Intel Acquires Wind River: Re-aggregation Continues in the Semiconductor Software Stack

• Intel’s acquisition of Wind River Systems sends a couple of different messages. From a semiconductor perspective, it suggests an ongoing trend of re-aggregation in the semiconductor supply chain as a means of more tightly-controlling device architecture for both hardware and software. From an embedded software perspective, the increasing trend towards consolidation that has been ongoing for several years just reached a new plateau. Intel has effectively taken out the largest player in the embedded software market leaving only a few independent companies.
  
  MOTIVATIONS AND TRENDS
  
  Several embedded software companies have been acquired in the past 10 years. The most notable acquisitions are shown in the table below.
  
  Table 1: Acquisitions of Embedded Software Companies
  
  

  Acquiring Company	Target
  Texas Instruments	Spectron Microsystems
  Motorola	Metrowerks
  Harmon	QNX
  Mentor Graphics	ATI
  Radisys	Microware
  Altium	Tasking

  
  Source: Company Press Releases
  
  While the motivations behind each acquisition may have been driven by different strategic or tactical requirements (Motorola was a good example of a tactical play to keep a valued customer, while Harmon’s was more a strategic play for the automotive market), the primary driving factor has been the need to control the device architecture at the hardware and deep OS levels, given the rising importance of software in the stack. Texas Instruments realized this early on and indeed has been very successful in capturing a significant lead in the wireless markets because of this advantage.
  
  Software is now becoming as or more important than the hardware, a trend that Gary Smith EDA has been calling out for several years. The Toshiba Keynote at ASP-DAC this year presented data from METI showing that the embedded software development cost was 60% of the total further supporting this hypothesis. There is no longer any doubt that design going forward is primarily going to be software centric and most semiconductor manufacturers, especially those that have a big stake in the consumer electronics market have recognized and acknowledged the trends. The free lunch that silicon vendors and device manufacturers have enjoyed for so long in terms of being able to squeeze out more efficiencies from the hardware and not really worrying about the software is coming to an end. And with it has come the recognition that going back to the days of turnkey systems may not be such a bad thing after all.
  
  INTEL’S INTERESTS IN THE EMBEDDED MARKET
  
  Intel has been selling embedded processors for many years now in parallel to their bread and butter PC and server businesses. Intel’s footprint in the embedded market has been much smaller in comparison to their PC marketshare. Margins in the PC markets are shrinking and the recent recession has significantly impacted both PC sales and server sales. Intel has tried to expand into the device markets before but has not been successful. However this time, things are a little different. Unlike 10 years ago, there is a new disruptor in the marketplace - multicore architectures. For the past 3-4 years, Intel has been steadily building up a software toolchest for multicore that goes beyond just VTune (a rather long in the tooth optimizing tool.) Intel needs to control the hardware and hardware dependent software layers to produce an optimized, low power architecture to appeal to device manufacturers especially in consumer electronics. You don’t need to look far to see an example of how some companies do this very well. Apple controls every aspect of the device: hardware, software and the user experience.
  
  WHERE DOES WIND RIVER FIT IN?
  
  From an OS perspective, Wind River brings two operating systems to the table. One is VxWorks, a proprietary real time operating system and the second is a linux distribution that they have targeted primarily at telecom and datacom companies. In addition to this, Wind River has built up a sizeable tools and middleware stack over the years, primarily through acquisition.
  
  From a channel perspective, Wind River’s biggest customers are the aerospace/defense and telecom vendors accounting for 50% of the market with the rest divided up among industrial equipment, automotive and consumer electronics. Intel does not have a strong presence in the aerospace/ defense or telecom/datacom markets. Buying Wind River may provide channel access to these markets where Intel can then hope to leverage its hardware products as well. Intel also does not have a large presence in the automotive markets. This has long been an area of strength for Motorola, not withstanding its recent troubles, and again Wind River’s relationships with some of the tier 1 and 2 automotive suppliers may provide a channel for Intel to enter those markets.
  
  Wind River has some presence in the consumer electronics markets, mostly printers and office peripherals though they have had some wins in set top boxes and a few other devices. Their strength in the smartphone and feature phone market is not that clear either. They are part of Google’s Android ecosystem and have a dedicated team of engineers to support the Android platform. They have had some wins in other smartphone markets as well. Intel has not yet signaled an interest in moving into the smartphone market though they are definitely going after the device markets with Atom and Mobilin. And entering the smartphone market (should that be Intel’s intention) is not that easy given that there are already several strong players with well-supported ecosystems. However, the sum of the parts may be greater than the whole and may have motivated Intel to acquire Wind River in preference to other independent software vendors that they could have acquired at a much lower cost but may not have given them the reach they are hoping to get in these channels.
  
  IMPACT ON THE EMBEDDED SOFTWARE AND EDA MARKETS
  
  This leaves the embedded software market in a more fragmented condition than previously. Wind River was the largest player in the market with revenues of approximately $350M. Green Hills Software, Mentor Graphics’ ATI division, LynuxWorks, ExpressLogic, Enea and MontaVista are the other players in the market but combined revenues likely are lower than those of Wind River if you ignore Microsoft’s Embedded division’s revenue. Each company has strengths in different areas as result of historical and market forces that have led to each vendor staking out a strong position in no more than one or two vertical markets.
  
  EDA vendors with the exception of Mentor Graphics have not really embraced the need to create stronger and deeper relationships with embedded software vendors. As the software problem becomes more and more significant in device architecture, they do need a viable strategy to understand how to approach software from an ESL perspective (this includes embedded software) or risk being marginalized.
  
  CONCLUSION
  
  Only time will tell if Intel’s strategy will pay off. Very likely support for other semiconductor manufacturers that Wind River currently supports will dry up. As noted, the embedded software industry is mostly smaller players that may not have the bandwidth to support new reference board implementations. In turn this could trigger either a new round of acquisitions, or semiconductor vendors might decide to invest in more internal developments for the same reasons that drive Apple – control the stack more tightly. EDA vendors need to carefully consider their options with regards to embedded software and ESL. If they don’t have a clear strategy they could find themselves on the fringes as opposed to being a key strategic ally for their semiconductor partners..
  
  Daya Nadamuni