The EDAC Market Statistics Service (MSS) report is out and ,given that it and the Gary Smith report are the only two comprehensive studies of the state of our industry, people will be tempted to draw direct comparisons. This is a temptation that must be avoided, since there is nothing more dangerous than comparing two things belonging to different species.
While the Gary Smith group looks at EDA in general, the MSS contains, with one notable exception discussed later, only information gathered from EDAC members. Although its membership represents a significant portion of EDA companies representing all aspects of our industry, EDAC's membership, especially outside the US, does not include all EDA vendors.
Statistics for the EDA industry are highly biased by the fact that the four largest companies in the industry (Cadence, Magma, Mentor, and Synopsys) represent about 80% of its business, making the behavior of the remaining companies less significant in a statistical study. It is impossible to figure out what the rest of the industry is doing in individual market segments since the figures for the largest four companies are not reported separately. And yet, that is what really interests me.
The revenue from the largest four companies to me represent the present health of the industry, while the revenue of smaller companies and startups represent the potential health of the industry, in other words, they are the leading indicators useful to predict future growth trends. The big four rely mostly on multi-year corporate deals that are less sensitive to market fluctuations. Large customers, just like large vendors, are wary of fluctuations that complicate forecasting and thus make their relationships with financial analysts more complex to manage. After all stock prices and to some extent executives compensations are significantly impacted by how financial analysts see their revenue and profit projections.
On the other hand, in a growing market small companies can be firmer in pricing and find a bigger Total Available Market (TAM), while in periods of slow or even negative growth they must discount their products in order to generate sales.
The good news is that both studies report double-digit growth in 2006 for EDA: Gary Smith reports an 11% growth (although his study is not quite complete yet), while EDAC reports a 15% growth. Significantly the EDAC report relies solely on data reported by its members, while Gary points out that his team also interviews customers who might be less aggressive in generating the numbers.
The SIP controversy
I am going to deal with the Semiconductor Intellectual Property (SIP) sector in the EDASC report first, since this has been a source of controversy in the industry. I believe that whether or not SIP revenue should be counted as EDA revenue would not have become a point of debate is EDAC had only included the revenue reported by its members in the total. Unfortunately, as this market became established and grew, his members generated only a small portion of the sector revenue, not enough, in fact, to establish a clear picture of this market. Therefore, EDAC chose to include in the total the publicly reported revenue from ARM and Rambus, both of which are not EDAC members. This led to accusations that EDAC was trying to inflate the numbers and project a profile of the industry that was healthier than reality. Three observations are in order.
Some other observations
- Instead of following the rule for all the other market segment used to make sure that enough data is used to statistically study a sample, it could have reported SIP numbers with a footnote stating that statistical analysis of this sample was not considered relevant due to the sparse data available.
- It is very easy to extract the "foreign" data from the total because it is reported separately. So although all of the revenue is used to arrive to the annual EDA revenue, one simple subtraction give you the "pure" number that excludes the two outsiders.
- The SIP sector is growing so rapidly that next year EDAC might want to consider dropping the outsiders, since the total revenue reported by its members now amounts to roughly 10% of the total EDA revenue, making the data within the segment statistically relevant.
The MSS contains data showing expected trends, like the increase in revenue in segments like ESL and Analog and Mixed Signal. But it also contains some surprises like a decrease in revenue in the DFT sector. I also found surprising that the revenue from Design Flow and Methodology Services had a negative trend, just when most designers are beginning to use ESL tools. The report also indicates that most verification and other compute intensive tasks are run on Unix platforms and not Windows. Yet Microsoft is unlikely to risk yet another antitrust legal battle for "just" a $2 billion increase in revenue.
And finally compounded revenue growth in the SIP segment for EDAC members since 2003 is more than double than what the two outside companies have achieved in the same period.
All in all the MSS makes for interesting reading and provides a valuable planning tool, even to those companies that are not EDAC members. In trying to understand the present market and especially in extrapolating its behavior into the future, it helps to be able to rely on more than one opinion. Let's hope that our industry will continue to benefit from more than one market statistics report.