LG’s own stack, but combined with Qualcomm’s MSM8655 Snapdragon for CDMA and the app processor. The new Samsung Droid Charge is one of the few early LTE phones to exclude Qualcomm, using the firm’s own LTE modem and a CDMA chip from Via Telecom, the only real CDMA baseband rival to the San Diego giant, but only seen in low cost or Chinese CDMA models so far.
However, only a few of the score of vendors which have announced LTE basebands are shipping as yet, and volumes are low, so the rankings will change. Several suppliers are waiting in the wings to grasp that opportunity, and it is not just Nvidia that Qualcomm needs to watch. Another new challenger is Renesas Mobile, whose CEO, Ikuyo Kawasaki, has targeted 20% to 30% market share within the next few years (from an estimated 3% now), even though the firm was only formed as a separate entity six months ago, when the Renesas parent firm acquired Nokia’s modem business, with its 1,900 employees. As well as the staff, the purchase brought Renesas an LTE protocol stack that Kawasaki sees as his chief weapon in becoming a Qualcomm competitor. The low hanging fruit will be to take over TI’s Nokia business as the US firm exits this space. Will Strauss, principal at chip analyst firm Forward Concepts, told EETimes: ‘There’s no question that Qualcomm is the number one in modems for 3G, HSPA+ (and CDMA); TI is the number two in 3G and HSPA (thanks to Nokia); and Renesas should take over much of that after 2012, but not much before then.’
As well as the modems, Renesas this year announced its contribution to the apps processor race, a dual-core device based on the ARM Cortex A9 MPCore with Imagination Technologies’ multicore PowerVR series for graphics. As a latecomer, Kawasaki says he will seek different channels to those of Qualcomm, in order to differentiate Renesas and take a lead in the move towards the LTE cloud, where many different gadgets will include 4G links. ‘Just as Qualcomm understands the telephone business, we believe that Renesas Electronics understands systems businesses. Our application processor will be designed into a number of platforms including cars, cameras, game consoles and other digital consumer devices ‘ across the board,’ he said in the interview. He expects to see LTE integrated into devices beyond handsets and tablets from 2013.
The rise of Renesas as more than just NTT DoCoMo’s tame chip supplier puts even more pressure on the established majors. While Qualcomm, as we have seen, has been steadily transforming itself, more radical change programs have been necessary at Texas Instruments, STE and Freescale, whose pressures have been more urgent. TI reacted to the loss of its near-complete hold on Nokia’s handset business – as the Finnish firm first adopted a more diverse supplier strategy, and then saw its market share decline anyway ‘ by defocusing on merchant basebands and prioritizing the OMAP apps processors, its burgeoning analog business, and its base station platforms. As recent comments from the firm regarding the continuing impact of Nokia’s problems showed, the transition is far from complete and will bring some pain with it, but the analog strategy is clearly working and TI has rejuvenated its wireless infrastructure offering in response to a newly aggressive Freescale. However, there is risk to OMAP from the lack of an integrated platform, especially as the acquisition of some modem specialists means that there will be fewer non-aligned partners for TI and its customers.
At STE, several years of mergers ‘ first an STMicro-NXP venture, then the addition of Ericsson’s processor business ‘ have probably delayed the shake-up of the product range and the company organization for longer than was ideal. However, it is in train now and STE is insistent that results will be seen from late this year, despite the frustratingly slow process of reducing reliance on legacy products. STE has rationalized its businesses post-merger, and refreshed its product range to address a rising level of competition. Last year it said announced a whole new smartphone chipset range to respond to Qualcomm as well as new rivals like Nvidia, but it is still being dragged down by weakness in its legacy products, and has announced further cost reduction programs.
STE, a joint venture formed in February 2009 between Ericsson and STMicro, aims to cut costs by $120m a year by the end of 2012, which will involve the loss of about 500 jobs from its total payroll of 6.700. This will cost about $55m, with most of the charges to be taken in the second half of this year.
The company cited ‘recent changes in the business environment and reduced demand for legacy products at certain customers’ as the reasons for further restructuring. In particular, it is likely to have suffered from Nokia’s recent troubles, since it the Finnish player is a major customer on the 3G side. Texas Instruments also recently blamed Nokia for a disappointing performance in its baseband business, which it is exiting. Another key problem for STE is a slower than expected ramp-up of China Mobile’s TD-SCDMA services, in which the venture is an advanced provider of device silicon.
It has admitted that the decline of its older products has not yet been offset by sales of its new ranges, such as Thor and NovaThor. It is reported to have won a place supplying Nokia for its upcoming WP7 smartphones, but Qualcomm ‘ currently the only company providing commercial chips for WP7 devices ‘ will be the first and primary vendor.
Last month, CEO Gilles Delfassy told an analyst day that he had withdrawn the target of achieving break-even by the second quarter of 2012. “These actions while necessary to strengthen the financial position of the company, will not compromise the execution of our new products and delivery to our customers,” said Delfassy in a statement. “We continue to gain traction on our new product portfolio and remain steadfastly committed to leadership in the smartphone and tablet markets.” In the first quarter, STE announced a net loss of $178m, compared with a year-ago loss of $154m, on net sales of $444m, down from $606m.
Meanwhile, Freescale has reduced its dependence on cellphones even more decisively than TI, though it pushes its app processors into tablets and ereaders. As well as wireless infrastructure, it has steadily put embedded processing at the heart of its new-look strategy, and CEO Rich Beyer was outlining the turnaround at last week’s Freescale Technology Forum in Texas. Beyer took the helm in 2008 and has overseen a painful restructuring, following the 2006 leveraged buy-out by private equity firms, which incurred a debt mountain of $7.6bn. Beyer has pared down the diverse product range to focus on key areas of differentiation and now sees everything as centered on the embedded processor ‘ surrounded by sensors, RF, analog components and software. These integrated platforms will become the basis of a bewildering array of smart devices, and Beyer was not shy of quoting the oft-cited Ericsson prediction that these will reach 10bn units by 2015.
‘Thanks to embedded processing, we have cut the cord to a desktop. We are creating devices that are always on, always connected and always with us; and they are the devices that conform to us, not the ones we need to conform to,’ he said in his keynote, adding that Freescale wants to cash in on this trend by riding the wave of ‘connected intelligence’. This means building IP infrastructure to support the new devices; develop processing engines for smart mobile device and cars; and target distributed intelligence in key verticals like telehealth, smart energy and Freescale’s automotive stronghold.
There are three key business divisions to support this (hardly unique) strategy – microcontroller solutions, networking/multimedia, and RF/analog/sensors. Every group increased 2010 net sales by double digit percentages over 2009 (microcontrollers by 43%, networking/multimedia by 33% and RF/analog/sensors by 30%).
Freescale claims it will not suffer from the cellphone baseband exit ‘ even though SVP Tom Deitrich admits that highly emotive decision was ‘gut wrenching’, given Freescale’s heritage. Although OEMs are increasingly looking for integrated SoCs, Freescale thinks these will become commoditized and that it can offer a different form of integration, more suited to the new device categories – a solution that includes a multimedia processor along with sensors, power management and other analog components, all pre-integrated. Sensor technology was highlighted as the most important new area for R&D resources, while Beyer named ‘power/RF and software to cultivate ecosystem partners as the other key investments on the development front. However, the chief competitive differentiator, he claims, remains Freescale’s deep understanding of processor architectures, particularly through supporting the Power Architecture it codeveloped along with ARM, and offering a wide range of products that are targeted at specific markets such as base stations, network infrastructure, mobile devices and cars.
Diversity is no bad thing ‘ a new report from Forward Concepts points out that, although basebands constitute the largest non-memory cellphone chip market at $12.7 billion in 2010, there are many other opportunities in the handset – $4.5 billion for power management units, $3.7 billion for RF transceivers, $2.9 billion for image sensors, $2.8 billion for RF power amplifiers, $1.5 billion for standalone apps processors and $1.2 billion for touch- and navigation-screen controllers. Other cellphone chips in the $1 billion-plus league include those for Wi-Fi, Bluetooth, MEMS and sensors, while GPS and FM radio are also significant peripheral chips each with over $500 million apiece in 2010 shipments. The ‘combo’ wireless peripheral devices, integrating Wi-Fi, FM/AM, Bluetooth, GPS and NFC, now constitute 8% of the cellphone component market, benefiting firms like Broadcom, CSR and TI.
But the highest expected 2011 growth rates in cellphone chips are all about mass data ‘ LTE basebands at eightfold (from a very low base of course); standalone apps processors at 17% and Wi-Fi at 15% (plus NFC at 94%, GPS at 33%, touch controllers at 25%, and MEMS at 14%). It is hard to argue against Qualcomm remaining the best positioned vendor to take advantage of these particular growth trends, but with Icera on board with its LTE and 3G+ modems, Nvidia is certainly one to watch.
This article first appeared in Wireless Watch our sister publication