Intel to pay $1.4bn for Infineon WLS
But does it really need it?
Comment Intel has embarked on a major shopping spree to counter the pressures on its traditional businesses, which prompted it to issue a results warning at the end of last week. That was swiftly followed by the announcement that it would acquire Infineon‘s wireless arm, as widely expected, for $1.4bn, hard on the heels of the purchase of security software firm McAfee.
While some elements of the proposed McAfee takeover were baffling, the general strategy was clear and necessary – diversifying the product range, increasing software and services elements as opposed to pure silicon, adding new value to that silicon, finding new routes into the wireless and embedded markets where Intel‘s progress has been obstructed remarkably effectively by the ARM- based community.
The planned acquisition of Infineon WLS (Wireless Solutions), which should close in the first quarter of 2011, is the direct opposite of the McAfee deal. At first glance it is more logical. After all, it will increase, at a stroke, the range and scale of Intel‘s activities in the market it most wants to penetrate, mobile devices. Indeed, only last week a report from analysts at IC Insights said that a major purchase of this kind was the best way to fend off the fate of being overtaken by Samsung as the world‘s largest semiconductor maker. But a closer look raises far more questions that it answers, most importantly, why does Intel want Infineon? It may not be for the most obvious reasons – the expansion in cellphones – but for less high profile aspects of the business, such as its connections to its current parent‘s strongholds in the auto and other embedded sectors, and Infineon WLS‘s specialized architectures for security, which could complement McAfee in a major on-chip security strategy.
The price was low – even a "fire sale", some thought – despite Infineon CEO Peter Bauer‘s insistence that there was no urgency to sell and he would hold out for a high price. Unlike its parent, Infineon WLS has been stable and profitable in recent times, but the merger process could still be fraught, especially with the involvement of German takeover and employment laws, the firms‘ very different cultures, and possible manufacturing tie-ins to the larger Infineon. Intel says it will address these risks by keeping WLS as autonomous as possible, though this in itself could dilute some hoped-for economies of scale and product integrations.
The customer base
The price was probably influenced by the likelihood that Intel will lose, over time, several of the major customers for which some commentators thought it acquired WLS. Infineon‘s main product strength is in cellular modem basebands, and in this sector it is currently ranked fifth in revenue terms and fourth in unit shipment share terms, and is estimated to be on track to ship as many as 300 million basebands in 2010. However, it could lose several of its largest customers.
Also, Samsung sees Intel as a key competitive threat in memory and its own growing app processor business. In recent years, Samsung has got far closer to Infineon, at the partial expense of Qualcomm – the San Diego giant will waste no time trying to fill in any gaps left by WLS, as will MediaTek, and they might even hope for Samsung‘s influence over Apple to facilitate that vacant baseband slot in the next iPhone (Samsung provided the app processor for the first three iPhones and codeveloped the A4 which powers the iPhone 4 and iPad. This is closely integrated with the Infineon baseband).
The likelihood of WLS losing out in future iPhones could be increased as Apple will have to use a Qualcomm baseband in the presumed CDMA version, and could find it attractive to tap the San Diego firm for all future models (especially the iPad, where it could use Qualcomm‘s innovative CDMA/UMTS Gobi modem).
The Apple-centric nature of the markets has tended to portray the Intel- WLS deal as, partly, a way for the latter to shore up its relationship with the iPhone maker, but really, even if Intel/WLS keeps that slot, the Infineon contribution to the iPhone is minimal, low margin and easily replaceable in future models. The two chips it supplies, the baseband and quad-band EDGE transceiver, total an estimated $14.02 in a total BOM of $187.51 for the iPhone 4, according to iSuppli, while the slots Intel would really like, the app processor and memory combination (controlled by Samsung) total $52.35.
Unlike Samsung, Nokia might welcome the new ownership of its strategic supplier for 2G and some low end 3G basebands and SoCs. The market leader has been growing closer to Intel at the high end, notably with their Atom oriented MeeGo software platform, and is an increasingly large customer for merchant basebands now it relies only slightly on inhouse or customized designs. It also has a joint development with Infineon focused on gigabit LTE silicon for future devices. However, the sale of its own modem operations to Renesas suggested that Nokia would transfer much of its baseband business to the Japanese partner in future, since it retains a joint venture relationship.
If the large baseband customers cannot be guaranteed to remain on-board, this must be about the ability for Intel to offer an integrated digital/RF platform, rather than just a standalone app processor. It has not done this since it sold off its XScale business, with integrated ARM processor and modem, to Marvell in 2006. Peter Clarke of EETimes commented: "Intel's move begins to look like an admission that by turning its back on communications in 2006 the company got it wrong."
The acquisition does give Intel the option to integrate digital and RF circuits and take a system level view of requirements and partitioning. That, together with its leading-edge manufacturing engine, does give it some key advantages in the market, but it is not clear that these will be enough to transcend the suspicion Intel is regarded with by some in the communications sector.
This suspicion stems from Intel‘s dictatorial stance in the early PC market. As Clarke put it: "This looks like a repeat of the Intel PC strategy where it took a system level view and started to tell OEMs: 'this is what your product looks like, this is the software, these are the components and pricing and this is the PCB layout'. And from there it was only a short step to telling OEMs what their margins were and how many of which types of product they should make each month."
And does Intel really need an integrated platform anyway, or its own baseband business, even one with an iPhone slot? Infineon already has one for low cost 3G and 2G, though many of its sales rely just on basebands, and of course mobile market leader Qualcomm follows the all-in-one approach. But others have increasingly moved away from modems. Intel certainly needs to improve its SoC activities, but it does not need to acquire a baseband firm in order to create integrated architectures around Atom – it could partner, as most of its mobile app processor rivals, like Texas Instruments, now do.
Intel failed in offering combined processor/baseband architectures before, and even previously giant vendors in this area, like TI and Freescale, have given up on the thankless margins and lack of differentiation to be found in modems (Nokia recently sold its own team to Renesas, while MediaTek continues to drive down the prices).
Samsung could overtake Intel in chips in 2014
Samsung could overtake Intel before 2015 and become the world‘s largest semiconductor vendor, with its mobile chips a key driver. This will put even more pressure on Intel to diversify its revenue streams and make a serious impact on the mobile and embedded devices market, though the acquisition of Infineon should improve its chances of staying ahead.
This is the view of analysts at IC Insights, who point out that Samsung‘s chip revenues have grown at a CAGR of 13.5 per cent between 1999 and 2009, while Intel‘s has slowed to 3.4 per cent in the same period. If the pattern continued, Samsung would overtake in 2014.
Intel‘s dominance has relied mainly on its microprocessors in PCs and servers, but it has had limited success in pushing its x86 architecture into mobile devices, which now represent the growth segment of the industry. Meanwhile, Samsung is larger in DRAM and NAND flash memory, where it is market leader, and is expanding its efforts and sales in many other areas that ride on mobile and embedded platforms – microcontrollers, microprocessors and wireless communication chips.
Although, in the past, it has not always won business with its own handset sister company, Samsung Mobile is now working more closely with the silicon arm and could be a vital lead customer in future – as in chips, Samsung is chasing the handset market leader, Nokia, aggressively and expanding its market share.
The main areas where Intel and Samsung clash are NAND flash and apps processors, both of which are growing on the back of growth in mobile gadgets. And Samsung has spent more on semiconductor capex than any chip vendor, including Intel, in six of the past seven years, says IC Insights. In particular, it has invested heavily in growing its ARM-based mobile processor business, whose flagship product is Hummingbird, and upgrading its foundry activities.
ARM and new markets
Intel could broaden its own mobile platform in two ways: purchase capabilities to boost the architecture it is building around Atom, or give up on trying to squeeze Atom into every market and support multiple platforms. Since WLS is wedded to the ARM architecture, we must assume the latter, especially as Intel said in its preliminary statement: "Intel is committed to serving WLS‘ existing customers, including support for ARM-based platforms."
Re-entering the ARM market is sensible, given the weight of market share and developer support the processor platform is acquiring in mobile, and even in markets Intel might have regarded as sacrosanct, like cloud servers and notebooks. WLS has an architectural ARM license, though this is confined to security applications, and a long relationship with the firm, which should be more positive than Intel‘s previous ARM related attempts – most recently, when it acquired the former StrongARM unit of Digital Equipment, turning it into XScale. WLS, at least, brings readymade channels and customers among the mobile giants, which StrongARM did not, leaving Intel to penetrate the unfamiliar territory ill-equipped.
But WLS‘ customers are not confined to the mobile majors. Its parent is a major player in automotive chips and this gives WLS key connections to an important expansion market for all chip firms targeting embedded devices, as cars become a new screen for internet and media consumption. And a key synergy that should not be underestimated is in security. Infineon has worked very hard on chip- level security and its ARM architectural license is centered on this area. The acquisition of McAfee indicated that Intel wants to dominate this segment as security issues become ever more complex with web access stretching to millions of diverse gadgets and embedded systems. In this respect, Infineon will bring far greater value than it does to the more obvious cellphone space. In its own statement, Intel said: "WLS‘ sale enables Infineon to expand leading position in markets for automotive, industry and security technologies," pointing to the real value.
The ARM deal was signed last November, and is targeted at high security SIM cards for multimedia applications. Infineon will integrate its proprietary security features into ARM‘s widespread CPU core implementation. The firm‘s 'Integrity Guard' hardware encrypts data across most parts of the physical chip implementation such as data buses and all memories as well as caches.
A further note of interest that emerged from the bare-bones statement was that Intel would use Infineon‘s baseband expertise to supports its plans to "accelerate LTE". We might all be forgiven for thinking Intel‘s plan had generally been to contain LTE in order to boost the WiMAX technology, over which it has far more influence. Infineon has some advanced LTE activities, including its R&D partnership with Nokia, and this will certainly enable Intel to participate in both 4G platforms, especially when LTE starts to appear in netbooks and notebooks (although, as in 3G, Intel could equally well have partnered in this area).
But it is clearly using its new acquisition plan to "out itself" in terms of the inevitable acceptance of LTE, and it said WLS "expands Intel‘s current Wi-Fi and 4G WiMAX offerings to include Infineon‘s 3G capabilities and supports Intel‘s plans to accelerate LTE. The acquired technology will be used in Intel Core processor-based lap- tops, and myriad of Intel Atom processor-based devices, including smartphones, netbooks, tablets and embedded computers." It will also enhance Intel‘s HSPA activities, via Infineon‘s 65nm HSUPA platform (XMM 6160), and upcoming 40nm HSPA platform (XMM 6260).
Copyright © 2010, Wireless Watch
Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.