The road to hell is paved with good intentions. And Cadence is about to find out how close it has gotten to corporate disfunctionality. It all started with what was a good idea at the time. The premise was that the industry was ready for change and that attracting someone from outside of the close knit EDA community would bring a different and thus more profitable approach to the business. Who better than someone who was in a senior executive position at the premier semiconductor company in the world? This is how Michael Fister became Cadence's CEO.
The announcement came just before the 2004 DAC, even if, to be precise, he did not assume the title of President and CEO until August of that year. I still remember the dismissive comments from both EDA executives and analysts about this "outsider" that would soon find out the "EDA reality".
It must be said that during his career at Cadence, Fister did try different things, and succeeded to increase Cadence's profits from about $49 million in 2005 to $296 million in 2007. In the process Cadence surpassed Synopsys in sales to establish itself as the largest EDA company. So, a Wall Street analyst consumed with short term returns had to like Fister.
The fundamental wrong assumption
Every observer of the EDA industry knows that IC design has been the major historical source of revenue and, more importantly, profit. And most observers will agree that the fuel to this engine has been the rapid deployment of new manufacturing technology by semiconductor companies. Gordon Moore prediction that the number of transistors in a given silicon area would double every 18 months, was turned into a law in the late 1960's and then was used to justify the most significant industrial progress of the last 43 years.
As a semiconductor guy who had seen the microprocessor as the most important component in world wide progress, Mr. Fister continued to expand Cadence involvement in this market, not seeing, as he should have, the coming consolidation in the semiconductor design and manufacturing business. By 2007, Synopsys, Mentor, Magma, and Cadence, were competing for the business of about two dozen customers with budgets above $100 million yearly. And things got worse this year, as many companies decided to skip full production at 65 nm and go directly from 90 to 45 nm.
And the future is not any more promising. The number of companies seriously interested in 35 and 22 nm fabrication technologies is getting smaller, not larger. The time between moving from one manufacturing technology to the next is growing, to the point that the prediction that 22 nm will be available for use in 2011 is less certain than it was six months ago.
Thus Cadence is now suffering from not having expanded its markets in areas that have not been traditional EDA markets: like system level design and design automation segments closely related to electronics. But it is worse than that! The most rapidly growing segment of EDA is the use of FPGA in design: and yet Cadence plays a miniscule role in this market segment.
The new behavior
Cadence made a few wrong assumptions on its way to becoming different, the most important being that it believed that behaving differently was the same as being different.
Following the Intel model, Mike Fister decided that Cadence was big enough to create and exist in its own reality, on the assumption that if the largest EDA company ignored traditional EDA events and organizations they would wither away into irrelevance. And in addition, this approach would minimize direct comparisons between Cadence and its competitors. So Cadence reduced its involvement in EDAC, stopped exhibiting at DAC and other industry conferences, narrowly managed its press relations, and built a monument to itself: CDNLive!
The differences between Intel and Cadence are so stark, that Mr. Fister is guilty of not understanding them, internalizing their meaning, and building a strategy founded on useful nuances borrowed from the semiconductor giant. Let's just stick with the obvious: Intel has superior engineering, marketing, and manufacturing capabilities than its one rival AMD. Before I get flamed, I admit that for brief periods of time, AMD has succeeded to equal and even surpass Intel in engineering results, but one has to manufacture and sell its great products for the largest margin possible, and this AMD has not done.
Cadence is not Intel: it has three large capable competitors, who can match and often surpass Cadence technology, its products are very price sensitive, customers are capable of negotiating a better price and the time on hand to wait for the end of the quarter, and does not hold a position of strength in all the phases of IC design, from conceptualization to manufacturing and test.
The final mistake
There are conflicting reports as to whose idea it was to attempt to purchase Mentor: some say it was Fister's idea, some say that Cadence Board of Director instructed him to "get it done". In either case the execution was a sequence of errors that will become a business school case study.
No one seems to dispute the fact that Cadence approached Mentor with the offer sounding like they were doing Mentor the favor of offering such a good deal since obviously Mentor had no better alternatives. It may not have been a "take it or die" offer, but it certainly looked that way. And then there was that contradictory financial situation. The deal, Cadence said, was not dependant on it receiving financing, yet without the financing Cadence could not afford the deal. And finally, all my sources tell me that there was no discussion between Fister and Dr. Rhines as to the fate of Mentor's executive team following the acquisition. Given these premises, the result was predictable. I agree that the unraveling of the financial markets and the softening of the EDA industry did play a role: timing could not have been worse. Cadence financial results were well below expectations, the credit markets dried up, and Cadence saw itself faced with having to increase the offer without having any way to pay for it.
When things go wrong, culprits are needed. And Mike Fister and the team of executives he brought from Intel are the obvious candidates. It is unfortunately not clear that their immediate replacements are any more qualified to lead an EDA company, so the search for a new CEO must be short and furious before the ship drifts so much off course to enter the EDA equivalent of the Bermuda Triangle. And of course, the trickle down phenomenon will take place. There will be one or more re-organizations, there will be one or more layoffs due to the necessary "refocusing" demanded by the new executive team, and there will be further deterioration of financial results for Cadence as its major customers will wait to see what the new Cadence will look like before committing to significant purchases. And do expect very strong attempts by Magma, Mentor, and Synopsys to make significant inroads in the Cadence customers base.
The EDA industry needs fresh ideas and a new approach: it did not and does not need a sartorial makeover.
Michael Fister Resigns -- Interim Office of the Chief Executive Created