April 14, 2006 -- After 23 years working in the EDA industry, it's quite difficult to walk away without imparting some words of insight on an industry that has truly helped make the electronics age ubiquitous and modernized our lives. So, from the sidelines, let me share some of my thoughts on EDA and its future with my EDA colleagues and semiconductor customers.
We've all anecdotally bemoaned the prospect of the EDA industry being in stagnation. Indeed, revenues have been stuck at approximately $4 billion annually for about five years now. And we've also heard about the extenuating circumstances and heard the rationalizations that have helped convince us that EDA isn't doing too badly in the face of this half-decade long semiconductor recession. But Dataquest data shows us that EDA expenditures among semiconductor companies shrunk from 2% to 1.75%, and in the overall scheme of things, the EDA industry has been losing ground.
As a result, Wall Street has lost a lot of its interest in this sector with little growth potential and a stagnant financial performance. The fact that semiconductor buyers, with their growing company revenues, have procured more EDA products at lower cost is a trend that EDA needs to be very concerned about because it's a slippery slope for our industry. It reveals that something is not quite right in the EDA industry and if we do nothing to reverse the trend through some conscientious efforts, it will only get worse.
Looking deeper, we can see several contributing factors to the degradation of EDA's value in the semiconductor customers' minds.
EDA tools lag behind advances in process technology
First, as nanometer processes moved rapidly into mainstream IC production, EDA tools have generally been behind schedule or not mature enough to meet users' needs and expectations (see EETimes 10/24/05 interview). As a result, several resourceful semiconductor companies chartered their own internal CAD departments to develop in-house tools, which did work for them in several cases. Dataquest estimates that such in-house development investment increased to 27%, the highest ever since the inception of the commercial EDA industry in the early 1980s. These resourceful companies claimed to have developed only: 1) those technologies/ capabilities not available from, or insufficiently developed by, EDA vendors, or 2) the interfaces needed to integrate various tools for more efficient internal design flows. However, these companies invariably took off the table the potential of a considerable new revenue source from EDA tools. In addition, these in-house charters enhanced a "we-can-do-EDA tools-ourselves" mentality among key EDA users.
The impact of EDA giants
Second, the increasing influence of the large EDA companies may have stifled EDA innovation and also impacted EDA economics and growth. Large EDA companies often offered packaged deals to major accounts to increase their account penetration. The purchasers of EDA tools always welcomed such deals since they, 1) gained much greater bang for their buck, and 2) achieved more integrated flows with arrays of tools from large EDA companies, even though some tools didn't work as well as others or as well as third-party tools. The average selling price (ASP), however, decreased substantially in all such volume purchases. Once semiconductor customers grew accustomed to these prices, and began to habitually demand such low ASPs in follow-on purchases, total EDA revenues began to suffer. This trend continues today.
Moreover, customers would use these discounted prices to push other EDA vendors, including start-ups, for similar prices. The result? Lower ASPs for start-ups - who need the revenue to fund innovation - would slow these companies and their product development. In parallel, EDA start-ups - where innovation flourishes and unique capabilities really get developed - had to become part of at least one large EDA company's design flow because their products usually had a hard time getting accepted by major semiconductor accounts if not integrated into an established design flow.
The cost of price cutting
Third, as a corollary to the massive power of the large EDA vendors, the price slashing among competing EDA products also caused a general decline in the EDA industry's ASPs. But such product and price competition has injured EDA start-ups severely because they lack staying power and have limited competitive capacity. This is where EDA's developing economic dynamics fundamentally endangers EDA.
A consequence of this EDA business environment is slower growth among start-ups. During last four years, medium-sized EDA companies with revenues between $2 million and $20 million were considerably fewer than those in the last decade. For a long period of time, the number of EDA start-ups stayed relatively constant. But most start-ups had, and continue to have, a difficult time gaining growth momentum for five years in a row.
What happens with start-ups who have to settle for lower ASPs? Ongoing lack of market visibility, limited sales channels, and less-than-full integration into the design flows of larger EDA vendors or major foundries constitute formidable obstacles to start-ups offering stand-alone products. If these start-ups cannot generate sufficient revenue to survive and grow to develop and support new nanometer EDA technologies, the delivery of many innovative nanometer solutions to semiconductor customers will assuredly wither.
Fourth, software piracy - with the sale of unauthorized software through the Internet or other channels at a fraction of its value - is a serious threat to our industry in terms of revenue loss and IP rights violations. The EDAC has been looking into this problem and it might want to find a way to identify the sources of such piracy and press them to stop proliferating EDA software or face legal challenges in their courts. Such an effort needs resources and time to achieve, yet it will be more productive than each EDA company independently scrutinizing and policing the piracy of its products.
The need to work together
The EDA industry can become strong and prosperous once again only if we work together constructively and do the right things as an industry. Admittedly, this will require a sea change in EDA mentality. Since EDA's inception, we have been mired by company-provincial insularity at the cost of action to drive the growth of the EDA market. This must change.
Conducting "status-quo" EDA business will inevitably lead to industry-wide decay or very limited EDA growth at best. We already see the signs of this decay. $4 billion should not be EDA's total revenue ceiling and we should not accept $4 billion as our revenue bottleneck. It will take discipline and sacrifice among EDA companies and strong cooperation from their semiconductor customers and to build a healthy and prosperous EDA future. The key is to uphold and vigorously defend EDA's product value, even under heavy pressure from users, investors and competitors.
Large EDA companies must take the leadership and stick to certain pricing rules agreed upon by EDA Consortium (EDAC) members. It's not about price-fixing or cartel-like decision making. And it's not about regulatory action from the EDAC. It's about unconstrained price cutting which can destroy any industry, including the EDA industry. It's why product-dumping, such as selling memory chips below production cost to gain market share, is illegal.
I've seen examples of sizable industries disappearing rapidly in the third-world countries due mainly to ruthless cut-throat business practices. Examples include Taiwan's jade and marble product industries, which became self-destructive because of over-zealous price cutting more than 10 years ago. Certain industry-wide discount guidelines that all EDA sales persons should abide by might be one way to retain product value and even help shorten the business-closure cycle. One such example comes from the golf equipment industry. I recently discovered that golf club manufacturers support a uniform pricing approach. Buyers can't get more than the standard discount from any golf club outlet in US. A few automobile brands have also started to uphold this limited discount practice.
But it's not just the EDA vendors who have to open their eyes. Semiconductor customers must start realizing that EDA is a key partner in their own success and act positively to foster such a partnership.
We all know that every ecosystem has its own way to live and thrive collectively and energetically. A larger ecosystem (the beneficiary) always learns to protect and assist the smaller sub-ecosystems to helps sustain it. Semiconductor customers have to recognize that, 1) most EDA vendors' tools are vital to their electronic design productivity, and 2) the challenge and high cost of developing next-generation nanometer EDA technologies, which, if compensated with a fair value, will encourage more such development. Their participation in helping their EDA vendors is essential to making these technologies available sooner, continually current to meet the user needs.
I experienced very positive partnerships of this nature between my previous companies and Micron Technology, AMD, STMicroelectronics and a couple of Japanese customers, among others. These EDA customers/partners helped develop full product capabilities with us and integrated them, step-by-step, into their design flows.
So what happens if the EDA industry and its tools continue to be under-valued, and if we allow this creeping decay to continue? In the electronics high tech arena, when EDA loses, the semiconductor industry will suffer from insufficient tools to enable increasingly challenging designs. Support and productivity will decrease as users move farther into the nanometer realm. There's a proverb to describe such a mutual dependence and loss: "Your teeth feel cold when lips are gone."
On the other hand, to win customers' trust, the EDA industry must stay innovative and provide needed nanometer capabilities to users in a timely fashion. Unique capabilities that solve urgent problems for customers are the best way to sustain EDA's value. Users are willing to pay, with little discounting, for real solutions to their difficult problems. Nanometer SOC products have become very large and very complex, with more and more on-chip memory and analog components, all of which are extremely challenging to design, verify, optimize and manufacture. EDA solutions of value include full-chip verification, ESL, DFM, DFY, power reduction, SI, mixed-signal analyses, EMI and several others.
Because of the their focus and diligence, EDA start-ups have a much better shot at developing innovative nanometer products. But they need assistance in bringing their next-generation technologies to market. Other than early funding, they can use the cooperation of potential customers as well as the EDAC or even large EDA companies. They are like new graduates entering a society where they need to be cultivated to fulfill their potential to the society through mentoring or assistance programs. As an example, the EDA Consortium (EDAC)s's Committee for Emergent Companies can do more mentoring than its current program of sponsoring a couple seminars in a year. Start-ups constitute a majority of the EDAC membership. They are worthy of more EDAC's help.
Semiconductor customers, by and large, have taken the initiative at propagating their ecosystem. Customer requirement specifications have been very useful for EDA start-ups to target their technology and product development. The recent ITRS roadmap is an excellent example of the semiconductor industry specifying its needs for the future to its suppliers to guide their suppliers' development. Also, major semiconductor foundries have been actively working with EDA vendors to address DFM and DFY issues. On top of these, it is imperative that the semiconductor industry reassess the the value of EDA and rekindle a partnership spirit with EDA.
On the EDA industry side, large EDA companies need to broaden their support to help start-ups in the design-flow integration of their products so that some next-generation technologies are more readily accepted by semiconductor customers. Constructive -, not competitive - partnerships between large and small EDA companies may bring forth fruitful results. As an example of a win-win EDA partnership, my first company, EPIC, survived in 1988 with an investor and OEM relationship with Valid Logic. At the EPIC IPO, this early investment resulted in more than a 20X financial return to Cadence Design Systems (which had acquired Valid).
Whether the EDA industry will have a bright or a gloomy future depends on how well everyone in the same boat works together as partners and execute with a new spirit of community and a long term vision of mutual reliance. If we all pull together, the total semiconductor-EDA ecosystem grow and prosper. The time is now, but it' both fortuitous and foreboding. The decay has starting to creep in. The question for everyone in the EDA industry is, "How much do we want to cure the problem?"
By Sang Wang, Ph.D.. Sang Wang co-founded EPIC Design Technology and Nassda Corp., and took both companies public. He has a Ph.D. in electrical engineering from Stanford University.