Handset Component Technologies

Helps clients match technology solutions with demands for displays, batteries, cameras, storage and media, as well as semiconductor components, including baseband & applications processors.

March 1, 2015 18:03 srobinson

So here we are on the eve of another Mobile World Congress and my thoughts turn to what amazing new devices and technologies are going to be revealed this year.

One of the topics that I’m going to be following closely is displays, and I’m looking forward to getting my hands on the latest curved or even flexible displays, high colour-gamut displays, displays with built-in solar cells, and especially high-resolution smartphone displays. 

There are many smartphones on the market now with a QHD resolution display (2560 x 1440 pixels, often incorrectly referred to as 2K) including the Samsung Galaxy Note 4, Galaxy S5 LTE-A, LG G3 and Motorola Nexus 6. However, the first couple of smartphones to incorporate a QHD display were from much less well known vendors Vivo and Oppo, with their Xplay 3S and Find 7 models respectively. 

It’s unlikely that we’ll see an Ultra HD display in a smartphone this week. A 5.5-inch UHD (3840 x 2160 pixels) display doesn’t exist yet – not officially at any rate – and equates to a pixel density of around 800 pixels per inch. I don’t know any vendor that has a commercial grade 800ppi display yet, so I think we are going to have to wait a little longer for a UHD-equipped smartphone.

But there are enough challenges with a QHD display, let alone a UHD one, and even if a display with 800ppi was available and production ready now I don’t think we would see it in a 5.5-inch smartphone for another 6 to 12 months.

Going back to the title of this blog, why might QHD displays burn two holes in your pocket? Firstly the cost and secondly the heat. QHD displays are not cheap and the high resolution display necessitates other highly-specd components too.

As well as sporting an amazingly high resolution display many of the latest smartphones are also able to capture UHD video content at 30 frames per second or higher. But some users of the early devices were reporting thermal overload to the point where the device even shut down because it became too hot.

So, why are some smartphones with a QHD display and UHD video capture causing such thermal problems? There are several reasons:

- Number of pixels. A QHD display is made up of 3.69 million pixels, each of which has 3 sub-pixels (RGB) with 8-bit colour depth. That’s 88.47 million raw bits of information for each frame. 

- Frame rate. If the display panel is refreshed 30 times per second then the total number of bits of information could be as high as 2.65 billion bits per second. That’s without compression of course. 

- Codecs. Fortunately H.264 and HEVC (H.265) compression algorithms reduce the data flow to the display significantly. HEVC is approximately 50% more efficient in terms of bitrate than H.264.

- Hardware vs software implementation of the codec. The H.264 standard was ratified about 10 years ago and the most efficient implementations of it are done in silicon. However HEVC is still new and hardware implementations are only just coming onto the market now. Software implementations running on the CPU are not as energy-efficient as the dedicated hardware implementations.

Of course, I can’t finish without saying something about the latest Samsung/Qualcomm spat. Samsung reported last month that its upcoming Galaxy S6 would no longer be powered by the Qualcomm Snapdragon 810 processor due to thermal issues. It’s possible that the 810 has some power consumption challenges; for a start it’s based on the off-the-shelf ARM-Cortex A53 and A57 cores rather than Qualcomm’s own customised 64-bit architecture, which we are likely to hear more about this week and which normally offer significant power/performance improvements. However, Qualcomm has dozens and dozens of customers for its Snapdragon 810 processor, including the new HTC One M9 announced today, and none of them apart from Samsung is claiming thermal issues. Could it be that there is a hidden agenda here and that Samsung is using this as an opportunity to promote its own new octa-core Exynos 7420 processor instead? Time will tell.

So, the challenges with Quad HD displays will be all-but forgotten in a year’s time; the displays themselves will be much cheaper and the processors that drive them will be much more efficient and less power-hungry. Looking ahead to this time next year I predict that I’ll be talking about the challenges of UHD displays in smartphones and tablets, rather than QHD. Can’t wait.

Stuart Robinson

December 26, 2014 14:27 skundojjala

The LTE baseband landscape is evolving and the year 2014 saw improved activity in terms of competition, price tiers and market penetration. Qualcomm, the current LTE baseband incumbent, finally saw some LTE competition and is challenged by HiSilicon, Intel, MediaTek, Marvell, Samsung and others. Broadcom and Ericsson have already exited from the baseband market.

  • MediaTek, a late entrant in the LTE baseband market, is poised to become the solid number two LTE baseband vendor in 2014, thanks to LTE ramp in 2H 2014. The company is currently using a stop-gap dual-chip LTE solution (LTE slim modem plus 3G AP) to compete in the LTE baseband market. MediaTek started shipping its LTE APs now, but 2014 shipments are expected to be dominated by slim modems. We expect MediaTek’s 2015 LTE shipments to be largely dominated by integrated APs.
  • Intel continued its progress and scored multiple high-profile LTE designs at Samsung with the help of its XMM 7260 Category 6 LTE baseband solution.  Intel’s SoFIA 4G chips in 2015 could help further.
  • HiSilicon made significant inroads in the LTE baseband market and is now shipping Kirin-branded LTE APs in volume. The company scored multiple LTE smartphone design-wins at its parent company Huawei.
  • Marvell continues to benefit from the Chinese TD-LTE market. However, the competition is intensifying in the Chinese LTE market and Marvell’s share is expected to come under attack in 2H 2014. Marvell is expected to fight back with low-cost LTE APs in 2015.
  • Samsung is trying to reinvigorate its Exynos AP chip growth with the help of LTE APs. The company made limited progress so far. Samsung’s in-house LTE slim modems and LTE APs are maturing and we expect Samsung Mobile to use in-house LTE modems in more flagship and mid-range models in 2015.
  • Spreadtrum is currently a minor player in the LTE baseband market. But the company is expected to significantly improve its shipments with the help of 5-mode LTE slim modems and APs in 2015.
  • Single-mode vendors such as Altair, GCT and Sequans are now focusing on non-handset markets.

Strategy Analytics estimates that Qualcomm’s LTE baseband volume share declined to less than 80 percent in Q3 2014 compared to 95 percent in Q3 2013, thanks to increased competition. Qualcomm still has a significant edge in terms of product roadmap and design-wins. The company has already announced its 5th generation LTE chip MDM9x45, which is expected to become available in 2015. We believe Qualcomm has a multi-year lead in LTE integrated APs for smartphones and tablets. However, 2015 is expected to see more LTE competition, thanks to improved vendor product portfolios with 64-bit LTE APs. In addition, low-cost LTE handsets in emerging markets are expected to intensify the price competition in LTE basebands.

Sravan Kundojjala (@SKundojjala)

September 30, 2014 22:18 sentwistle

Leading TSC vendors Atmel, Cypress and Synaptics have developed bespoke noise reduction techniques which are enabling new display implementations able to accommodate multi-finger touches and 3D gestures, including touch detection through gloves, hovering, tracking wet or sweaty fingers accurately across the screen as well as providing fine-tip stylus support.

The smartphone handset OEMS faced cut-throat competition and needing to differentiate their products, are introducing larger and more sophisticated screens with new and advanced touch capabilities. This is placing an increasing burden on touchscreen controller (TSC) suppliers, who must design their products to meet ever more stringent technical specifications.

Altering the physical and electrical properties of touchscreens and accommodating advanced new features means that state-of-the-art TSCs increasingly need to have faster scanning and higher data processing capabilities, more sophisticated power management to minimize power consumption and, in particular, much improved signal-to-noise ratios.

  • The latest touchscreens are able to accommodate multi-finger touches and 3D gestures, including touch detection through gloves, hovering, tracking wet or sweaty fingers accurately across the screen as well as providing fine-tip stylus support.  TSCs now use advanced algorithms to detect these very small signals whilst rejecting the larger, unwanted signals from the user’s hand, from power supplies and other noise sources.
  • The display, particularly an LCD display, is a major source of noise, and the introduction of new integrated sensor/display designs exacerbates this problem. Leading TSC vendors such as Atmel, Cypress and Synaptics have developed bespoke noise reduction techniques which are enabling new display implementations such as display-integrated in-cell and on-cell touch systems and even single-layer in-cell sensor designs.

Further Client Reading:  Clients of Strategy Analytics Handset Component Technologies (HCT) service can read the full Insight report “Touchscreen Controllers: Enabling New Features and Advanced Capabilities in Smartphone Designs

Steve Entwistle

June 25, 2014 13:31 skundojjala

HiSilicon, Huawei’s in-house silicon business unit, has launched its first LTE-integrated applications processors Kirin 910 and Kirin 920 in 2014. HiSilicon already designs slim modems (Balong series) and stand-alone applications processors (K3V2 series), but this integrated product helps it to gain a significant advantage over the competition. HiSilicon is the world’s first LTE Cat 6 apps processor vendor with its Kirin 920 while Qualcomm and Intel’s Cat 6 LTE chips are slim modems only at the moment.

The Kirin 920, designed in 28 nm, integrates multi-mode Cat 6 LTE modem, quad Cortex-A15 / Quad Cortex-A7 big.Little octa-core CPU, ARM Mali T628 GPU and Tensilica HiFi3 DSP. The Kirin 910, designed in 28 nm, integrates multi-mode Cat 4 LTE modem, quad Cortex-A9 CPU, ARM Mali 450 GPU and Tensilica HiFi2 DSP. The Kirin 910 featured in multiple smartphone and tablet products from Huawei and the Kirin 920 featured in Huawei’s latest Honor 6 Android smartphone.

With these products HiSilicon joins the commercial LTE apps processor vendor club, which includes Broadcom, Marvell, NVIDIA, Qualcomm and Samsung. Broadcom, however, has announced its intention to sell or shut down its baseband business. MediaTek will introduce its commercial LTE apps processor products in Q3 2014 and Intel will release its SoFIA LTE chip in the first half of 2015. Spreadtrum is also likely to introduce its LTE apps processor later this year or early next year.

Like Apple and Samsung, Huawei is also now focusing on vertical integration and its network infrastructure relationships with operators have probably helped Huawei to bring multi-mode LTE 6 chips ahead of many of its baseband competitors. We estimate HiSilicon had less than 1 percent share in baseband and smartphone apps processor markets in 2013. However, an increasing percentage of Huawei’s smartphone shipments now carry HiSilicon’s chips, which bodes well.

While HiSilicon LTE chips are unlikely to threaten Qualcomm’s market share any time soon, it is likely that these chips could affect Marvell, MediaTek and Spreadtrum’s LTE chip supply ambitions to some extent. In our view, Qualcomm is well-differentiated with its LTE basebands, associated RF and connectivity technologies. We believe Huawei will continue to source LTE chips from Qualcomm for flagship phones outside China. Huawei’s LTE apps processor chips also validate Broadcom’s decision to exit the baseband market. It remains to be seen whether Apple, like Huawei and Samsung will design its own LTE chips.

Sravan Kundojjala (@Skundojjala)

June 2, 2014 17:21 skundojjala

Today, Broadcom announced that it is exploring strategic options for its cellular baseband business, including a potential sale or shutdown. The company hasn’t identified any potential buyer yet. With this announcement, Broadcom joins the long list of players who exited from the baseband market including some high profile player such as Analog Devices, Freescale, Infineon, NXP, ST-Ericsson and Texas Instruments.  Broadcom is still a minor player in the baseband market and the company had less than 5 percent revenue share in the cellular baseband market in 2013, by Strategy Analytics’ quarterly cellular baseband tracker estimates. Broadcom has been investing in baseband technology for more than ten years with little success to-date. The company has had some success in 2G and 3G markets, but never really capitalised on its momentum with tier-one vendors such as Nokia and Samsung. Despite not having a major market share in the baseband market, we always viewed Broadcom as a capable player, thanks to its strong relationship with Samsung, proven integration capabilities, proven execution with tier-one OEMs and strong cellular IP. Previously we said that 2014 was shaping up to be an important one for Broadcom’s baseband business and only a successful LTE execution will secure the long-term outlooks for its Mobile & Wireless business.

We estimate Broadcom has spent over $3 billion on cellular baseband related R&D since 2007 without profit. The baseband market is quite R&D intensive. The rationale for Broadcom to invest in basebands had been to expand its connectivity business to mid-to-low range smartphones where a complete platform approach is relevant, allowing Broadcom to capture more dollar content while expanding in the high-end at the same time. Broadcom feels its connectivity share in premium smartphones is relatively secure, thanks to its technology differentiation, strong customer relationships and quality of its products. In contrast to its connectivity chip business, Broadcom’s baseband chips were neither leading edge nor first to market. This situation left Broadcom in catch-up mode in basebands and only upcoming LTE SoCs and thin modems would have made it a high-end player in the baseband market. Unlike Qualcomm, Broadcom viewed its baseband business as supplementary to its connectivity business. For Qualcomm the baseband modem comes first and everything else is a revenue expansion opportunity.

Broadcom’s 3G baseband shipment year-over-year growth fell to just 4 percent in 2013 after experiencing a buoyant 193 percent year-over-year growth in 2012. This can be attributed to intense competition in the 3G baseband market and the maturity of 3G baseband technology. According to Broadcom, the company is expected to have just $200-$250 million baseband revenue in 1H 2014. We believe that the barrier to entry in 3G basebands is low now and as a result the market has seen rapid price erosion in recent quarters, leaving vendors with very little margin.

In 2013, Broadcom acquired Renesas Mobile’s LTE assets to accelerate its LTE product introduction. Broadcom also announced that the company is targeting at least $100 million in LTE revenue in 2014. At least through Q1 2014, Broadcom appeared to be on track with LTE product and business milestones. During its Q1 2014 earnings call Broadcom’s CEO announced the company’s upcoming Cat 7 / Cat 9 / Cat 10 basebands and SoCs, which was unusual for a company that typically announces its products closer to launch. We view this early announcement of its long-term product roadmap as an indication that Broadcom was attempting to make its baseband assets more attractive to potential buyers.

Currently the LTE baseband segment is the fastest growing sub-segment of the baseband market with above-average ASPs. Qualcomm is the market leader in the LTE baseband market with over 95 percent revenue share. In 2013, revenue from LTE basebands accounted for over one-third of total cellular baseband revenue in 2013 and we expect this growth to continue for the next few years.

This announcement leaves Qualcomm, Intel, MediaTek, Spreadtrum and Marvell as the key players in the baseband market. It remains to be seen whether Broadcom will find a buyer for its baseband business as historically Freescale, ST-Ericsson, TI and Renesas Mobile all struggled to find buyers for their baseband businesses when they put them up for sale. ST-Ericsson and Renesas Mobile, however, found a buyer for a piece of their baseband business and we expect Broadcom’s attractive LTE roadmap could attract a player with market share expansion plans.

Sravan Kundojjala (@Skundojjala)

September 12, 2013 14:09 srobinson

Every so often a technology development comes along that makes you think "wow, that could really change the world!" That thought occurred to me this week when I read about a new wireless charging technology...

Ossia, a company that has been in stealth mode developing yet another wireless charging technology, has finally revealed to the world what it has been working on in secret for the last six years.

The technology, called Cota (which probably stands for "charging over the air" although I haven't seen confirmation of that yet), was demonstrated on-stage at a TechCrunch event called Disrupt in San Francisco last weekend. Cota's inventor, Hatem Zeine, showed his system charging an iPhone from a distance of about 10 feet and said that it would work round corners and through walls, like Wi-Fi. In fact, like Wi-Fi, it also works on the same 2.4GHz frequency, delivering a focussed beam up to 1 Watt at a maximum distance of 30 feet (10 metres).

The prototype transmitter is huge and incorporates half a million components, but will be shrunk to the size of a desktop PC in the next year or two and will cost a little over $100 according to Zeine.

The prototype receiver will also be shrunk into a single chip and could be embedded within a cellphone, TV remote control, games controller, Bluetooth headset, smoke detector or anything else that is currently battery-powered. 

One of the many clever aspects to Cota is its configurability; again, like Wi-Fi you can set up your transmitter to be open to the public or closed to only one or a few devices.

Some questions still lurk in the back of my mind though:

- How safe is it? It's claimed to be inherently safe, "as safe as Wi-Fi", but then not everyone is convinced about the long-term health effects of Wi-Fi either.

- How efficient is it? Existing wireless charging standards are up to 70-80% efficient. It will need to be as good. Zeine claims that in the commercial version of the system the more transmitters there are the better will be the efficiency.

- How will it work in a vehicle? The automotive industry is embracing other wireless charging standards. The physical size of a Cota transmitter may be a problem for in-vehicle usage, unless the small individual components and transmitters can be dispersed within the vehicle.

The number of wireless charging standards was already high, and the addition of a new, potential disruptive one this week, must be a concern to the existing standards bodies, consortia and organisations:

- Qi, the standard with the most products on the market currently, is an inductive power solution that has been developed by the Wireless Power Consortium and has the backing of over 160 companies including TI, Fulton Innovation, NTT docomo and a host of handset manufacturers. We're aware of about 20 smartphones integrated with Qi so far and the list is growing. The WPC aborbed one of its competing standards, PowerbyProxi earlier this year; PowerbyProxi had developed a more loosely-coupled inductive system but pulled the plug on launching its own standard and joined the WPC in May 2013.

- A4WP, the Alliance for Wireless Power, is a consortium founded by Samsung and Qualcomm. Qualcomm's WiPower technology is incorporated into the company's new "Toq" smartwatch, and is based on the A4WP specifications. Intel's "Wireless Charging Technology" (WCT) will also be compatible with the A4WP standard.

- Power Matters Alliance, developed by Duracell Powermat, has a widely-used standard (Starbucks for example) that is partially compatible with Qi. Its downfall is its tightly-coupled operation with very little spatial freedom.

I've seen too many announcements over the years of promising battery technologies that are going to solve all of our smartphone-charging problems, and not seen any of them come to fruition yet, so I'm cautious about this announcement too, but cautiously optimistic.

With so many competing and incompatible wireless charging standards vying for space it's debateable whether any of them will succeed in the long-run. Wireless charging technology is still too immature (as this week's revelation by Ossia has demonstrated) for the industry to settle on one standard, but one standard is exactly what is needed if wireless charging is ever going to more than a niche fad. 

Stuart Robinson

September 11, 2013 13:33 skundojjala

Yesterday, Apple announced its new iPhone line up including the iPhone 5S and iPhone 5C. The interesting part from our perspective is that the iPhone 5S features a 64-bit A7 applications processor, which is the mobile industry’s first commercial 64-bit smartphone chip. Apple beat specialists like Qualcomm, Broadcom, Samsung and NVIDIA to bring the first 64-bit smartphone applications processor chip to market. Apple seems to have at least a 6-9 month advantage with 64-bit commercial smartphone chips, in our view. Intel’s 22 nm-based 64-bit Merrifield smartphone chip will debut early next year and the 64-bit Bay Trail chip for tablets will debut this month.

Apple surprised with its semiconductor prowess as the 64-bit A7 chip packs 1 billion transistors. For context, Intel’s 3rd generation Core product Ivy Bridge has 1.4 billion transistors. The A7 chip’s transistor count compares well with desktop and notebook processors released in 2010-12 time frame. Apple, an architecture licensee of ARM CPU cores, is clearly demonstrating its vertical integration advantage with tight integration between hardware and software. It seems that the A7 chip is manufactured using 28 nm process technology (TSMC or Samsung is unknown at this point). Apple claims its A7 chip is twice as fast as its previous generation A6 chip, while doubling on graphics performance. While unconfirmed, the A7 chip most likely integrates Imagination Technologies’ PowerVR SGX6-series GPU (codenamed Rogue).

ARM, the key CPU IP supplier to smartphones and tablet chip companies, revealed its 64-bit plans in 2012 by announcing ARMv8-based Cortex-A50 series. The 64-bit Cortex-A53 and A57 chips eliminate the RAM limitations of 32-bit chips and OEMs will be able to add more than 4GB of RAM. At that time, ARM announced that first 64-bit commercial chips will be available in 2014. But, Apple’s architecture license must have enabled it to bring its own chips to market ahead of ARM’s schedule.

While the use cases for 64-bit chips may be limited in the near-term, application developers are likely to find a few. One potential application is Adobe Photoshop. We feel that the 64-bit chips are more appropriate to big screen devices like tablets and Apple is likely to announce an A7-based new iPad in future. Apple kicked off 64-bit mobile chip wars at a time when 32-bit multi-core penetration in smartphone apps processors reached two-thirds of the total AP market.

We estimate that vertical vendors such as Apple, Huawei and Samsung captured 25 percent of volume share in the smartphone apps processor market in 2012. By our estimates, Apple ranked number two in the smartphone apps processor market and number one in the tablet apps processor market in 1Q 2013. Samsung’s industry first big.LITTLE chip Exynos 5 Octa chip coupled with Apple’s industry-first 64-bit smartphone chip announcements this year suggest that these vertical vendors have high ambitions to lead the technology curve. However, vertical vendors currently lack cellular SoC products, which leaves enough room for merchant vendors to continue to grow in the fast growing smartphone and tablet apps processor markets.

Sravan Kundojjala

September 9, 2013 16:01 skundojjala

We have been tracking the multi-core smartphone apps processor (MCSPAP) market developments since the introduction of such chips back in 2011. Last month, we have released our MCSPAP report, which details single-core, dual-core, quad-core and octa-core smartphone processor shipments across stand-alone and integrated categories up to Q2 2013.

Some of the highlights of our multi-core smartphone apps processor tracker report:

  • Based on Strategy Analytics research, multi-core chip penetration in smartphones increased to 66 percent in 1H 2013, up from almost zero in 2010. MCSPAP shipments registered more than two-fold year-on-year growth in 1H 2013. This multi-core proliferation can be attributed to strong efforts from chipset companies to bring advanced features at a lower cost. Chipset vendors now offer MCSPAPs at various price points and thanks to chip vendors’ efforts, dual-core smartphones are now available at sub-$100 price points in emerging markets.
  • Multi-core smartphone apps processors (MCSPAP) first out-shipped single-core APs in Q3 2012, thanks to strong traction for flagship chips such as Qualcomm Snapdragon 600 and low-cost Cortex-A5 / Cortex-A7 chips from multiple vendors.
  • Quad-core smartphone applications processor shipments registered five-fold growth in 1H 2013 compared to 1H 2012, while single-core smartphone applications processor shipments declined by 14 percent in the same period.
  • Qualcomm led the MCSPAP market with 43 percent volume share followed by Apple, Samsung, MediaTek and ST-Ericsson in 1H 2013. It was during 4Q12 that Qualcomm overtook Apple to become the leading MCSPAP vendor and continued in 1H 2013. Spreadtrum led the single-core segment in 1H 2013.
  • Qualcomm’s multi-core Snapdragon chips gained strong traction and featured in multiple flagship smartphones including the HTC One and Samsung Galaxy S4 among others.
  • NVIDIA, despite being an early entrant in the MCSPAP market, lost momentum in smartphones. However, NVIDIA’s upcoming LTE Tegra chip Tegra 4i could generate more volume in the smartphone market.
  • HiSilicon, Huawei’s in-house silicon designer, also gained strong traction in MCSPAPs with its K3V2 quad-core chip, but volumes were low.
  • Broadcom, Intel, Marvell and Spreadtrum started shipping their respective MCSPAPs in 1H 2013 and these vendors are still at the beginning of their multi-core product cycles. We expect these vendors to ramp up in the next few quarters.
  • In terms of baseband-integration market trends, single-core has the highest baseband-integrated chip penetration while quad-core segment has the lowest baseband-integrated chip penetration. This can be attributed to time-to-market advantage with stand-alone applications processors. Qualcomm, a strong proponent of integrated chip, used stand-alone approach to gain share in multi-core chips initially. We expect Broadcom, Marvell, MediaTek, Qualcomm and Spreadtrum to proliferate the market with low-cost baseband-integrated quad-core chips.
  • There are plenty of interesting multi-core technical approaches emerging such as ARM’s big.Little, NVIDIA’s 4-Plus-1 and ST-Ericsson’s eQuad.

Overall, multi-core penetration in smartphones continues to rise and by end of this year multi-core penetration could cross 75 percent. While it is debatable how many number of cores would be enough, the market is expected to see the first true octa core chip from MediaTek later this year. We think quad-core is the sweet spot in smartphones for years to come and we also expect the increased emphasis on GPU and DSP computing from chipset vendors, not just CPU core computing.

Sravan Kundojjala

September 4, 2013 16:27 skundojjala

Today, Broadcom announced the acquisition of Renesas Mobile’s LTE assets including its production-ready and carrier-certified LTE dual-core cellular SoC and patents. This acquisition jumpstarts Broadcom’s LTE progress and puts the company on the LTE map ahead its previous schedule. Broadcom’s original LTE-Advanced chip BCM21892 is expected to debut in early 2014 and would only have contributed materially in 2H 2014. Broadcom is still a minor player in basebands and the company had about 4 percent revenue share in the cellular baseband market in Q1 2013.

Renesas Mobile’s parent company Renesas Corporation announced the shutdown of its modem activities in June 2013. Before the shutdown, Renesas Mobile scored an LTE dual-core SoC design-win with Samsung, but that win came too late for the company. Renesas Mobile’s financial struggles, lack of global customer exposure and unsuccessful venture to integrate IP from multiple companies (Renesas (Hitachi plus Mitsubishi), NEC and Nokia IP) all took a toll and contributed to its collapse. ST-Ericsson, like Renesas Mobile, also struggled with organisational integration challenges and was eventually was taken over by part owner Ericsson, which has taken charge of ST-Ericsson’s 4G LTE thin modems. Unlike Broadcom, Ericsson is only focused on thin modems.

This is Broadcom’s second major LTE-related acquisition after Beceem Communications (October 2010). Broadcom said that the LTE roadmaps of both companies will converge and the company will ramp up its acquired LTE dual-core SoC and will sample a quad-core Cortex-A7 LTE SoC in 2H 2014. We believe with these kinds of products Broadcom can address the mid-tier LTE segment. Broadcom tends to announce products close to launch and we believe the company might release more competitive LTE SoCs (perhaps A15 / A7 combination) in the future.

Renesas Mobile’s LTE SoCs feature Imagination Technologies’ GPU core IP and we expect Broadcom to replace Imagination’s GPU core IP with its own VideoCore IP in future versions. It’s rather surprising that Broadcom didn’t mention Renesas Mobile’s MP6530 quad-core LTE-Advanced SoC (28 nm-based big.LITTLE A15 / A7). Renesas Mobile planned to mass produce this chip in 2H 2013 originally. We believe this product schedule may have been pushed out in order to start off quickly with what’s on hand (Renesas Mobile’s production-ready dual-core LTE SoC).

Currently the LTE market is dominated by Qualcomm, which captured 97 percent revenue share in Q1 2013. We believe Qualcomm’s share will no doubt prove unsustainable in the face of increasing LTE competition next year from Broadcom, Intel, Marvell, MediaTek, NVIDIA and Spreadtrum, which will all produce multi-mode LTE chips for smartphones and tablets.

Broadcom has a successful track record of acquiring and integrating assets that can jumpstart its market share. We believe Broadcom’s proven ability in acquisitions coupled with its strong wireless product portfolio and financial strength will enable Broadcom to emerge as a strong alternative to Qualcomm in the next 12-18 months.

Sravan Kundojjala

March 18, 2013 17:52 skundojjala

Today, Ericsson and ST Micro announced the breakup of the ST-Ericsson JV with ST-Ericsson taking over the 4G LTE multi-mode slim modem product line, while STMicro will oversee the existing products including legacy modem business, RF, Power Management and NovaThor integrated apps processors. Earlier, in April 2012, ST-Ericsson announced the transfer of its stand-alone apps processor R&D activity to ST Microelectronics. In addition, ST-Ericsson will put its connectivity business up for sale, but the company hasn’t identified any potential buyer yet for that business. We note that previously, several big names have left the baseband market including Analog Devices, EMP (Ericsson Mobile Platforms), Freescale, Infineon, NXP, and Texas Instruments among others. The baseband market requires intensive R&D for a company to remain competitive.

ST-Ericsson struggled since its formation in February 2009. The JV hasn’t been able to produce a single profitable quarter during its existence so far. Part of the JV’s struggles can be attributed to duplication among legacy products, transition to a new product roadmap and constant management changes. ST-Ericsson was formed by combining the modem assets of ST Microelectronics, EMP, NXP and T3G (TD-SCDMA consortium). Clearly, the JV struggled to integrate multiple companies and execute on its original plan to become a leading mobile chip company both in Europe and globally.

ST-Ericsson failed to establish itself as a true competitor to market leader Qualcomm over the last four years. ST-Ericsson got stuck in continuous product transitions to fill gaps in its product lines while the market moved on rapidly. The company lost momentum in the TD-SCDMA market and later lost significant revenue opportunities at Nokia. In 2012, ST-Ericsson saw some success with its excellent NovaThor U8500 dual-core chip at Sony and Samsung and the company shipped about 25 million NovaThor chips in 2012. However, this late success couldn’t bring ST-Ericsson to profitability, given its growing debt burden and legacy products. Based on our estimates ST-Ericsson ranked number-five in the baseband market in unit terms in 2012.

The latest breakup announcement jeopardises the relatively successful NovaThor “ModAp” product line of ST-Ericsson. Ericsson said it will solely focus on multi-mode 4G LTE slim modems in future, and will pursue licensing opportunities for its NovaThor baseband-integrated apps processor business.

Ericsson said the company wants to be the number three player in the slim modem market, and the company has given a 18-24 month time frame to achieve that. We think this self-imposed target is really aggressive unless the company has un-announced design-wins with Apple and Samsung. Currently, Qualcomm and Intel are the dominant 3G/4G multi-mode baseband players in the market. Ericsson said its LTE-Advanced slim modem M7450 will ramp in 1H 2014 and the successor product M7500 will ramp in 1H 2015.

To achieve the number three position, Ericsson would have to score iPhone or Galaxy S design-wins as the market for slim modems is relatively limited outside of Apple and Samsung. Based on Strategy Analytics estimates, baseband-integrated applications processors dominated the smartphone modem market with 62 percent unit share in Q3 2012. Ericsson could potentially pursue M2M, USB dongles, tablets and other non-handset markets as an opportunity to expand its slim modem business, but this would put it in direct competition with Intel and many small, innovative LTE baseband suppliers such as Sequans, Altair and GCT Semi.

We were somewhat surprised by the lack of buyers for ST-Ericsson’s modem business given ST-Ericsson’s 4G LTE products, which are production-ready. We have been saying for a while in our baseband tracker reports that the JV's concerning financial performance would eventually make it as a takeover target. We feel that the slim modem business is not a long-term fit for Ericsson. In retrospect, we think that Ericsson and ST Micro’s venture to create a European cellular chip powerhouse was always going to be a challenge, one that ultimately ended in failure, not just because of the difficulty of integrating disparate cultures from different companies, but also because the formation of ST-Ericsson coincided with a drastic decline in the joint-venture’s top customers, Nokia and Sony Ericsson (now Sony).

Sravan Kundojjala