Daniel Nenni

With a family of six boys and an incorrigible Dad, eating out was an adventure with the many preferences, complaints, and all out bad behavior. We were banned from the nice places and bored with the fast places, until a Smorgasbored came to town. Swedish for buffet, it was serve yourself, all-you-can-eat, quantity over quality, a perfect place for six growing boys to practice gluttony. The lines were long at first, other restaurants suffered at first, but soon the prices went up and the quality went down so the smorgasbords left town.

EDA started with perpetual software licensing (good for the lifetime of the product) but quickly learned a healthy business cannot survive on 15% annual maintenance fees from existing customers. Existing customers were often neglected in favor of new business, thus the EDA software subscription license (lease) was born. From my memory Avant! gave birth to it, which enabled Avant! to manage near linear growth based on a then loosely defined software subscription GAAP.

In early 2000 the Smorgasbored business model came to EDA, using unfair competitive pricing practices to stifle the innovation of emerging companies. “Why buy a best in class point tool when you can get a mediocre one for free?” From my memory it was Cadence that started it all but others followed quickly, with Magma perfecting it. I remember one of the top semiconductor companies switching flows from Synopsys to Magma. The engineers preferred Synopsys for technical reasons, the CFO chose Magma for a fraction of the cost which prompted my favorite EDA quote “Engineers don’t always get the tools they want!” Synopsys today offers a smorgasbord or “Enterprise” Semiconductor IP business model which will fail for the same reasons that EDA is failing today, innovation stifling pricing!

On the other side of the business model tracks we have gainsharing. Enlightened semiconductor IP companies charge an up front per-project licensing fee with a gainshare royalty stream paid by customers and/or foundries. ASIC Services companies charge a nominal upfront project fee (IP/NRE) and offer a per-chip gainshare price. While EDA revenues have traditionally been flat, they declined an estimated 18% in 2008.  2008 semiconductor IP revenues on the other hand were +8%, and ASIC Services revenues +25%. The math is simple here, EDA is where semiconductors begin, the semiconductor market is worth $250B per year, EDA is $4B per year, where would the semiconductor industry be without EDA again?

Gainsharing can maximize productivity and performance, improve working relationships, simplify work, improve quality of results, align goals and secure long term commitments. It is vital that both companies are married to gainsharing and it needs to be thoroughly researched and documented to ensure successful implementation. The key metrics for EDA ganisharing include: engineering resources (people, software, compute cycles), design time-to-delivery, die size, power consumption, performance, and yield.

EDA gainsharing is not just possible, but probable, being the best way the industry will thrive again. I have seen it done, have done it myself,  believe it. Otherwise the food fights will continue, customers will forever dictate pricing, and EDA will be but a footnote in semiconductor history.