You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!

  • EDAC Name Changing for ESDA, but what about IP ?

    The EDA Consortium (EDAC) has changed name for Electronic Systems Design Alliance (ESD Alliance). That’s a good reminder that IC are developed (thanks to Design Automation) to be integrated into a System. A wide design ecosystem support system development, including embedded software, design intellectual property (IP), embedded software, advanced packaging (3D, TSV,…) and design service companies.

    Let’s focus on the design IP segment, which is now “officially” the largest category according with the results published by EDAC for Q4 2015 with $702.2 million and +9.2% growth (see the picture below).

    Article: SemiWiki.com Analytics Exposed 2012-edac-results-q4-2015-min-l2l.jpg


    I decide to focus on design IP for several reasons ; at first because we will see that the growing behavior of the segment monitored on a 20 years timeframe exhibit an amazing vitaly, almost imune to economical crisis. The second point justifying to look at the design IP business more carefully is that this segment is underestimated if you only look at EDAC results. Not because EDAC or ESDA doesn’t do a good job, but simply because the IP revenues reported quarterly by EDAC are generated by IP vendors being part of the organization. By definition, an IP vendor who is not an EDAC member will not report IP revenues to EDAC, and there is no way for EDAC to take these revenues into account.

    Article: SemiWiki.com Analytics Exposed 2012-edac-1996-2015-min-l2l.jpg


    Thanks to EDAC, we can monitor the revenues generated in the five categories by the members during the last 20 years, or 80 quarters. The dynamism of the Silicon IP (SIP, in clear blue) appears through the strong growth rate, SIP passing from (almost) zero in 1996 to $700 million per quarter in 2015. CAE, the former largest category, has grown from $300 to $650 million on the same timeframe. Or, if you prefer, it took only 5 years for SIP (2010 to 2015) to make a move which took 20 years to CAE (passing from 300 to 700).

    Let’s zoom into the 2008 to 2011 period, covering the strongest economic crisis since 1929. SIP category has been impacted, strongly declining in 2009, like the other categories. But, it took only 5 quarters for SIP to come back to the pre-crisis level, when it took 10 or 15 quarters for CAE or IC Physical to recover. Why SIP has recovered much faster than tools categories?

    Let’s take an example: a team is developing a SoC in 2009, and R&D cost has to be lowered due to the crisis. The project manager may decide to lower the EDA investment and buy 4 seat licenses instead of 6, for example. But if you need a GPU, or MIPI CSI PHY or PCIe PHY and Controller, taking the decision not to buy these IP and develop it in-house is possible, assuming you have the right design resource, but certainly a lot more risky than buying a Silicon proven IP to a vendor. We can assume that SIP is now a strategic piece of System-on-Chip development and IP outsourcing is clearly growing year after year, with 13% CAGR 2005 to 2015 as extracted from the above picture.

    Article: SemiWiki.com Analytics Exposed 2012-interface-ip-2005-2015.jpg


    I have no problem with the data collected by EDAC as the result is representative of the IP business. When compared with data collected by IPnest for the Interface IP segment (in the $500 million for 2015) we observe the same growth and CAGR, as you can see on the above picture (issued in 2010, but the 2014 forecast fit with actual data with +/-4% error).

    The problem with IP category from EDAC comes from the number of IP vendors reporting their revenue, as only five of them are EDAC members: ARM Ltd., Cadence, Sonics, Synopsys and Mentor Graphics. These five vendors have reported slightly less than $2 billion revenues coming from IP in 2015. It’s easy to guess that the total IP revenues have been higher, but the good question is how much higher.

    I have built a non-exhaustive list of missing IP vendors, indicating their 2015 revenue when available from their annual report:


    • Rambus ($300M)
    • Imagination Technologies ($245M)
    • CEVA ($60M)
    • Faraday ($25M)
    • eMemory Technology
    • Kilopass Technology
    • Sidense
    • Aragio
    • Dolphin Technology
    • Andes Technology
    • Cortus
    • GUC
    • Arteris
    • Silicon Image
    • Arasan Chip Systems
    • PLDA
    • CAST
    • Northwest Logic
    • True Circuits
    • Mixel
    • Analog Bits
    • Silicon Creation
    • Silabtech
    • M31
    • Discretix
    • Sarnoff
    • Chips&Media
    • + several dozens


    We can observe that four of them are reporting IP revenues for a total in excess of $630 million in 2015 ! I have listed another couple of dozens of IP vendors, either privatly owned, either reporting globally their revenue from IP, design services and NRE. If we assume an average revenue in the $10 million range for these companies, that makes $230 million to be added. If we consider that probably 30 to 50 companies are missing in the above list, we can evaluate the revenue generated by the IP business (not reported to EDAC) is most probably in excess of $1 billion… and this makes $3 billion in 2015.

    In conclusion, the SIP category is certainly the most important of the chip design ecosystem, but the figures shared by ESD Alliance are underestimated, the real business figures being in the 40% to 50% higher. What could be done to correct this issue ? Maybe the ESD Alliance could launch a promotion campaign to convince the missing IP vendors to become ESDA members, offering a discounted subscription fee, as most of the missing companies are not as large as the current members… Beeing creative, we certainly could find other options to better know the real size of the Silicon IP market.

    Whatever the selected solution, there is a real need to more accurately know the category which is now the most important in the chip design ecosystem !

    Eric Esteve from IPNEST