Intel only needs one copy

Hi Paul, I agree with Mike Demlers point which points out that ?value? should be at the center of EDA licensing . Duolog has an innovative licensing model where we tie our ROI (the value) directly to our licensing in what we call ?complexity-based licensing?. Duolog?s background as a chip-design company has had an impact on how we license our EDA technologies, where, in the past our engineers have been frustrated by having to choose between lower costs and less features(and therefore less ROI) whereas now we deliver our tools fully-loaded with no feature limitations. Instead, we?ve based our licensing model purely on the complexity of the systems that the tool can process. For our register management tool, Bitwise, the main feature is that it captures registers, bitfields and memory maps for IPs/Sub-systems/SoCs and autogenerates specifications, RTL code, verification infrastructures, SW headers etc. We automate ?bitfield collateral? and therefore the value-add depends on the amount of bitfields in the system. If we look at the usage models we can see that IP developers can specify their IPs with <500 bitfields so we can create a license for this type of user at a lower cost than a license for a system with 30,000 bitfields. A solution that automates 30,000 bitflelds will have a bigger ROI than a solution that automates 300 bitfields and therefore the license can be scaled accordingly. To that end an IP team can have access to much cheaper licenses and correspondingly then don?t need to take any high-end resources which can be a common with standard licensing. So it makes more sense that a chip development has access to a range of licenses at varying costs rather than a host of general licenses that have varying feature limitations. To this end, ?value? is at the center of our EDA licensing and coupling licenses to complexity is a way to deliver a shared ROI. Regards - Dave

I have always been curious why small EDA vendors with niche products don''t host the software on their own servers and provide a secure connection and workspace for each customer. It would allow them to monitor and update bugs real time, eliminate tool compatibility issues and revision deployment issues, and charge appropriately for the service. Hardware, runtime and electricity cost would be minimized

EDN - thanks for creating this blog. Intelligent conversation, w/o editorial moderation. Perfect. How long before the EDA marketing types start to dominate this forum?

New business models are never accepted ... until they are accepted. Netflix changed the way people expect to pay for movies. iTunes changed the way we buy music. Similarly, EDA companies have constantly changed their licensing models and customers now expect to buy VPAs with remix rights or eDaCards. These were new and unexpected business models at the time they came out but now we can't even imagine going back to the old models. The trick is to come out with a new business model that best meets the customer's needs for flexibility.

Sorry Paul, but I have to disagree with you on the analogies. I do think it says a lot about EDA (and none of it good) that there is so much continuing debate over what is a viable business model. Other industries mature. Why does EDA continue to struggle with an identity crisis, to the point that we need to search for other "analogous" businesses in a quest to make sense of it all? #1: Selling "enough" software means "enough hours of use". >Time is only one measure of value for a tool, but by no means is it the only one. What about accuracy, capacity, reliability, etc., etc.? We can charge for that, and for most products that I have managed... we have. #2: "Intel only needs one copy?. "They have a hard time building license demand naturally." >In my personal experience, that is not true either. The problem is that vendors have typically saturated major accounts with "all you can eat" deals, which dilutes the value of all the tools to the point that nobody could tell what they were worth individually. If you want to establish the value of EDA, don''t shove all the tools into one big undifferentiated bag of software! VPA''s (volume pricing agreements) not withstanding, Moore''s Law would "naturally" lead to more (and more valuable) tool usage as the size and complexity of chips increased, especially for a company like Intel. When a tool demonstrates real value, there are still many opportunities to grow business. #3: "A nice analogy is the restaurant business". Hmm... I''m not getting that one at all. Once again, this totally misses the point that pricing in other competitive businesses is based on value... as it should be! You said that "rationally a restaurant might charge by the minute", but that is "not what the customers expect". It''s not a matter of what customers expect... it''s a matter of what they value! Is it the ambiance, the creativity of the chef, the location? Or is it all just food? (Dog food, if you recall the infamous Joe Costello incident). If there is a problem in the business model of EDA, it''s that too many vendors have lost all sense of value. Some whine that the value should be higher... just because. What other business works that way??? -Mike

I've been on both sides of this business (I'm a user now) and the issue is a real one - and was 15 years ago. One other way of creating perceived value is to have lots of knobs to twiddle. That makes the user feel like an expert, and have more involvement in the tool. One tool we use now has a button that optimizes the settings - and I wonder how this is going to affect perceived value. I've done floor demos for a tool where you filled out a simple form, pushed a button, and it pumped out some valuable IP - but charging any decent amount of money for that is a hard sell. Maybe the only path to success is having an effective monopoly on one class of tools. I can't see any others.

This is an interesting discussion and is similar to a recent posting at the theAsicGuy.com blog. I always thought it was weird that synthesis tools and simulators were licensed using the per floating concurrent seat model. This gives the vendor no real incentive to improve the performance since they will sell more tools the longer it runs for. I guess benchmarking keeps them honest, but last time I checked most EDA vendors prohibit benchmarking in their license agreements. I think you''ve hit the nail on the head with the "Intel only needs one license." It''s the big companies with the central CAD teams that have most influence over EDA, and it''s in their best interest to keep it per-concurrent seat licensing. The small startups on the other hand are the ones who lose out on this deal. At PDTi SpectaReg.com we offer a per-user subscription year and we try and add as much possible value per user login.

EDA is dead. Everyone is guilty. The semis helped kill it by refusing to pay reasonable prices for the tools. The big guns -- IBM, Motorola/Freescale, TI, etc. -- are the most to blame here. The EDA companies helped kill it by agreeing to these prices to meet near-term sales quotas. After all, a $5 million contract looks *great* on a salesperson''s sheet (nevermind that the list value is somewhere closer to $50 million). Later, when they realized they couldn''t fund further development, the layoffs and closures begin. Then the semis complain that they''re not getting their money''s worth and insist on further price reductions. The vicious cycle continues. This is all very similar to what''s happening in the airline industry. I''m quite glad I got out of this. I do feel sorry for my colleagues who decided to stick with it, though.

I don''t necessarily see it as a sales problem. That''s just an excuse from a salestard who isn''t hitting his numbers. You gotta show value to get my money. That''s what most sales guys will fail to admit... is what is the true value of their technology in terms of the overall piece of the silicon pie. Look at the EDA business. Unfortunately its started up be either serial entrepreneuers (who know nothing about technology) or engineers (who know little about running a business). You need both... you need a good idea and you need a model which makes financial sense. Just because a tool helps me out, doesn''t mean I need to pay you enough to keep you afloat, because I''ve got the other option, which is to do without you or go to a competitor. Your comparison to the restauarant business is great, because they suffer from the same problem. I''ve been to alot of great places which have awesome food... but go out of business. At the same time, there is are web pages to hunt down the McRib. Do I want a nice steak? Hell ya. We all do (sorry Vegans, you know when nobody is looking you want a steak... come one, green beans suck) But McDonalds can open chain after chain selling McRibs, while steak joint after steak joint shuts down. There are some great steak joints which sell great products and have been around for years. Don''t give me a McRib and charge me for a New York strip. That being said, I really wish the EDA industry sold Happy Meals.

Most EDAs provide less than customers expect and then charge annual rental(whatever they call them) fees as well. Offer the customer more than they expect, at a good price and your sales increase. EDA needs to learn that customers expect value for their money; something they do not seem to understand.