The improving global economy, coupled with a major drive for efficiencies and for new markets, is creating a boom in M&A in the semiconductor industry. This is the busiest year for such deals since 2011, and much of the activity is centered on the diverse opportunities in the internet of things (IoT) space.
After a turbulent few years, many chip companies are seeing their results stabilize and their cash reserves mount. Intel and Qualcomm are particularly cash-rich, and under pressure to use some of that money to grow their businesses as they engage in head-to-head struggles across many sectors, not least the nascent IoT. Intel, Qualcomm and Samsung, the top three players, have over $100bn in cash and equivalents between them, over 80% more than in 2010, according to Bloomberg calculations.
Large suppliers also need to make acquisitions in order to drive growth, as the overall semiconductor sector mature ‘ the industry’s revenue has increased by an average of 5% a year since 2010, less than half the rate in the first decade of the century.
Recent deals have included RF Micro Devices’ purchase of TriQuint and Analog Devices’ acquisition of Hittite Microwave, both of which prompted sharp rises in the buyers’ share prices. The main motivations for deals of that nature are to offset slow organic revenue growth by taking on a smaller, but more rapidly expanding firm; or to move into new product areas. As the recession lifts in hi-tech, efficiencies are more rarely the main driver, though they remain important.
Other transactions are driven by a slowing market space and the drive to consolidation. In mobile processors, there has been significant shake-up in the modem and system-on-chip space, and may be more to come. As margins fall and growth rates slow in cellular devices, the chips are increasingly concentrated in the hands of a few giants ‘ Qualcomm at the high end, and some Asian providers such as MediaTek on the commodity side.
Broadcom and Nvidia have both struggled to make their mark on cellular modems and could look to offload the activities they acquired with Renesas Mobile and Icera, respectively ‘ though it is questionable whether they would find a buyer. Other companies which have exited this sector, such as ST-Ericsson and Texas Instruments, have not succeeded.
In low margin, high volume areas – many analog markets, and some technologies for the IoT ‘ there will be a push for just one or two large players to seize the whole space. Texas Instruments is hotly tipped to make further acquisitions to increase its share in analog, the main source of its business now, and to use up excess capacity in some of its plants. Atmel is often mentioned as a potential target in this context. TI is a good example of a company which quit mobile device chips once they became a commodity, and is now relying on one segment it can dominate, and one where it can find significant growth (industrial embedded chips for the vertical IoT markets, where TI is the market leader, but with only about 8% of a fragmented space).
Against this backdrop, analysts are busily playing the game of ‘spot the target’ and Wall Street watchers are pointing at Semtech and M/A-Com, and even the veteran of the whole industry, Fairchild Semiconductor.
This week’s deals in the wireless chip markets were made by the DSP IP leader, CEVA, which has added to its connectivity portfolio by purchasing privately held RivieraWaves; and Atmel itself, in another IoT-driven deal, to buy Newport Media.
RivieraWaves was founded in 2010 and has pulled off the rare feat of building a successful business rapidly around an IP licensing model. Almost its entire staff of 29 people in France’s hi-tech hotbed, Sophia-Antipolis, are R&D engineers, and it has signed more than 25 customers for its WiFi and Bluetooth connectivity IP, including some substantial names like Renesas, Dialog and Celeno, and others which remain secret.
No wonder, then, that the start-up appealed to CEVA, the ‘ARM of digital signal processors’, which has been expanding its own IP licensing model into other technology areas, and was pipped at the post by Imagination Technologies in 2012, when the two licensors battled to acquire MIPS for its ARM-rivalling processor platform.
Licensing is a good business at the moment. Many chip vendors, especially in emerging economies, look to license rather than design most elements of their platforms, to reduce their R&D costs and tap into broader ecosystems. And on the other hand, some device designers, and even some cloud and network equipment makers, are turning back from merchant chips to those designed inhouse (as seen at Apple and in some ZTE base stations, for instance), but similarly, are not looking to design every building block from scratch.
Despite RivieraWaves’ small size, it will boost CEVA’s licensing revenues ‘ recently about $23m a year – by 25%, CEO Gideon Wertheizer said on an analyst call to discuss the deal. That makes the purchase price of $12m upfront, and another $7m contigent on performance, seem very reasonable, especially as CEVA’s larger size should enable it to scale up its new business quickly, and it has locked key engineers in for two years.
The two main segments which CEVA will target with its new IP will be existing customers who build small cells and other infrastructure based on its DSP platform; and companies which require hardwired connectivity for SoCs aimed at mobile and wearable devices. In the former category, Wertheizer told EETimes: ‘We see among our customers an almost immediate need to run software-based WiFi algorithms, in addition to LTE processing.’
In the second category, the main growth opportunity will be in connectivity for IoT devices ‘ consumer at first, and then industrial. Here, Imagination rears its head again, and is seen as the main rival, having unveiled a configurable all-in-one IP solution late last year, integrating WiFi, Bluetooth and FM, and named Ensigma.
Prior to its new purchase, CEVA was working with Antcor of Greece on WiFi and developing Bluetooth inhouse ‘ its TeakLite-4 DSP v2 architecture, unveiled earlier this year, supports Bluetooth dual-mode (classic and Low Energy) connectivity. The two Bluetooth activities will merge over time, though CEVA is already working on future Bluetooth iterations, as well as on new WiFi extensions such as 802.11ah for long range, sub-1GHz links, and 802.11p for vehicle-to-vehicle communication.
Meanwhile, hard on the heels of RFMD’s and ADI’s purchases in the RF chip space, another player in that segment, Atmel, is to pay up to $170m for Newport Media ($140m on closing and $30m subject to meeting revenue targets over the next two years). Newport made $43m in 2013.
Newport is well known for its mobile TV activities but the main attractions for its new parent are its low power Bluetooth and WiFi offerings, which the larger firm says will “enable Atmel to offer the industry’s most complete wireless portfolio of smart connected devices for the internet of things’.
This indicates where the next wave of M&A cash will be spent in the wireless semiconductor space. Whether they are adopting a licensing, fabless or manufacturing model, every player needs to position itself for the IoT, and in a fragmented and ill-defined market, that means either having the scale to cover a very wide range of specialized chips for specific applications and sectors; and/or controlling an integrated set of core horizontal enablers, among which the Bluetooth LE/WiFi combination must be the most important.