Yet another chip vendor is set to exit the brutal handset market and concentrate its growth plans on the internet of things. Ailing Japanese supplier Renesas is to shut down the LTE modem business it acquired from Nokia in 2010, but has high ambitions for the in-car sector.
The auto space is one of the early stars of the wider internet of things (IoT) space, bringing fast connectivity and even some processing power to every kind of appliance. Established powerhouses like Freescale are increasing their efforts in 3G/4G embedded chipsets; major smartphone chipmakers like Qualcomm and Nvidia are expanding here too; while others are fleeing handsets altogether and pursuing IoT growth.
Renesas is among this group, which also includes Texas Instruments and Fujitsu. The Japanese group is to close the LTE modem business which it built largely around a unit acquired from Nokia in late 2010. Although the division had advanced technologies ‘ in particular, vying with Qualcomm to offer integrated multimode LTE basebands at an early stage ‘ it struggled to make an impact on the market.
Japan-based Renesas paid $200m for Nokia’s modem activities but amid tough market conditions across its businesses, it is pulling in its horns and focusing on its core business. It had previously hoped to sell the modem firm as a going concern but has apparently failed to find a buyer, despite reports of interest from Chinese organizations.
Renesas will continue to supply non-LTE modems and products already in mass production and with customers in place. However, this is assumed to be an interim measure since the unit has made ‘significant losses’ since its acquisition, and its parent was in no financial position to implement and expensive turnaround plan. It secured a $1.8bn bail-out from the Japanese government last year.
This marks the end of an ambitious plan by Renesas to offset stagnation in many of its core businesses by chasing the high growth mobile sector, but given the established might of Qualcomm, Intel/Infineon and others, this was probably an excessively optimistic strategy in a time of global recession and deep problems in the parent firm. It had seen Texas Instruments, Freescale, Analog Devices and ST-Ericsson before it, all pulling out of handset modems, and only ADI managed to find a buyer for the business (which it sold to MediaTek).
Mobile processor makers have been buying up modem specialists in order to chase Qualcomm’s integrated handset system-on-chip platforms, but not big divisions with bloated workforces and complex legacy product lines. Renesas Mobile would have been a far more difficult purchase than, for instance, start-up Icera (which Nvidia snapped up for its 4G modem technology), or Beceem (which went to Broadcom). The notable exception has been Infineon, another stalwart of the auto and embedded spaces, which sold its wireless unit to Intel ‘ probably the last major deal for a unit of this kind, and a chance that STE and Renesas missed.
Various factors have made it tough for second tier mobile players to survive in the smartphone modem or system-on-chip space. One is consolidation, with the power consolidated in the hands of two OEMs ‘ Apple and Samsung ‘ and their suppliers. Both have tight control over their supply chains and leave little oxygen for those which are excluded, including Renesas Mobile and ST-Ericsson (also defunct).
The remaining crumbs can still provide a living for a company with a specialized niche such as 4G-only, and/or very low cost base ‘ Sequans and Altair in the former category, Chinese challengers in the latter ‘ but neither was the case for Renesas Mobile, with its legacy products to support and its 1,500 staff, 1,100 still in high-cost Finland (plus 300 in India and 30 in China. All these are expected to be laid off during a closing-down process that will last for the rest of the year).
Renesas Mobile was becoming increasingly involved with its parent’s in-car activities, which could have seen a common architecture being developed to span mobile and embedded devices. This is an important growth strategy for the entire ARM-based processor community, with many of the approaches developed for larger products now being adapted for the car and then the wider IoT. Even without its modem division, Renesas will continue to develop its R-Car series of automotive SoCs, the latest of which, the R-Car H2, incorporates eight ARM cores. It adopts ARM’s big:little approach to design (combining high performance cores for intensive tasks, with lower power cores for routine activities, lowering the overall power consumption of a system).
Fujitsu is taking the same approach, and like Renesas, sees the in-car market as a rare source of light in times of turmoil (its system-LSI business is being merged with that of Panasonic in its latest bid for profitability). It has produced its first big:little SoC for the auto and industrial markets, combining two high performance Cortex A15 cores and two low power Cortex A7s, along with ARM’s Mali-T624 quad-core graphics processor to offload some tasks from the CPU. The chip is targeted at applications where users control systems via on-screen graphics.
ARM says it now has 17 licensees of the big:little technology, and commercial implementations are now starting to emerge, two years after the concept was launched. The flagship is Samsung, whose Exynos Octa powers some Galaxy S4 models, and after the Renesas and Fujitsu announcements, more should follow soon. Other publicly declared supporters are CSR, Hisilicon, Marvell and MediaTek, though a high profile naysayer is Qualcomm, which claims its own approach to power management in multicore processors is superior to big:little.