Wireless Industry Analyst

In my overview of Qualcomm’s fiscal second quarter 2008, I mentioned that the company’s broad range of mobile products positions it to exploit the cellular boom optimally. Before we look at the products themselves, it is illustrative to first peek into a few aspects of the mobile market that Qualcomm addresses.

According to Strategy Analytics, the mobile handset market grew 12% y-o-y while CDMA-based handsets (includes CDMA2000, WCDMA, EV-DO, HSDPA) grew 26% y-o-y. It is forecasted that by 2012, there will be about 1.6bn 3G subscribers in the world. The 3G goldmine, being CDMA-based, presents Qualcomm with an opportunity to monetize on every phone using these technologies. (You can find more on this topic here and here.)

While Europe, Japan, South Korea and the US continue to migrate rapidly towards 3G, emerging markets such as India, China and Latin America present equally bright opportunities for CDMA vendors. The challenge is to cater not only to the elitist user in the traditional markets, but also to the utilitarian consumer in the emerging markets looking for low-cost feature-phones.

Mobile computing and data will also be future growth drivers. As wireless networks mature, and convergence devices start to permeate the market, computing on-the-go will become a staple. Data-cards for high-speed broadband in the mature 3G markets will be an important component. Complementing this is the market for embedded broadband modems in internet-ready laptops.

Mobile computing not only opens up the opportunity for broadband modems but also for the core chipset itself. The challenge will be to make smaller, lower-power processors. Intel, for example, is targeting this market with its Ultra-mobile PC and the mobile internet devices (MID) market. The company hopes to revolutionize computing through the MIDs for the mainstream market targeting social networking and location-based services among other things. Apple recently acquired PA semi signaling its seriousness about convergence devices. As the mobile-phone packs more and more features and laptops become more nimble and mobile, the so-called convergence devices, as signified by such industry events, will emerge as the future of communications.

For Qualcomm, each of these product areas presents opportunities that can steadily grow its revenue. The company not only stands to make money on its own chips, but also for every CDMA-based chip sold in the coming years. Apart from the areas mentioned above, Qualcomm will also benefit from its portfolio of connectivity solutions (GPS, Bluetooth, mobile TV, and to a certain extent, WLAN.) In my next article on Qualcomm, I will focus on its strategy to address these market segments.

[Long Qualcomm at the time of writing]

[Originally for Sramana Mitra's site]

I value Marvell at $21 per share. As we have evaluated in this series, the company draws its strengths from the stable revenue coming from its storage, ethernet and WLAN businesses. Its primary weakness is its perceived inability to control its expenses that have resulted in a poor profit/loss record despite growing revenues. The growth driver is its application processor business that will benefit from 3G and smart-phone market growth.

If Marvell plays its cards right, its wireless business can grow to contribute around 40% of its revenues in five years. This will come at a CAGR of up to 30% supported by a sustained market share in the application processor space. The communication processor and the connectivity solutions businesses will help sustain the application processor market share. This requires Marvell to actively develop competitively priced high-performance solutions and in-house platforms in the future.

‘Tavor’ is a very good strategic initiative. It also needs to successfully ship its WLAN, Bluetooth and FM combo-chip. If it can deliver on its envisioned low-cost GPS solution, perhaps as a single chip with the other connectivity solutions and also integrate these in a platform, then it will be a force to reckon with. Perhaps, as I have alluded to here, a merger with SiRF may work out well for the company.

These strategic initiatives will not make much more financial sense if Marvell fails to trim its enormous expenses. I have assumed that it will be able to turn around and cut its operational expenses to about 35% of its total revenues. If the company fails to do so rapidly to achieve nominal expense targets, then the valuation will fall down to $9. Today, its operational expenses are significantly higher resulting in a loss for fiscal 2008. It is true that the agreements to source from the Intel foundries have resulted in some efficiency loss. But I don’t think they alone account for the high expenses. It is becoming clearer to me that Marvell has to further trim its work-force to at least demonstrate a commitment towards making itself more nimble.

In summary, if Marvell keeps a tab on its expenses, its value will double up. If it can also gain baseband communication processor market share, its value can increase further. The aggressive management’s iron-hand administration and the Sutardja family’s personal stake in the company will drive both efficiency and business development. So, while $10 is a good price to pick up the stock, the realization of my valuation will depend on how quickly Sehat and his team can achieve this turnaround. For those who own this share, be ready to let go between $15 and $20 if you don’t hear of design wins for its single-chip XScale solution or if you don't see efficient expense management. For those of you hoping to make a quick buck, I would suggest looking elsewhere despite this stock’s undervaluation just given its uncertainty.

[Originally for Sramana Mitra's site]

In the last two parts of this series, we looked at Marvell’s position and strategy in the wireless connectivity solutions market. These solutions serve another important purpose - to make Marvell’s cellular and handheld products competitive and complete. In this part, we will take a look at whether Marvell can successfully leverage its acquisition of Intel’s XScale business to create another line of business that will see the 20% growth that the management envisions.

Marvell offers the Monahan family of application processors based on the XScale architecture. It also offers baseband communication processors that are capable of GSM, GPRS, EDGE and WCDMA. Its solutions are part of the Motorola Q, Motorola ROKR2 and RIM Blackberry Pearl among other phones. Another notable product is the Samsung i780 that features its PXA310 application processor with Qualcomm’s MSM6200.

Marvell today commands about 25% of the mobile application processor market next to TI, but does not have as much of a play in the baseband market. The Santa Clara-based company sampled an integrated single-chip solution with HSDPA data capabilities and an application processor last year. Marvell hopes to see ramped up volume production and revenue in the second half of this year.

‘Tavor’, as this single-chip solution is code-named, will hold the key to Marvell’s ability to retain the RIMM account moving forward. Moving forward, Marvell hopes to grow with RIMM that commands around 50% of the smart-phone market share. This will be contingent on RIMM retaining its market share and also Marvell delivering competitive baseband solutions. You can read about competing baseband solutions here. ‘Tavor’ will also help Marvell retain its application processor market share as competitors move towards similar single-chip solutions.

Marvell also aims to bring smart-phones into low-cost mainstream leveraging its core competencies. CEO Sehat’s comments on the company’s GPS strategy further ratify this line of thought. The ASUS 3G smart-phones branded as Vodafone 1210, which use Marvell’s communication and application processors, is a good example. The synergy in this relationship comes from the fact that both Marvell and ASUS are committed to low-cost products. (Incidentally, this low-cost philosophy is one solution to bring the convergence movement to the emerging markets. You can view this as complementary to the elitist band of phones from Apple, RIMM or Palm like Ms. Mitra points out in her recent post here.)

At the moment, Marvell is relatively behind the leading chipset vendors in its development efforts. It may hence not get a favorable look from existing handset vendors as it seeks a meaningful baseband communication processor market share that will justify its XScale acquisition. However, it will look to establish new relationships (like ASUS) to grow with its customers’ market share gains. Much like Broadcom, Marvell has a good relationship with PC/laptop vendors like Dell. This positions the company to understand and help these vendors enter the mobile space should they decide to. Bolstered by stable application processor sales, this risky strategy has huge upsides with relatively minor downsides.

In my last post on SiRF, I mentioned that SiRF will likely merge or get acquired. For those of you missed it, you can read it here. I finished with a teaser suggesting a merger with Marvell as one of the options. Let us look at why this is possible. Before you read on, I should warn you that this is just a speculation. So take it with a pinch of salt and more of a synergy analysis.

To reassert my point on SiRF, the company might have stayed a little too long as a GPS company without acquiring other connectivity solutions or pairing up with a complementary solution provider. With mobile driving GPS growth, SiRF may find itself marginalized in this space despite its leadership. You can find earlier coverage on this topic here, here and here.

Marvell, on the other hand, lacks GPS capabilities. Sehat Sutardja, Marvell’s CEO, suggested that GPS has to be commoditized to have only an incremental cost impact on the mobile phone silicon. All said and done, if it cannot have a competitive GPS solution in the next couple of years, it will lose its baseband communications processor customers. The integration of application and communication processors also implies potential share loss in the application processor market. You can read more on Marvell’s connectivity solutions capability in my valuation series here and here.

If SiRF merges with Marvell, the latter will have a complete portfolio of connectivity solutions complementing its WLAN, Bluetooth and FM capabilities. It will also get DVB-H for mobile broadcasting. This will allow the Santa Clara-based Marvell to develop its own mobile platform if it finds relevant demand. I agree that Marvell already has enough financial headaches and operating expense issues to take care of. These will certainly be deterrents against any new merger. But we should note that SiRF is a fairly nimble company. So, Marvell should get favorable margins from the fabless SiRF.

The combined entity can definitely benefit from synergies. Both companies have RIMM as its customer. But in the future, this position will be challenged given aggressive pricing and bundling from the leading vendors. Broadcom, both companies’ main competition, has the potential to displace both of them from these accounts. A merger can help ward this situation off. Marvell’s efficiency can help lower the costs of GPS as it envisions. Besides, SiRF’s strong IP position in GPS will give the company good leverage during licensing discussions.

Finally, while I don’t want to read too much into it, I cannot ignore the Banatao connection. Diosdado Banatao, is now the chairman and interim CEO of SiRF. Given his reputation as a successful serial entrepreneur, his appointment as the interim head itself seems to point to an M&A. it also makes the Marvell angle more likely. He was intimately involved with Marvell as an early stage investor and as a board director until 2003. He knows the Sutardjas very well. He knows Marvell’s work culture, strengths and product portfolio and should have a better sense of the synergies between these companies than anyone else out there. So we can trust him to at least give this idea a good thought.

In summary, it is possible that SiRF and Marvell may merge. It may not be a bad strategy for either company. I will leave the price of such a deal to a SiRF valuation analysis that I am planning to do soon.

(For the interested reader, I have done an extensive valuation analysis of Marvell which will complete this weekend. You can read the first few articles of this series here, here and here)

[Long SiRF at the time of writing]

Forbes.com reported late Tuesday that Apple (AAPL) has agreed to buy PA Semi, a 150-person microprocessor design company. Coming on the eve of its quarterly earnings announcement, the deal rumored to be worth $278 million has caught many chipset and handset vendors by surprise.

PA Semi was started in 2003 to design and develop high-performance low-power PWRficient^TM processors targeting high-performance embedded computing and control markets. The team is headed by the visionary designer Dan Dobberpuhl and is backed by Bessemer Venture Partners, Highland Capital Partners, and Venrock Associates. The 64-bit multi-core processors, the company claims, is “three to four times lower power than other similar processor platforms available today.”

I will leave further details of this deal to the frenzy of articles you will find in the web tomorrow. To me there are two very interesting questions to be addressed. Why should Apple buy PA Semi? What are the industry ramifications? I will pen my thoughts on the first question here and leave the second to a sequel.

The Apple spokesperson declined to comment on the plans and purposes of the acquisition. But there are definitely some obvious angles that I can think of.

Firstly, Apple knows very well that the future is about mobility, future is about technology convergence. Mobility, in turn, is limited by power and battery life. Whether it is the iPhone, the iPod or the Air, Apple has pushed its vendors towards efficient power management. Moving into the future, the company is perhaps looking to use this as a further differentiator from the innumerous clones springing up in the market. It wants to be the only one to leverage the design superiority of these PWRficient^TM processors for most, if not all its mobile platforms.

Secondly, PA Semi’s homepage proclaims the words ‘green computing’ in bold letters. Combine this with the timing of the announcement – Earth Day – and you get the message that Apple is trying to drive. It is committed to low-power, it is committed to going green.

Thirdly, an in-house application processor development unit, especially for Apple, will enable tighter and more efficient integration. This will also allow for further form-factor and power optimization. Customized software applications that enhance the user experience can be written. Furthermore, the internal development and optimization strategy will make it difficult for the copy-cats to reverse engineer. This certainly creates more value for any mobile product that comes from the Apple stable.

In summary, Apple has made this acquisition as a very well-thought out strategic initiative. While it is unlikely that we will see PA Semi-enabled products in the next year or two, you will likely find it as the application processor not just for the iPhone, but for most of Apple’s future convergence devices – mobile phones, music players and laptop replacement devices. In a sequel, I will look at what this means to the other chipset vendors sharing the luscious mobile pie.

(For the interested reader, I have covered iPhone and its impending 3G successor in great detail in my site. You can access it here, here and here.)

Michael Canning resigned last week as the CEO of SiRF, the leading GPS silicon provider. Besides the events leading up to it, the announcement itself reaffirms my belief that SiRF’s end-game, whether it likes it or not, is to merge or get acquired.

As early as last summer, when SiRF’s shares were soaring, I questioned its strategy of focusing on stand-alone GPS solutions unmindful of the bigger trends towards integration and convergent connectivity solutions in the mobile world. Mobile companies like Broadcom and NXP (which recently got acquired by STM) acquired smaller GPS companies. SiRF, on its part, did not attempt to pair up with WLAN or Bluetooth solution providers but instead focused on a DVB-H product for mobile broadcasting. Contrast this with Atheros, a WLAN provider with comparable market cap last year, which has expanded its portfolio to have Bluetooth, GPS and Ethernet over the past couple of years and you will notice a flaw in SiRF’s thinking so far. You can read my thoughts on the Broadcom’s Global Locate acquisition here and Atheros’ GPS moves here.

Despite the booming GPS market and its acquisition of Centrality last year, SiRF may have just let the mobile market slip away from its hands. Broadcom especially seems to be doing a stellar job in stealing customers away from SiRF. The loss in market share led to bad quarters and the company’s share price has since tumbled to its nadir. While I think that this price drop is a little harsh, it does surprise me that most people did not see this coming. More insights and explanations about the SiRF strategy can be found here and here.

Now, with Qualcomm, Broadcom, STM-NXP and TI having GPS capabilities, SiRF has limited room to maneuver in the mobile space. Additionally, companies like Atheros and CSR are broadening their portfolios to fill in the gaps. Faced with the increased possibility of being marginalized, it looks like SiRF will have to align itself with the others who have a realistic chance in the mobile space or with a microprocessor/computer chipset company to sell with laptops.

So far, the deterrent for most companies trying to get SiRF was the latter’s huge market capitalization. That is not true any more. Besides, the company is now on its back-foot. A company like Intel can easily grab the company. But of late, I feel that a merger with Marvell is a good possibility as well. I will let you think about this for a day. Check this space tomorrow for my reasons why.

[Long SiRF at the time of writing]

[Originally for Sramana Mitra's site]

As we continue to analyze the company’s wireless products, it is illustrative to look at its connectivity solutions strategy. In the last segment of this series, we discussed Marvell’s WLAN business. In this piece, I will analyze its Bluetooth and GPS strategies, which I consider as vital elements of tomorrow’s convergence devices.

Marvell hardly has any presence in the Bluetooth market. However, its Bluetooth, WLAN and FM single chip solution, the company claims, is gaining traction. While this will help Marvell retain its WLAN customers, it is unlikely to draw Bluetooth customers away from the incumbents, most notably Broadcom and CSR. Broadcom is in a position to exploit the high Bluetooth attach rate in mobile phones to increase the WLAN uptake through a combined product. Marvell will likely find it more difficult.

Marvell announced a single-chip Bluetooth 2.0 + EDR solution last summer. Marvell is hoping to capture some of the lucrative headset market with this product but may be a little late to do so. I am, however, encouraged that it is pursuing an active R&D effort for Bluetooth which is imperative if it wants to be a successful mobile chipset vendor.

Marvell’s GPS strategy is more interesting. While most of its competitors either already have a good GPS solution or have quickly acquired its capabilities, Marvell has not made any tangible acquisition move as yet. When questioned, CEO Sehat Sutardja did acknowledge that GPS will be imperative for a number of mobile applications in the next two years or so. But he continues to make the point that the company’s GPS roadmap will be cost-driven. Marvell’s strategy is to internally develop GPS solutions that will add incrementally minimal cost to its platform. The company thinks that the success of GPS-based services will hinge on its large-scale adoption in the low-end mobile phones. While this may be posturing to discount its obvious handicap at this juncture, I do acknowledge that it is interesting out-of-box thinking from the company.

Marvell is unlikely to make an impact in the stand-alone Bluetooth market. From the looks of it, it is also not pursuing the GPS stand-alone market. Both efforts are directed towards bolstering the mobile platform components instead. Whether these will pay off entirely depends on the scale of the design wins Marvell can get for its XScale based mobile platforms.

[Originally for Sramana Mitra]

So far in the Marvell series, we have looked at the storage and Ethernet business areas. We also briefly touched upon the overall company strategy. As we move on to dissect the company’s wireless business, we will start with its position in the WLAN market.

Marvell has a good presence in the WLAN market buoyed by the success of its low-power embedded solutions. These products primarily targeting the mobile segment have enabled Marvell’s penetration in the gaming devices and music player markets. Marvell is in Microsoft products, RIMM phones and in the iPhone, of course. Unless Broadcom can steer Apple towards its WLAN/Bluetooth/FM product, perhaps together with a bundled GPS chip as well, Marvell will continue to be in the rumored 3G iPhone. You can read the rest of my iPhone 3G analysis here.

Marvell does not, however, have as big a presence in the PC segment dominated by Broadcom and Atheros. The company is looking to change this situation with its Top Dog^TM product that is advertised to give throughputs up to 600 Mbps. Though it is presumably targeting the high-end laptop market with this product, Marvell also claims traction and design wins in the enterprise and retail access point markets. The company has 10-12% of the WLAN market share today. Most of this comes from the high-growth cellular and mobile consumer electronics product segments. I estimate that the WLAN business, that has grown around 20% y-o-y recently, contributed about 10% of the company’s fiscal 2008 revenues.

WLAN as an enabling technology will grow at about 15% CAGR for the next few years. So, Marvell’s ambitions of retaining its current growth in WLAN depend on its ability to retain and grow its market share. This, in turn, will depend on four things.

Firstly, the company’s 11n strategy of going for performance vis-à-vis price should pay off. The incumbents, Atheros and Broadcom, have very competitively priced products and in my mind understand the price-performance trade-off much better than Marvell. Then, you have relative newcomers Ralink coming up with low-cost solutions.

Secondly, even in the embedded space, its sustenance is in question with Atheros coming up with equal or better products both in terms of throughput and power consumption.

Thirdly, of course, is its ability to get design wins for its cellular platforms so the connectivity solutions can be bundled with the baseband chips.

The final factor is its ability to keep pace with the changing demands from both the PC and mobile customers. The Bluetooth, WLAN and FM single chip, for example, is almost imperative for the company to stay ahead of its competitors in the embedded space. The concept is great, but it also demands an equally good execution. The company claims positive traction for this product with product ramps in the second half of fiscal 09. I am anxious to see how this will play out for the next two years.

In summary, it appears that Marvell has betted on some of its new product strategies to sustain its growth. With the competition getting tougher, the company may find it difficult to maintain status quo here. But it has a good presence in the cell-phone and mobile consumer electronics segments. It also has a great track record of delivering on its growth so far. So, if the company executes on its WLAN product roadmap, I am willing to conclude that the company will still be able to achieve about 20-22% growth y-o-y from that business.

[Originally for Sramana Mitra]

In the previous articles in this series, we looked at Marvell’s product strategy, briefly reviewed the fiscal 2008 financials before dissecting the storage business area. Let us now take a look at Marvell’s position in its Ethernet semiconductor business.

The Ethernet business is yet another strong-point in Marvell’s product portfolio. The company is a strong second place finisher behind Broadcom in this market with a share of around 20%. Marvell’s switching solutions enable voice, video and data traffic for the enterprise networking, carrier access and small office/home office (SOHO) networking markets. This is well complemented by the Gigabit Ethernet transceivers, communication controllers, the Yukon^TM family of Gigabit Ethernet PC connectivity products and the Link Street Gateway^TM products.

The stability of the Ethernet market and the company’s strategy to pick up accounts which will tap on the company’s operational excellence has together helped it grow its market share so far. The Ethernet business area has been growing steadily at about 20% for Marvell in the past few years. I estimate that this business area contributed to around 20% of Marvell’s fiscal 2008 revenues.

Well equipped with one of the most rounded product portfolios in this business area, Marvell is looking to acquire more market share from Broadcom. With some aggressive pricing and bundling strategies, the company appears to have made some inroads into key Broadcom accounts over the last couple of years.

The competition has also widened and threatens Marvell as much as it does Broadcom. However, the competitive landscape is such that there will be design ‘steals’ going back and forth resulting perhaps in a stable aggregate market share for its leaders. So, moving forward, I expect Marvell’s Ethernet business area to grow at about 9%. With the Ethernet business itself growing modestly at 5.5-6%, this translates to slight market share increase. For the interested reader, I will recommend reading my piece on Broadcom’s enterprise networking business area where I have detailed the Ethernet market and the competitive landscape.

In summary, the Ethernet business area will continue to provide another stable revenue stream for Marvell. I do not, however, expect this to be a bigger than average growth driver for the company.

Two blogs I read carried some coverage on the LTE IPR agreement that I wrote about here.

GigaOM pointed out that there will be a push-pull effect with the IP developers on one side and the operators and handset vendors on the other. But, it is perhaps unfair to characterize Qualcomm's IP position as the cause of 'pain and suffering'. I can see why other companies are seething at Qualcomm’s demands – the San Diego company takes a large chunk of their margins. But then, it makes sense from Qualcomm’s perspective and even for the US telecom industry as such. If there is one US Company that is thriving and very successfully holding its own amidst the dominant European presence in the mobile space, it is Qualcomm. It is the high margin IP-based revenue that has allowed the company to get here. The company diverts this revenue back into state-of-the-art research that advances mobile communications. At least that is its business model and that is what the others are trying to break through the legal battles.

Dean Bubley also had some good points that that I concur with. The IP-rich companies are ominous in their absence. Besides, it is a good way of positioning LTE as the fore-runner in the 4G race. Like he says, it will be interesting to see how the WiMAX world reacts to this.

Talking of WiMAX, did anyone notice Nextwave in this list? What is a company that put its weight behind WiMAX not long ago doing here? Does it say something about the future of WiMAX or the winner in the 4G race?

[Long Qualcomm at the time of writing]

An LTE patent framework is being pushed by a handful of companies to restrict the total IP licensing fees to single digit percentage of the handset ASP. Companies include Alcatel-Lucent, Ericsson, NEC, Nextwave, Nokia, Nokia Siemens Network and Sony-Ericsson. The announcement further calls upon interested parties to join this alliance to ‘stimulate early adoption’ of LTE. Though I can see the reasons for such an arrangement, I am not very optimistic about its large-scale success.

Part of the problem with LTE or any OFDMA-based standard is the diffused nature of the IP. Many more companies have a stake in the IP pie now than for example, in WCDMA in which Qualcomm has close to 30% of essential IP. So, if there is no upper bound on the royalty rates, the costs can potentially be more prohibitive than the existing CDMA standards and will have a direct impact on the handset ASP. You can read more about my OFDMA, LTE and WiMAX coverage here and here

Another thing to note is that this has been tried before in the 3GPP with WCDMA way back in 2002. Nokia, NTT Docomo, Ericsson and Siemens were among those who initially agreed to cap the cumulative royalty at around 5-6% of the handset ASP. The move did not generate widespread interest. Qualcomm, InterDigital and others with a strong IP presence have stayed away from such deals and are now reaping rich rewards. The WCDMA total royalty rates have hence remained much larger than the envisioned single digit numbers. You can read more about the IP strengths of Qualcomm and InterDigital here and here.

The handset vendors and carriers want to make sure that no one company becomes an Achilles heel in the productization and large scale proliferation of these technologies. However, even if chipset vendors join the bandwagon, these standards will still run the risk of an outsider company staking its claim on essential IP. So, while the 3GPP has tried hard not to repeat the 'Qualcomm effect' again for LTE, it will be interesting to see how this strategy plays out given the more diffused nature of OFDMA IP.

[Long Qualcomm at the time of writing]

In my last article on the STM, NXP merger, I analyzed the JV from STM’s perspective. The merger is a clear sign of further consolidation in the 3G space. To understand this further, it is also illustrative to look at how other players in the vendor space can be impacted by this merger.

The importance of this merger primarily comes from my thesis that Texas Instruments has fumbled in its 3G strategy, leaving its huge market share wide open for other vendors to steal from it. Especially at risk is its huge Nokia account. The Finnish handset maker has already moved towards a multiple vendor strategy sourcing 3G chips from STM. Last year, it committed about 200 engineers to STM for the latter’s 3G chipset development program. Nokia is essentially focusing away from chipset IP and development. This has worked against TI and into the hands of STM and Broadcom (whose EDGE solutions are now being sourced by Nokia.) You can get further insights into TI’s wireless strategy and its potential pitfalls in my valuation series here, here and here.

The JV can further consolidate this relationship between Nokia and STM. Further, the complete portfolio of connectivity solutions will make it very attractive for the future smartphones from Nokia and the others. So, the JV will likely add to TI’s wireless woes. The impediment that I see is performance. The JV will still be behind Infineon, InterDigital, Icera and Qualcomm on performance. The hope is that a renewed R&D thrust will help narrow the gap in future designs.

Broadcom is perhaps another vendor for whom the merger can cause headaches. The Irvine-based company has in the past made its ambitions apparent – to work diligently towards the coveted third spot in the mobile vendor matrix. Broadcom has been very aggressive in its mobile campaign, not just with product announcements but also in its legal battles with Qualcomm to defend its IP position. The bottom-line is that it is also looking to grab market share from TI. If the JV can improvise on the connectivity solutions and build a stable and complete platform, it will give Broadcom a run for its money. But Broadcom will still have the time-to-market advantage for now. You can read more on Broadcom’s wireless outlook and my valuation analysis here, here and here.

Qualcomm is less likely to be threatened by the JV though the latter now has access to Samsung. Qualcomm’s technology leadership, its support network and its longer-term view of the mobile space places it on firmer ground. Infineon, which is likely to be at the heart of iPhone 3G, will now be a distant second in the European vendor matrix. InterDigital, Infineon’s 3G partner, will have to rely on performance as it tries to gain in the smart-phone market. Marvell and Icera among others will also come up with niche selling points to counter such consolidation.

So, while the price STM paid for the merger can be debated, there is no doubt in my mind that it is a consolidation that will send some vendors scampering. I have a lot of questions in my mind now. Will someone pick up Freescale as well? Will TI target an acquisition that will give it 3G baseband capabilities, perhaps InterDigital or Icera? Will InterDigital and Infineon further formalize their strong alliance to gain scale? Well, there is consolidation in the air. I can’t wait to see how the vendor matrix ends up a couple of years from now.

ST Microelectronics (STM) and NXP announced last week that they were going to combine their wireless chip manufacturing capabilities to form a joint venture. This is an industry consolidating event that may well define the direction in which the mobile value chain is headed.

As per the deal, structured as a merger, STM will have control on the JV with about 80% stake. The company will pay NXP $1.55bn to close the deal in the third quarter. The merger combines two of the bigger wireless chipmakers, which together had about 10 percent of the global market in 2007, according to iSuppli. STM CEO Carlo Bozotti claimed that the JV will create a global leader with a more optimistic 14% market-share in wireless chips.

The companies’ combined wireless revenue for 2007 was about $3bn. They reported that the synergies are expected to save up to $250mn in 2011. Nokia, Samsung and Sony-Ericsson are among the customers for the JV. The company will primarily compete with Broadcom , and to a lesser degree with Infineon and Freescale, hoping to take market share away from TI. You can read more about TI’s precarious position in the mobile space and how such events aggravate this situation from my valuation series here, here and here.

Bozotti justifies the JV as a move to compete with the giants (Qualcomm and TI) on scale. “This is a business where scale is fundamental,” he says. The economy of scale, both in R&D and manufacturing, gives the leaders substantial ASP gains. The JV will also have a stronger IP position than either STM or NXP. This, in turn, gives it greater negotiating power in IP cross-licensing discussions further increasing its margins.

Besides establishing a strong R&D team to develop 3G solutions and beyond, the joint venture has what I would call a complete portfolio of connectivity solutions. These technologies – WiFi, Bluetooth, FM and GPS - will be integral to tomorrow’s convergence devices. This allows the company to chalk out a product roadmap that integrates these solutions into a state-of-the-art single stop platform that will be attractive to handset vendors.

The wireless industry, especially the 3G chip vendor space, will consolidate and the stronger players will pick themselves out of the crowd. The JV is STM’s statement of arrival. Over the last couple of years it has made tremendous strides culminating in design wins from Nokia and Sony Ericsson. The JV will perhaps not make an immediate impact in its product line. It, however, expands its customer base and also positions STM to exploit the convergence market over the next few years. The need now is to make use of the synergies and deliver on its design wins. The need now is for agility and execution.

[Long Qualcomm at the time of writing]

[Originally for Sramana Mitra's site]

In the prequel, I looked at Marvell’s fiscal 2008 financials focusing especially on the company’s expenses. As we move ahead, let me dissect its position in the storage industry in general and the hard disk drive market in particular.

Marvell offers a wide variety of storage products for hard disk drives, tape drive electronics, optical disk drives and storage subsystems technology. The portfolio consists of discrete components such as the preamplifier, read channel, hard disk controller, and tape drive controller and also System-on-Chip solutions efficiently incorporating multiple components into a single integrated circuit.

Marvell is one of the leaders in the storage industry. It has about a third of the hard disk drive silicon market share supplying solutions to Western Digital, Samsung, Toshiba, Hitachi among others. All credit to the company’s management for having successfully gotten Marvell this far despite starting as rank outsiders in this area. I estimate that storage products, revenue from which have grown over 30% y-o-y for Marvell, today account for over 50% of its revenue. The company also seems to have made the right moves with the mobile 2.5 inch hard drives. The mobile hard disk drive business has thrived in recent quarters increasing sequentially by 39% in unit shipments due to favorable market share shifts.

Though the competition has consolidated with the 2006 merger of LSI and Agere and the more recent move by the former to purchase Infineon’s storage assets, Marvell’s success story in storage will continue. Helping it is the ASP expansion coming from the transition to SoC products from discrete components. Besides, the market share shifts among the hard drive vendors, especially in the 2.5-inch mobile drives, have helped Marvell’s customers. Thirdly, the mobile drive market is itself accelerating rapidly. These should come together with the 10% y-o-y HDD unit growth rates and Marvell’s R&D efforts to sustain growth.

In summary, despite the cyclical nature of the HDD market and the consolidation of the competition, Marvell has mastered the space enough to lead the way. In the medium term, this business will continue to grow at 20% y-o-y helping provide Marvell a stable revenue source.

[Originally for Sramana Mitra's site]

In my previous article on Marvell, I provided a brief overview of the company strategy and product portfolio. Before we move on to dissect the various business areas, let us take a brief look at the fiscal 2008 financials. Marvell’s total revenue for fiscal 2008 is $2.9bn at a gross margin of 48.7% and an operating loss of 3.6%. This represents a 29% increase from fiscal 2007’s revenue of $2.24bn. The net revenue increase is due to increased volume shipments of storage SoC, a full year’s revenue from the Intel acquisition and the printer business from the Avago acquisition.

The gross margin decrease from 2007’s 50.8% is primarily due to the lower margin cellular and handset products acquired from Intel. Marvell was contractually obligated to purchase from Intel under a supply agreement. With the terms of this agreement being met, the company has partly transitioned to its fabrication partners and is looking to get incrementally higher margins this year.

The operating loss was incurred mainly due to the 50% increase in the R&D expense. The company spent $971mn on R&D as compared to fiscal 2007’s $645mn. This comes directly from the additional salary and related costs of $161mn (about 5.5% of revenue) incurred in fiscal 2008 related to the Intel and Avago acquisitions. Fiscal 2007’s R&D expenses were also 80% higher than the fiscal 2006 numbers due to the personnel increase coming with the Avago, UTStarcom and Intel acquisitions. While the lay-offs and restructuring measures may bring the R&D expenses back to the 30% range, the total operating expense will be about 47%. This is alarmingly high as compared to the semiconductor industry average about 32%.

Marvell’s current cash and cash equivalents stand at about $600mn, 20% of its annual revenue. Broadcom, its similar-sized competitor has a cash position of about $2.3bn or 60% of its annual revenue. Western Digital accounted for 17% and sales to Asian customers represented 84% of fiscal 2008 revenues.

While Marvell has a broad portfolio, a significant portion of its revenue comes from the highly cyclical and competitive hard disk drive industry. Besides, even for a fan of long-term vision and strategic spending like me, the expenses and their outlook are staggering and out of proportion. But for fiscal 2006, when the OpEx was about 38% of the revenue, the company has in recent years had 45% or greater OpEx. It seems that Marvell’s growth comes with very high associated costs. Sehat reiterated in the recent conference call that the company is committed to a reduction of costs. I wonder how quickly and efficiently this can be put in place.

Marvell provides application-specific standard products (ASSPs). The company has substantial expertise and intellectual property in the areas of DSP, embedded ARM-based microprocessor, analog and mixed-signal integrated circuit design. These strengths are responsible for Marvell’s phenomenal rise in the storage market. Today, the Santa Clara-based company (incidentally, Atheros’ back-door neighbors and competitors) leverages this expertise to develop high-performance System-on-Chip (SoC) solutions for various markets.

Marvell’s product portfolio includes solutions for data storage, Ethernet, cellular, wireless networking, personal area networking, video-image processing and power management solutions. Marvell serves a variety of markets including convergence of voice, data and video in consumer electronics goods. Marvell’s customers include storage companies such as Samsung, Toshiba and Western Digital, networking companies like Cisco, Juniper Networks and Foundry Networks, and cellular companies including Research in Motion (RIMM), Motorola, and Palm.

Marvell, in the words of its CEO Sehat Sutardja, is “inherently a product cycle company.” He continues to say that the company’s growth rate is directly related to new product adoption and transition. The strategy is to look at long-term investments that can grow on a top-line basis at 15-20% y-o-y. In the context of the current situation, this directly relates to Marvell’s thinking behind the acquisition of Intel’s applications and communications processor business.

We can divide the company’s products into three broad business areas namely storage, Ethernet and wireless. Besides these, other major products include solutions for VoIP, printing, digital video processing and power management. With this as a background, we should be able to take a deeper look at Marvell’s fortunes moving forward

The web is abuzz with the news that the latest iPhone 2.0 beta software revealed whose chipset was at the heart of the impending 3G iPhone. The fairly innocuous and cryptic word ‘SGOLD3’ embedded in the code points to Infineon’s SGOLD-3H baseband processor. While this is news for many enthusiastically expecting the 3G iPhone, it is only a confirmation of what I have been saying since last fall – Infineon and InterDigital will be at the heart of the next iPhone.

The Infineon SGOLD-3H together with the MP-EH platform is a natural solution for Apple. Infineon’s SGOLD2 and the MP-EU platform form the basis of iPhone V1 and Apple has spent considerable time and energy optimizing it for performance, stability and power consumption. It makes sense to carry the good work forward with MP-EH which looks very identical to MP-EU rather than reset it with another new platform. You can read the rest of my prediction analysis from last fall here and here.

While Infineon’s SGOLD-3H appears to have been the default solution for the 3G iPhone, Apple definitely will not regret its choice. The Infineon solution is a Category 8 HSDPA solution, meaning that it supports incoming data rates (equals download speeds) of up to 7.2 Mbps. It has a fractionally spaced equalizer implementation of the receiver (which I think is predominantly InterDigital’s IP) that allows the realization of nominal data rates in mobile conditions. Not surprisingly, Infineon was one of the three companies picked as the 3G performance leaders by Signals Research. You can read more of my SGOLD3 musings here.

It is also important to acknowledge the other winners if the MP-EH and SGOLD-3H are in the next iPhone. First and perhaps the biggest will be InterDigital. Infineon’s 3G solutions use InterDigital’s stack. InterDigital is bound to get per-unit royalty and its deal with Apple last summer is a good pointer towards this direction. The King of Prussia company is Infineon’s behind-the-scene partner for 3G. My extensive coverage of the InterDigital connection can be found here and here.

Marvell, I think, will continue to supply the WiFi provided the interoperability issues it faced with its previous offerings are ironed out. With the Garmin Nuvifone round the corner, it is quite possible that iPhone will have GPS too. Broadcom is in a nice position to offer a bundle of its WiFi, Bluetooth and GPS offerings and may come out a big winner despite losing on the baseband front. As with the chipset itself, I expect Apple to retain most current component vendors who have met Apple’s serious performance expectations last time around. More on the iPhone 3G potential features can be found here.

I am glad that my iPhone 3G predictions are expected to be true. While the actual winners will not be determined until a tear-down, this is good intermediate information. Infineon is trading close at $7.37, close to its 52-week low of $6.26. InterDigital is trading around $20 today. If you have not read my InterDigital valuation series, this is perhaps a good time to do it from here. With the details of the iPhone 3G consolidating, now is perhaps a good time to pick this stock up even if you are looking at some quick short-term profits.

[Long InterDigital at the time of writing]

Bristol-based Icera Semiconductors announced today that it was acquiring Sirific Wireless for its complementary mobile RF capabilities. With this all-share merger for an undisclosed sum, Icera now hopes to offer a more complete 3G+ mobile chipset. (Read earlier coverage of Icera here and here.)

Sirific, founded in 2000, is a fabless semiconductor company developing CMOS RF transceivers for various mobile technologies including 3G and beyond. It boasts of expertise in digital RF, one of the more difficult design problems in mobile circuitry. It has raised around $60mn till date.

With a series C funding closed recently, Icera has now raised a little over $140mn in venture funding. So, the combined entity has over $200mn in venture money – an impressive amount demonstrating the VC confidence in this startup. Icera plans to raise a further $100mn later this year positioning itself for an IPO at the London stock exchange in 2 years.

Is it justified in dreaming big? Why go for Sirific?

Well, Icera has a niche product with its soft modems. Soft modems are highly customizable and upgradeable. The technology and the concepts involved, if rightly done, can be used for any wireless standard allowing for easier integration that is vital for tomorrow’s convergence devices. Icera’s thinking is that Sirific’s RF frontend and digital RF expertise will complement its revolutionary Livanto soft modem solution. This will not only allow Icera to come up with more complete, in-house platforms and reference designs but also offer a roadmap to highly integrated baseband + RF products in the future.

Icera is beginning to gain traction in the wireless data card market that it is currently targeting. It has signed a deal with Option, Europe’s laptop wireless data card leader and aims to capture 33% of the global data card market by the end of next year. But with the move towards in-built wireless broadband capabilities, products such as Qualcomm’s Gobi and the convergence devices phenomenon, data cards will become superfluous in the future. So to create a larger impact, the company has to make it big in the higher volume mobile phone space.

It has, however, not gotten nearly as much traction in the mobile space. What it needs is a strong alliance with a company like TI that offers application processors. Like I mentioned for InterDigital, such an alliance will allow it to piggyback on TI’s support network and economy of scale to win market share for its baseband (now baseband + RF) business. In turn, this will also help TI retain its 3G market share.Earlier coverage on this topic can be found here, here and here.

Icera has the technology to take it into the future. It needs to however sustain in the fiercely competitive mobile vendor space. The data card market will help on that front in the short term. Besides positively approaching the mobile phone market with the strategic acquisition of Sirific, Icera has certainly made a bold statement indicating that its IPO ambitions are real. The flotation valuation of such an IPO will be determined much by the design wins it secures in the mobile broadband space and perhaps on a few other strategic alliances.

[Originally for Sramana Mitra's site]

In the previous article, I discussed what an acquisition of InterDigital can do to TI. What then will happen to the cellular industry and the 3G value chain?

If TI acquires InterDigital, I think there will be further market consolidation in the 3G space. Firstly, Nokia and Motorola will be happy that they can compete with ‘Qualcomm inside’ phones from Samsung and LG. Until Nokia and Qualcomm cool off on all their IP issues, it is unlikely that the Finnish company will get chips from Qualcomm. So, the entire Nokia business, which until now was TI’s monopoly, is now wide open. An InterDigital acquisition is one way for TI to retain this market. Also, Samsung and LG may also get a viable alternative to pursue a multiple vendor strategy.

If TI acquires InterDigital, its main competitors will feel added pressure moving forward. Qualcomm, the undisputed performance leader so far, is now being challenged by InterDigital, Infineon and Icera. The Qualcomm advantage still comes from its unsurpassed support network and the scale of operation. But TI can complement InterDigital to match Qualcomm on these two fronts as well, thereby erasing its competitive advantage.

If TI acquires InterDigital, Broadcom will take a bigger blow as its wireless strategy seemingly rests on capturing TI’s market share both in baseband and in the application processor market. Infineon, which currently has a strong alliance with InterDigital, and which I suspect has key elements of the latter’s 3G advanced receiver IP, will run the risk of being marginalized in 3G. Similar arguments can be made for Marvell, Freescale and ST Microelectronics among other chipset vendors.

If TI acquires InterDigital, it can potentially generate up to $10 billion of additional revenue over the next five years. The handset vendors like Nokia will be happy to see crucial 3G patents in the hands of ‘one of them.’ This will be an important factor securing TI design wins into the future. Assuming that a deal is closed at my $75 valuation of InterDigital, the sale price will be around $3.5bn. Under this condition, I will value TI between $36-$37.

Clearly, the two companies have distinct cultures and business model. But if these issues are dealt well, the combination can be explosive. What amazes me about InterDigital is the immense value it offers to companies with completely different business model. If it considers being acquired, it will be hot property. While I am not predicting an acquisition here, I think this is an option that TI should think about. Perhaps it is thinking about InterDigital, or even Icera, or has an accelerated in-house 3G/HSDPA program. Or it is just waiting out for the 4G promise, dodging the 3G curveballs thrown at it in the interim.

[Originally for Sramana Mitra's site]

In my recently completed series on Texas Instruments, I pointed out that TI was not in a great position as far as 3G is concerned. I subsequently also suggested that the company ally with InterDigital to prevent further damage. Let us take this a step further and examine what it will mean for TI to acquire InterDigital. I understand it takes much more than a 500 word essay to make an acquisition happen, but a quick look in this direction will give some new insights into the wireless industry as well.

If TI acquires InterDigital, it will have a viable path towards sustained 3G baseband market share. It will have a quicker time-to-market for advanced HSDPA products. An OMAP-Vox Solution with the 3G HSDPA baseband receiver from InterDigital and its own OMAP can be a competition killer. Even a bundle of IDCC’s SlimChip solution with OMAP will sell well. The new product can marry TI’s manufacturing excellence with the SlimChip’s proven performance. The interested reader can find a more detailed analysis on the combination product here in my blog.

InterDigital’s ASIC business is a small portion of its value. But for TI, the combination of 3G intellectual property (IP) and a competitive product is most appealing. In contrast to InterDigital, TI has never had an aggressive IP strategy. It has, however, leveraged its IP to obtain favorable cross-licensing deals with Qualcomm and the others. If it acquires InterDigital, TI can use its new-found IP position to further improve its margins by getting even better licensing deals. More importantly, it will have the handset vendors once again queuing at its gates for their next designs.

An acquisition will work well for TI. In the sequel, I will look at how the industry value chain will take it if TI made this strategic move to get InterDigital and also give some numbers for such a deal.

[Long IDCC, QCOM at the time of writing]

Recently, Signals Research Group published an independent performance report comparing HSDPA solutions from various vendors. The test results, coming as they did from a credible firm, surprised many by concluding that InterDigital(IDCC), Infineon (IFX) and Icera were the winners. I was not surprised though.

Over the past few months, I have extensively covered InterDigital and its SlimChip solution. The SlimChip’s advantage comes from its advanced baseband receivers with firm theoretical underpinnings. Much like Qualcomm, InterDigital has a strong systems team that designs the optimal system first and then seeks to make the cost-performance trade-off. The SlimChip is well-positioned to complement the market-leading applications processors such as TI’s OMAP to break into the smart-phone market. For more coverage on InterDigital, I will refer you to my earlier articles here, here and here.

Infineon is also not a surprising topper at least for two reasons. Firstly, I am betting that Infineon will be in the impending 3G iPhone. The Apple angle implies that Infineon has to match the former’s push for innovation with a high-performance, low power chipset. Secondly, Infineon has a very friendly alliance with InterDigital. The Infineon chipset uses InterDigital’s 3G stack. Besides, I suspect that InterDigital has provided substantial baseband IP for this chipset including the advanced receiver algorithms that have made SlimChip superior. You can read my iPhone suspicions here and here.

Icera is a Bristol-based startup which has raised over $100 million in venture funding till date. Its Livanto solution is the world’s first wireless soft modem. Along with the Adaptive Software solution, Livanto supports HSDPA,WDCMA, GSM and EDGE. The software solution implies that very high precision advanced algorithms can be programmed without spinning a new chip. This not only allows for extremely good performance, but also implies that the Icera chipset can handle any improvements in the standard or receiver algorithm. You can find earlier coverage on Icera here.

I had suggested earlier that an alliance with InterDigital will do TI a world of good. Watch out for Icera as well. Soft modem is the future. This company founded by Broadcom and ST Microelectronics founders already has a pioneer status and can come at a much cheaper price than InterDigital if TI wants quick 3G capability.

In summary, the three companies are not surprising winners, after all. But the test results have challenged Qualcomm’s Unique Selling Proposition (USP) – its baseband performance. Qualcomm, for the moment, can rely on its economy of scale, service and support to stay ahead. This may also be broken if, for example, TI throws its weight behind InterDigital or Icera.

[Long Qualcomm, InterDigital at the time of writing]