What Happened
Yesterday, Synopsys (NASDAQ: SNPS) and ARM (NASDAQ: ARMH) announced a licensing agreement enabling Synopsys to distribute ARM's Fast Models and create models of ARM Cortex Series processors.
VDC’s View
The market for virtual platforms has undergone significant change over the past 18 months, seeing what accounted for the acquisition and/or consolidation of the top 4 vendors (CoWare, Synopsys, VaST, and Virtutech) into what is now two companies (Intel/Wind River and Synopsys).
The acquisition spree early in 2010 came at a time when many companies across the embedded landscape were struggling to shed debt burdens exacerbated by the financial crisis, while others, fueled by proverbial ‘war chests,’ were able to make strategic technological and/or business acquisitions. This dynamic is certainly not a unique characterization of the recent period of financial turmoil, but is nonetheless relevant and a fitting depiction of the events that led to what equated to a massive market consolidation.
The extension of Synopsys’s and ARM’s partnership, while not overly surprising, helps underscore why this change in the competitive landscape depicts much more than just a significant and accelerate market consolidation.
It was not that long ago that ARM poured significant resources (e.g. acquisition of Axys Design Automation in 2004) into its Development Systems (tool) division to build up a suite of ESL solutions that included a commercial virtual platform product, System Generator. Before too long, ARM had sold off ESL assets such SoC Designer and repositioned its System Generator product as Fast Models, a product intended not to drive significant direct revenue but rather one geared at enabling the success of its processor IP customers.
Meanwhile, Synopsys stockpiled a number of the other early leaders in the space (Virtio, VaST, and CoWare) to build out a broad virtual platform solution set to complement its EDA product portfolio while Wind River now uses Virtutech’s Simics product as a means to buttress the value proposition of its traditional software development tool offerings. Imperas, Cadence, and Carbon Design Systems have all subsequently entered the market with products targeting the virtual platform opportunity. The wide range and churn of market participants is perhaps both cause and symptomatic of the diverse technological use cases that range from hardware design to the facilitation of software development.
Clearly, however, no consensus has emerged identifying the most effective pairing – or even complementary engineering discipline (software or hardware) – that can best elevate virtual platforms’ value proposition to the masses.
Over the long term, we anticipate that the comparatively higher volume of software developers (as compared to IC/SoC designers), combined with the growing amount of end device value and project development labor dependent on software, will drive more virtual platform suppliers to recast the articulation of their value proposition to target the larger demographic.
Perhaps the larger question, however, is whether or not a commercial market for the technology will still exist in five years.
Open source/collaborative initiatives such as OVP and the recently announced TLMCentral can help drive the evolution and adoption of the technology, but they can also simultaneously cannibalize – or at lease potentially limit –vendors’ ability to extract premium tool purchases. Although we can point to the Eclipse ecosystem and how software development tool vendors have remained somewhat successful over a long period of time since Eclipse’s inception, we cannot underestimate the impact of the processor/processor IP landscape evolution. The aforementioned shift in ARM’s System Generator/Fast Models strategy as well as Intel’s 2009 acquisition of Wind River are both indicative of the growing pressure on the semiconductor community to enable/speed overall end device functionality and development in order to promote additional current/future socket wins and/or royalties.
One thing is for certain: Across embedded runtime and development solution domains, the growth in software content and end device complexity is driving a reevaluation of incumbent embedded system development solutions. Solution vendors from all domains are now scrambling to monetize the creation and/or enablement of this next generation value-add and, in many cases, are dissolving long-established market boundaries and business models in the process.