Cadence Design Systems: Retooling the toolmaker
New CEO shakes things up at Cadence, but faces many challenges in making the No. 2 EDA company No. 1 again.
By Erik Sherman -- Movers & Shakers, 11/1/2004
It's said that No. 2 always tries harder. And that observation might be doubly true when No. 2 used to be No. 1.
 |
| Michael Fister, CEO, Cadence Design Systems |
The decision to appoint someone from outside of the EDA industry as CEO surprised many people, but Cadence is hoping that Fister's managerial experience and combined technical and marketing background will prove useful. At Intel, Fister served as general manager of Intel's enterprise platforms group and was responsible for the design and marketing of several household-name processors.
At Cadence, Fister took control of a company that in 2003 had recorded its second loss—$17.6 million on about $1.12 billion in revenue—in the previous 10 years, according to figures from financial site Hoovers.com. Expenses have been mounting at Cadence, with SG&A (sales, general, and administrative) costs hitting 66.9 percent in 2003, up from 65.1 percent in 2002 and 57.4 percent in 2001.
Looking at the first two quarters of 2004 (third-quarter results were preliminary at the time of writing), SG&A was running at roughly 68.7 percent on $552.80 million in revenue. Even with gross profit margins of 92.7 percent—roughly double that of the average technical or system software vendor—high costs make profits hard to come by.
Cadence's finances must also be understood in context. Annual revenues have been declining since 2001. That might sound worrisome, but to grasp what is happening, it's necessary to discuss the concept of revenue recognition.
According to Cadence's 2003 annual report, an "increasing portion of our product business consists of subscription licenses." Under such arrangements, customers pay for the license over a longer period of time, which spreads the revenue out rather than concentrating it in a single lump payment. One-time payments can artificially pump up the numbers because a customer paying up front will likely not be doing business again for a number of years.
Therefore, a dip in revenue numbers may be a healthy sign; it indicates that the company is avoiding the temptation of pushing through poorly constructed deals in order to pack revenue into a given quarter. The other advantage of subscription-based revenue is that money appears regularly, not just when the company releases a new version of the software. The indirect advantage to customers is that engineers face less pressure to rush a software release out the door prematurely in order to meet revenue goals.
Fister has had to get up to speed fast and move quickly to deal with the roller coaster that is the semiconductor-tools industry. Business seems to have picked up a bit this year, with design starts in the consumer-electronics space up 20 percent. Devices such as large-screen televisions and MP3 players hit a sweet spot for Cadence. "We do a lot of mixed-signal capabilities, and we feel we've had the best technology going there," Fister says. Computer peripheral chips represent another strong area. For example, Intel's new BTX reference design incorporates significant changes in form factor and features.
Boosting revenue is fine, but the expense problem remains, and it won't improve of its own accord unless revenue climbs substantially. This is where Fister's experience at a hardware company could help. Software businesses have a certain habitual response in down times. "You start looking at what these companies do when things get tough, and their first instinct is to lay off people in services," says Jeffrey Tarter, an analyst with SoftLetter who reviewed the Cadence numbers. "They dump the people who are providing the profit margins and they hire more engineering and sales people, and the death spiral begins."
Yet support is a particularly important area for an EDA company to fund. "It's not a question of if you're going to be calling support, it's when," says John Cooley, editor of DeepChip.com, a news and review site for EDA users. "And that's true for all EDA tools. The problem is that the tools are so complicated and involved." According to Cooley, who runs surveys of his audience, Cadence often receives mixed and even poor marks on customer service and support.
Another area where Cadence could experience problems is in how it combines products. "The most important thing is to look at the toolkit and make that as strong as possible," Fister says. "We'll focus on the toolkit and take these disparate pieces of technology and apply them to customer domains."
But the kit-based approach Fister mentions is one of the main gripes EDA customers have, according to Cooley, who refers to it by another name: forced bundling. "The 'We're going to lock you into our stuff and only our stuff' [attitude], it doesn't fly in this industry," he says. "People want to choose individual tools."
The problem for users is that each of the top vendors—Cadence, Synopsys, Mentor Graphics, and Magma—has products for various stages of the development and testing process, but is usually known for being good in only some of them. Engineers want what they consider the best set of design tools, but they have to buy the bundle or kit to obtain the single product. "The real tool flows include products from all sorts of vendors," Cooley says. "I challenge you to find any EDA customer that uses a Cadence-only flow or a Synopsys-only flow or a Magma-only flow."
What that effectively does, Tarter says, is to devalue the applications that people actually want, locking EDA companies into a bundling approach in order to justify the prices that they need to stay in business. And as the bundling continues, the companies then rationalize the situation, claiming that customers are committed to the entire suite, when in reality they might only use one or two applications out of a dozen.
Other potholes await Fister as he settles into his position. Many EDA tool sales are tied to producing chips for newer and smaller geometries, like 90 nanometers. These require new tools, which is good news for the EDA vendors.
But many chip companies have been slow to adopt the latest process technology, partly because of technical difficulties and partly because their particular products might not require the leap.
Fister has built a significant list of goals. For instance, he wants more direct contact with the users of chips—his customers' customers—in order to improve the company's understanding of the market dynamics that ultimately drive the business. In addition, he wants Cadence to continue working on making its products interoperable with products from other companies.
