Here’s something interesting - this is a presentation of Wally Rhines, Mentor at DATE ‘06. If we go over the presentation, we see that one slide has the revenue of the Big 3 as a function of time. The interesting part is that this includes acquisitions . So what we see is that even after acquisitions, there is an almost flat (slightly down) growth. Hmm… why then any acquisitions?
The answer is also provided by Wally Rhines himself. Another slide marks the projected revenue ** if acquisitions were not made** - it is almost 28% lower than what their 1998 revenue would have been. What it means is that acquisitions were opportunity cost. Kill them off after buying them, but dont let them grow unfettered.
How is any of this related to current state of the EDA industry?
The most important line in the whole gobbledegook that I wrote above is - revenue growth is almost flat. Why? Because of the buffet deals; because of the pricing models where the Big Three sell a thousand licenses, each worth peanuts; because a small startup cant even buy a decent server for the price of a few similarly priced licenses (and if you know what a startup is about, you would know that you would be selling atmost a couple of licenses for the first few years.) Now imagine that you were a VC and I as you for your money? Q: “What’s your revenue projections?” A: “About 50,000$ in the first year - about 10 licenses” Q. “You do realize I can make more money selling Google Adwords on a porn site right?” A. “umm… but we have projected revenues of 10 million in 5 years after we have proven our superiority” Q. ”If you have proven your superiority”
Q.”How much money do you need?” A. “Well .. about 10 million dollars” Q. “10 million !!! whatever for? Google was bootstrapped for 100,000 by Bechtolsheim” A. “Well you see there is the frontend which has to support Verilog, Verilog-2001 and VHDL. As well as a few esoteric library, power, timing formats. We need to buy the, since there is no opensource alternative. Again, since the people who know EDA are in the industry and earning a fat pay package, we need to entice them. We need to buy 5 servers with 2gb RAM to run those 10million gate designs”. Q. “Why cant you hire fresh grads - cost you less” A. “They would’nt know jack about EDA” Q. “It was nice meeting you. Dont call me, I’ll call you”
Ladies and Gentlemen, our revenue model is broken. Creative selling of licenses is not going to do jack about it. Companies thrive on undercutting revenue(lets not even go into “we’ll make a product that is 50% better” ). Since the startup costs are high, we need a higher revenue returns.
I do not know what the solution is - is it a Salesforce.com type model, based a few dollars per minute of computing time? Or is it a Microsoft Office kind of a model, in which you need to be a monopoly to do anything about it. Or is the solution, as Dev pointed out - easy availability of basic blocks of opensource code (a la apache, mysql, php, etc.) whcih will drastically drive down development costs (both in terms of manpower and software resources?) I personally think the last one is a good community model - but the question is, who will be the next Linus Torvalds to spearhead, even a Verilog/VHDL/SystemC parser architecture