Future of Wireless Connectivity

Replacement of wire with wireless links had been a holy grail for quite sometime. Cordless phones, a relic from 1980’s, is a testament to this urge of human society. With the advent of personal computers, the jumble of wires in an average house has grown manifold. Consumers have waited long for a solution to relieve them from the ever growing hassle of wiring their computers, TV, DVD, set-top-boxes and sound systems. The knowledge gained in the last decade has propelled a slew of start ups to make our world cable free.

For the uninitiated, when we talk about wireless connectivity, we refer to two different types of networks called WLAN and WPAN. For short ranges up to 10m, devices communicating with each other are part of Wireless Personal Area Network (WPAN). Distance larger than 10m is covered under the rubric of Wireless Local Area Network (WLAN). WiFi (based on IEEE 802.11 standard) with around 100 Mbps and Bluetooth with 3 Mbps have come to dominate the WLAN and WPAN space respectively.

Unfortunately, as it exists today, neither the WiFi nor the Bluetooth has the throughput to replace the cables from our high end audio visual consumer equipment. Sensing an opportunity, startups like Quantenna and Amimon have developed the MIMO platform on the foundation of 802.11. They are claiming transmission rates of 0.6-3 Gbps. With the provided chipsets, OEM/ODM can make their DVD, set-top-box and display units wireless and accessible from anywhere in the house.

In the last five years, ultra-wideband (UWB) technology has materialized to address the WPAN. With lower energy per bit than WiFi, UWB has long promised to be a superior technology. UWB based on MB-OFDM can deliver 0.48Gbps over short distances. When the UWB development started, it specified a data rate that was an order magnitude higher than anything that WiFi had to offer. In the interim, proprietary WiFi solutions from Quantenna and Amimon has caught up and more importantly moved ahead of the UWB in terms of throughput. However, it remains to be seen whether these MIMO based solutions can be competitive with UWB on cost and power consumption. At least, in one respect, WiFi do have an advantage and that is its better coverage range.

UWB, so far, has been unable to penetrate the market even with the 2nd generation of products. There is a significant risk that it can be squeezed out of the space by the high speed variants of WiFi from one side and more ominously from the multi-Gbps short range solution from SiBEAM on the other side.

We will have to wait and see who is going to become the next CSR and Atheros in the new generation of wireless connectivity. The heated race is on.

Consolidation in cellular chip suppliers

The recent announcement of the EMP merger with ST-NXP wireless entity is another milestone in the consolidation of the wireless chipset suppliers. A mature industry vaporizes margins and forces consolidation. It had been a forgone conclusion that there were too many players in this exceedingly price sensitive business.

A cursory look at the list of companies competing in baseband ASICs until a few years ago reveals, that more than half have either ceased their operation or had been acquired.

TI
Qualcomm
Freescale(Motorola)
Infineon
Broadcom
Mediatek
Icera

NXP(Philips)
ST
EMP
ADI
Intel
Agere
Silicon Labs
Prairecom
Skyworks

From the list above, Infineon and Freescale wireless businesses cannot survive as independent entities. With the current wave of consolidation, it makes a lot more sense for them to merge. Infineon has a proven ultra low tier portfolio and Freescale has a strong presence in high tier 3G. The combined entity can then count RIM, Apple and Motorola as their customers. Furthermore, operational efficiencies can result from such a venture as they do have overlapping portfolio.

This will then eventually leave three to four major players to fight in the market. Broadcom who has been investing heavily in cellular space, does not have a significant market position. It has a tough life ahead of it. However, to its credit, it has the fighting zeal, and with Nokia on its side may come out on top.

Jha's move to Motorola and phone chipset suppliers

The recent surprising news of Qualcomm's COO Sanjay Jha, to lead the handset division of Motorola will have a significant bearing on the handset semiconductor players.



Jha had been with Qualcomm since early days, he joined the company as a senior design engineer and then rapidly rose through the ranks to lead QCT (the ASIC wing of Qualcomm) as VP of Engineering by late 90' s. In the presence of CEO Irwin Jacob at the helm of Qualcomm (son of the founder Irwin Jacob) , further growth was impossible at Qualcomm. Thus joining Motorola as co-CEO, and the promise to lead the Mobile Device unit's spin off is a smart natural choice.



Motorola handset business had got out of making ASIC's under Mike Zaviroski. The ASIC design was transferred to the semiconductor division in 2002. Since then there had been a drive to adopt the PC makers model, which is, buying off the shelf IC's with more of an emphasis on branding and supply chain management. What prevented the rapid adoption of this model?; the legacy software which was tied closely to the legacy chips. PC manufacturers, luckily did not have to deal with the software issue, they had Microsoft.



In a way, this dependency benefited the incumbent chip suppliers. The only way to dislodge the incumbent was to provide Platform solutions, comprising both hardware and software at a price point compelling enough to justify the switch. This dependency had benefited Freescale (in Motorola), TI (in Nokia), and EMP (in Sony Ericsson) and ensured that the revenue kept on coming as long as the dependency sustains.



Then, with the falling ASP of phones and the competitive products from the competition like Qualcomm and Broadcom, sooner or later a dent was bound to occur in this relationship. Motorola in the recent past announced relation with Texas Instruments, essentially cutting out Freescale from the future mix of mid and high tier phones for WCDMA market. Already, Infineon had been supplying chips for ultra low tier phones and Qualcomm was the supplier for CDMA products.



Qualcomm had been trying very hard for past several years to win the WCDMA sockets in Motorola. A few announcements were made last year, but nothing concrete occurred. Eventually, TI won the high tier sockets edging out Freescale from that business. TI already had a strong presence with their OMAP processors, which facilitated the win. With Jha coming on board, the dynamics of chip suppliers are bound to change. Jha who had been instrumental in pushing Qualcomm inside Motorola, is now inside Motorola. This can have far-reaching consequences for both TI and Freescale. Qualcomm has a very broad portfolio. It is entirely possible, that Freescale will even lose the low to mid tier segment of chips to Qualcomm. Wireless division of Freescale had been for sale, Broadcom backed out of the purchase late last year. With a losing customer base, the sale or merger of Freescale wireless assets will become all the more difficult.



Furthermore, TI will have to marshal all its resources to sustain the newly won sockets inside Motorola. Time will tell, whether it will able withstand the Qualcomm's onslaught but more and more it looks unlikely that TI will remain inside Motorola for long. Indeed, the move of Jha to Motorola, is a triumph for both Qualcomm and Jha.

Battle for Application Processors Market

In 1995, Dragon Ball running at a speed of 16.58MHz was adopted by the new kid on the block, PALM. It created a new niche in the semi-conductor industry called, Applications Processor (AP). Who would have imagined that AP would go on to become an integral component of many embedded systems like cell phones, GPS, and personal medial players.



In nutshell, Application Processor is a highly integrated micro-controller with gazillion interfaces. It can hook up with all sort of memory and storage like DDR, Flash, NOR, ATA. In addition, high-end processors, embed dedicated graphics, audio, video, and image accelerators for lower power consumption. They are targeted for a range of device from cell phones to portable navigation systems.



In the beginning, Dragon Ball was a developed by a small team of engineers in Motorola. The company itself did not anticipate its success. However, as it often happens, success breeds a sense of invincibility and complacency. Motorola ignored the requirements and feedback of its customers. TI and Intel entered the market with a bang with ARM based OMAP and StrongARM/XScale application processors. They offered added qualities of more speed, integration and a broader portfolio. One by one, Palm, Treo and others adopted OMAP and XScale. Along the way, Dell and HP jumped into the PDA business with Intel's AP.



Meanwhile, Motorola was actively developing MCORE as the belated replacement for the venerable 68x core. Feeling the heat from the OMAP and XScale, Motorola finally folded MCORE development and adopted ARM in the newer Dragon Ball rollouts. It subsequently relabeled the product line as i.MX.



The market for AP is big. Over the past few years, many new players have jumped into the bandwagon. Marvell eager to enter the handset market, bought the XScale based PXA portfolio from Intel in 2006 for over $600 million. This had given Marvell a foothold into the handset world.



Video and graphics processing are an integral feature of the high end AP's. Hence, it is natural for the graphics chipmakers like NVIDIA to enter the business. NVIDIA penetrated the market with the acquisition of Portal Player. Portal Player was once supplying chips for the iPod. Post acquisition, NVIDIA has now come out with ARM11 based Tegra APX running at 750MHz. It incorporates NVIDIA GeForce mobile graphics core and can play HD 720p video on a mobile device.

Qualcomm who had been making communication ASICs using ARM cores decided to take a step forward and introduced an ARM based Scorpion core running at 1GHz. However, it is unclear whether Qualcomm will market Scorpion as a standalone AP or will continue to provide it as part of its Snapdragon mobile platform.



Samsung is a late entrant, but had won Apple business by leveraging the Nand Flash portfolio. Samsung had combined the AP and nand flash into a single multi chip package. iPhone and other MP3 players uses Samsung AP's. It is debatable for how long Apple will stick with Samsung. In 2008, Apple bought PA Semi for $278 million with a possibility of replacing its existing suppliers with in house developed chips.



Many players can coexist in this market, but will eat into each other’s margins. Processors from second and third tier players like ST, NXP and Renesas will not gain traction. The bottom line for success; more features with less power in the smallest possible die.