key graf:

Cadence is no stranger to the world of hosted software and indeed could even be called a pioneer of sorts. Back in the days of the nascent world-wide web, EDA companies were testing different ways to approach new business models that would allow them to deploy software as a hosted service. The majority of these attempts however were difficult to implement partly because of bandwidth issues and partly because of security concerns. For those who remember companies like Toolwire and DevelopOnline, these vendors entered the market, believing that they could be successful in EDA design by primarily addressing PCB and FPGA design. Success they felt was sure because of the new approach and cost benefits they could bring in through this new deployment model. The market however was not quite as accepting of the new model and these efforts gradually sank into the background where they lived on mostly as research projects.

A quiet EDA revolution would be in order, i guess. Don’t take on the big guys head-on. Work from the inside (stealth, subterfuge, etc). At this point, you may have realized the broad strokes/motivational advice before is just because I have NO specific idea how it shall be accomplished. If we’re lucky, the current model will self-destruct. But a change is model is a response of a revolution rather than a revolution in itself.

If the big three support all open standards tomorrow and are perfectly interoperable, is the work done? or are we looking for a more democratic EDA industry? to hype it up a tad, Are we looking to save the industry from the big guys?

Aditya

Btw, I added a link to your blog in my “Blogs Of Note” section in my blog (The Tao Of ASICs @ http://asictao.blogspot.com/)

I appreciate you taking the time to post such thorough comments and analysis. Thoughtful give and take from people with perspective from real experience helps to push this concept forward. It won’t be overnight but we’re gonna get there.

Thanks,

harry

At the same time, I do share some of Gary’s views and I think there are several other factors that may make SaaS as exemplified by Salesforce.com not a reasonable scenario for the majority of EDA applications.

Gary brought up several things which I think I can boil down to two things (I discount “it’s been tried before”, because the world has changed):
1. The big guys are not motivated to change their licensing approach
2. SaaS doesn’t work for customized flows

I agree with Gary here on both counts. Let me talk about the first one for a minute:
The big guys revenue and profit is largely dictated by their large customers. They cannot change their pricing/licensing policies too drastically, because their entire corporate culture is geared towards the high-cost-of-sales approach.

A bigger problem however is that the EDA sales cycle is by it’s very nature very long. To understand why, you have to consider the economics. Companies are spending 20+ million on a chip design. A single turn can cost 2-4 million. And the cost of a schedule slip for a high volume product can easily be 10-20 million. The entire EDA budget for a project will likely be a small fraction of the project cost. Also, for highly technical software such as that found in EDA, it simply takes a while to determine how the potential adoption will impact a project. Finally, new tools can only be inserted into a project at specific times (early). The result of all this is that yes, companies would like lower prices on tools, but when push comes to shove, they still want to do a 6-9 month eval to make sure a tool works as advertised, is not going to adversely impact their schedule or quality, and to fit into the next available slot at the beginning of the next project.

What does the EDA company spend during this six months? You’re looking at at least a 1/2 AE and 1/4 sales guy at a minimum. With the sales guy fully loaded at $300K and the AE fully loaded at $150K, you’re looking at 75K just to support hte pre-sales campaign. So they’ve got to sell for $75K just to recoup their cost of sales, let alone R&D and G&A.

So the bottom line is that not an obvious way to address this long sales cycle and the resulting economic impact on pricing. The big customers seem to be unwilling to take alternate approaches. The AE’s and sales guys in the EDA companies are not motivated to change it (it’s their jobs that would suffer). And the EDA management seems to be content with the status quo.

With respect to Gary’s second point (SaaS doesn’t make sense for customized flows), I also think Gary is correct. There are lots of reasons for this. Let me enumerate a couple:
1. Performance. Most serious EDA apps (where the money is) require lots of performance. Think about place&route, layout, functional verification, analog simulation, etc. Local installation, where performance and compute resources are under user control, is a major plus in this area.
2. Interoperability with other tools. When a piece of software is hosted, it becomes much more difficult to interoperate with other tools through scripted flows. Scripting is a fact of life in every design I’ve ever worked on. So again, local installation becomes a major factor.
3. Data security. From my experience, companies are very gun-shy about letting any design data off of their corporate network. For most big customers it is simply out of the question.
4. There’s not a lot of upside for a web-hosted or offsite-hosted EDA. For CRM, there are advantages for web-accessible data so that your sales force has access from anywhere. For HR/payroll/accounting, there are other advantages. But for EDA, I don’t know of any compelling advantages to offsite-hosting, other than potentially reduced hardware costs and administration. But hardware costs are small. I can see administration, but I think this is outweighed by the interoperability issue mentioned above.

So all in all, I think their are a lot of barriers for the big guys to start selling their software via SaaS and making their licensing terms more economical.

All that being said, I do think that some of the points you outlined in your earlier article on the coming EDA revolution are valid. The first is that with the advent of more interactive web sites, high-speed internetworking, and web 2.0 facilities (blogs, social networks, video conferencing, screencasts, etc), and sites such as Xuropa (www.xuropa.com), small EDA companies are in a position to market their software on a large scale much more cheaply than they could in the past. The result will be that small EDA companies will be able to reduce their cost-of-sales and be able to compete with the big players more effectively.

Because of this low cost-of-sales, small EDA companies will be able to afford to utilize SaaS pricing models, such as monthly subscriptions, or metered usage. This could be attractive to some customers. For reasons I outlined above, I don’t think that web-hosted or offsite-hosted distribution is going to work. But for most customers, the biggest benefit of SaaS is affordability. So if you can provide affordable software that provides for at least partially local installation, that can be a reasonable offering.

You have to determine what’s held it back for all of these years. It’s probably more than one barrier, both from a technology as well as economic perspective. Salesforce is more than ten years old so SaaS is an established fact in the enterprise software market. But it hasn’t impacted EDA in any meaningful way…yet.

Disruption starts with unserved non-customers and customers that are unattractive to incumbents.

For startups SaaS allows them much more visibility into what their customers are doing (this was tried with the on premises model in EDA by a company called Aqueduct which was an add on which reported a lot more about design activity).