2000 and 2001 Should Be Good Years for EDA Companies
EDA and the Semiconductor Cycle
By John Barr,Needham & Co.
Surprisingly, EDA companies have not benefited from the rebound in the semiconductor industry in 1999 and 2000.
John Barr is the engineering software analyst at Needham & Co., the New York investment bank.
These four companies were worth $11.2 billion on March 31, 1999 and just $9.7 billion one year later. How could the stock prices of the customer base as represented by the SOX appreciate 151 percent, while EDA, as the key enabling technology, lost $1.5 billion of market value? People talk about severe price competition, poor management execution, inadequate tools and support, and even blame Wall Street for setting unachievable expectations.
In our view, the industry's poor performance in 1999 was primarily a result of the semiconductor industry cycle and how it relates to design automation purchases. We believe the EDA industry maintained growth in the face of a semiconductor downturn in 1998 through aggressive licensing and that the growth in 1998 led to a flat 1999. In our view, customers that received extended payment terms used budgets from future fiscal years and limited their ability to buy EDA tools in those later years.
We have long thought that EDA purchases were driven by design starts and adoption of new technology. We maintained that EDA purchases were a leading indicator of semiconductor health because tools are used to design new products. Therefore, EDA spending should be out of phase with the semiconductor industry.
We have weighed complex models considering systems and semiconductor R&D spending and excluding the impact of memory suppliers, PCB CAD, PCB CAE and design service revenues. While we believe such a detailed model might be an accurate predictor, we believe the simple relationship between semiconductor revenues and EDA revenues is quite revealing.
We maintain that the health of the semiconductor industry is a strong reflection of the state of the EDA customer base. Additionally, if systems companies are healthy and spending money on semiconductors, they are also spending money on design tools.
While 1999 and 2000 have been rough years for EDA, we believe 2001 should be a very healthy year.
The semiconductor industry's growth over the last five years shows patterns that are helpful in explaining the state of EDA today and why we believe 2001 should be a very good year.
The chart below shows semiconductor industry and EDA revenues from actual results in 1995 through projections in 2001. The current state of the business had its origins in late 1997. The semiconductor industry showed marginal growth after a downturn in 1996. The industry was looking for growth in 1998; instead we had the infamous double-dip down year exacerbated by the Asian crisis. In 1998, the EDA industry posted 19 percent growth in the face of a 9 percent decrease in semiconductor revenues. EDA revenues as a percentage of semiconductor revenues spiked to 2.6 percent, well above previous levels. We believe this 2.6 percent represented an unsustainable level and led to a flat EDA industry in 1999, despite 18 percent semiconductor growth.
We believe the EDA market is still absorbing the licenses sold in 1998 and 1999. Tad LaFountain, Needham & Company's semiconductor analyst, estimates semiconductor growth of 31 percent in 2000, while we estimate just 10 percent EDA growth. LaFountain estimates the semiconductor industry will grow 19 percent in 2001. In 2001, we believe this growth rate will start to show through to EDA and therefore we estimate 15 percent revenue growth.
How Did the EDA Industry Manage to Grow in 1998?
Of the $3.3 billion of EDA revenue in 1998, we estimate $300 million came from the second and third years of licenses offered by Cadence, Mentor and Synopsys. As the companies moved from perpetual licenses to shorter-term licenses, they offered major customers package deals and allowed payments to be made each year. Revenue on the license portion of these deals was recognized at the time the license was signed, yet customers are still paying for these deals two years later. In our view, customers that receive extended payment terms are really using budgets from future fiscal years and decreasing their ability to buy EDA tools in those later years. Had the semiconductor industry grown in 1998, it might have all worked, but the double-dip eliminated customers' ability to continue to increase their EDA expenditures.
In the EDA and Semiconductor Cycle chart above, we have normalized the EDA industry's growth by subtracting the EDA revenue from deals with long-term receivables and adding back payments on long-term receivables. We find that the normalized 1998 EDA revenues as a percentage of semiconductor revenues were still a cyclical high of 2.4 percent. However, we believe the percentage decreased to a healthier 2 percent in 1999. We believe this normalized estimate of EDA as a percentage of semiconductor revenues will decrease to 1.9 percent in 2000 and 2001.
We were particularly pleased to see that Cadence, Mentor and Synopsys all reduced the amount of revenue from term receivables in this year's first quarter. Cadence's revenue from licenses with payment terms extending beyond one year decreased to about $20 million to $25 million from a run rate of about $40 million to $50 million for the last three quarters and about $100 million to $135 million in the three quarters from third quarter 1998 to first quarter 1999. Mentor decreased to roughly no revenue after $30 million in fourth quarter 1999. We also believe Synopsys did not recognize revenue from licenses with more than one year of receivables. As a result, we estimate the EDA companies will do less than $100 million of revenue from extended payment term deals in 2000.
We believe that investors view EDA as a 10 percent growth business in 2000, as shown in the first chart, with troubled companies coming off a bad 1999. Synopsys is currently trading at a P/E ratio of 12 times and Cadence at 18 times estimated 2001 earnings before goodwill.
We don't think investors are giving the companies credit for the underlying growth rate, which we estimate at 23 percent in 2000 (shown in chart above), which is reflective of a healthy business and, we believe, worthy of higher multiples. We believe the higher growth rates and higher P/E multiples will start to show through as investors focus on 2001. Even if the semiconductor industry does not achieve the $233 billion forecast, we believe it should be a much better year for EDA growth and stock prices.
Oh yes, we also believe design starts and technology challenges for 0.18-micron, system-on-a-chip design will make EDA more important than ever in 2000 and 2001. We just think that the semiconductor cycle will allow customers to spend on EDA and design methodology, independent of the technology challenges.
electronicnews.com |
