categories

HOT TOPICS

Subscribe to our Feed

Future of EDA: Addendum

Posted on Friday, Oct 27th 2006

Recently, Gartner fired their EDA research team, part and parcel, with veteran analysts Gary Smith and Daya Nadamuni. Here’s John Cooley’s report on the departure with comments from many industry stalwarts. This takes out the last remaining independent analyst that covered EDA as an industry.

There is a problem in EDA. Small $4 Billion market that is not growing. No financial energy. No IPOs since 2001, very few acquisitions, alarmingly few financings, closure of EDA-specific venture firms like Telos, overall lack of innovation.

As it is, very few people understand the EDA business, and it is much too complex for Wall Street to fathom. There were a few VCs who knew how to invest in EDA – Mike Schuh at Foundation, Lucio Lanza, Jim Hogan and Bruce Bourbon at Telos, Pierre Lamond at Sequoia, and a few angels, most notably, Prabhu Goel. Others who have invested in EDA, like Theresia Ranzetta of Accel who invested in Jasper, don’t really know much about EDA.

Probably one of the best small companies in EDA today is Atrenta, led by Ajoy Bose, a veteran of the industry. Atrenta recently raised a Series D round at a high valuation, but from obscure VCs, and I don’t think any of them understands EDA either. Says something, when all the top VCs sit out on a late stage deal with serious revenues!

Furthermore, if you talk to people inside the industry, many of them cannot wait to get out and do something else!

So what is going on in the industry?

For solving insanely complex sets of technical challenges, the industry doesn’t offer sufficient financial rewards anymore. There was a time when the big 3 players – Cadence, Synopsys, Mentor – used to engage in active M&A, making it worthwhile for entrepreneurs and VCs to invest. Now, Mike Fister is at the helm of Cadence, and brings with him the NIH philosophy of Intel. Mentor is struggling. Synopsys doesn’t feel the pressure to acquire like before. Thus, the innovation machine has stalled. Add on top the fact that Cadence decided to pull the plug on Telos, their Venture arm, and one of the very few EDA investors is now dead.

If you read John Cooley’s report on Gary Smith, it appears that Cadence and Synopsys got pissed off with Gary, and pulled the plug on him too. Well, anyone who has operated in an oligoply, knows, that the media for the industry is completely dependent on the 2-3 large players who dominate the market. In Mechanical Design, the media sucks up effectively to Autodesk, PTC, Dassault, and as a result, survives. There is very little startup activity, and 95% of the coverage goes to larger players. Gary Smith, however, chose to support those who innovated. Magma, the challenger. Startups. And he pays. It also appears that the financial infrastructure around the Design Automation Conference (DAC) is also precariously waiting to collapse, especially with rumours that Cadence will pull out of the conference.

All this should be alarming for the trillion dollar semiconductor industry. If EDA, as an industry, collapses, or becomes a monopoly / duopoly, the innovation in semiconductors will also stall. I had written a few pieces earlier on the Future of EDA. Here they are again:

Future of EDA |
Mentor Graphics: Target for Silverlake?

And here are two pieces on two growth areas within EDA that are still receiving some funding:
DFM Vision – Clearshape, Blaze, Aprio are working on this area.
RTL Hand-off and Predictive Prototyping – Atrenta could crack this code.

Hacker News
() Comments

Featured Videos

Comments

nice artical. Though not sure how many cared about this analysis. What eda people worry about is infighting between top 3/4 and resulting collpase of innovation and creativity.

nano Sunday, October 29, 2006 at 8:51 AM PT

Why no system-level design approaches? Do you think they are a dead end?

-Sandeep

Sandeep Monday, October 30, 2006 at 6:36 AM PT

Not really. In fact, they have generated a lot of interest lately. However, I believe it is a category that is difficult to productize. You will see, most companies are doing large chunks of service business (CoWare, Forte). They have also raised money, and at this point, I doubt that they will raise much more. But no one shows any interest in acquiring these companies either.

So, my advise, if you’re doing an ESL startup – bootstrap, don’t raise money, do service projects, and develop a cash business.

Sramana Mitra Monday, October 30, 2006 at 10:11 PM PT

Is’nt this what they said when Synopsys, Cadence , etc. were formed as well?

I do think that system level (or some variation , a.k.a algorithmic synthesis) will be the next step forward. Organisations like ST Micro have also committed in a big way for system level design. I think more and more companies are interested in tapping their existing repositories of IP in addition to creating new ones.

That is where system level design will start creeping in.

Sandeep Tuesday, October 31, 2006 at 8:02 AM PT

No. Not at all.

System level design may be of interest to companies like ST, but I don’t believe you can, very effectively, form independent companies on that premise. Doing a small project inside a large company, and building a company are entirely different ball games.

Sramana Mitra Tuesday, October 31, 2006 at 11:22 AM PT

I would say, if PCB design tools are the the 1st generation , schematic entry tools are 2nd generation, IC Design tools are 3rd generation tools which will last for another 4-5 years, the 4th or next generation tools are system level or ESL design tools ..Though it might take considerable amount of time (say another 7-10 yrs) ..it is not really far off from reality when the two curves ( IC Design and ESL Design curves :- X axis would be million of gates, Y axis would be productivity / no of engineers designing per million gates) would meet. If not , atleast they will be close enough…I have been in the field applications for reasonable amt of time and have been talking to some very large design grps at big design companies ..they all are willing to move to ESL Design..as I mentioned earlier until the companies see that the productivity per engineer per million gates is justifiable and profitable, moving to ESL Design will take little more time….speaking of RTL -Predicitive analysis, I have done enough work and as someone already pointed design companies will always want accuracy and they ask for timing, area , power and congestion metrics..Tera tried this, failed…..I did extensive work & analysis on predictive analysis or prototyping solutions
and they dont correlate well with the implementation tools once you move to 65nm and below ( as parasitic delays dominate cell delays) ..and if your predicition doesnt correlate well your mainstram implementation tool ..what good does it bring to the designers…So companies like have Atrenta will have reasonable success till 65nm becomes the main node…Lets see if atrenta can use their predictive analysis and predict if their tools work on 65nm node :)

Kiran Friday, November 3, 2006 at 1:36 AM PT

Integrated electronic is going 3D. The current version, first stage, of it is System In Package (SIP) where mostly existing dice are used to form the required functionality.
The next step and the potentially huge opportunity for the EDA industry is when chips will be built from scratch with 3D in mind, where all the functionality is assumed to be implemented using a SIP where all of its dice are designed from scratch.
Imagine a CPU with 2GB “on chip” memory, tons of analog , super fast digital and multiple sensors; where brand new EDA will optimize it in 3D domain assuming different materials, allowing for what used to be internal busses to travel between dice avoiding congestion, optimizing for power, heat and mechanical stress, placing and routing over multiple dice, functionally and physically analyzing the whole thing etc. etc. Every single function that is now performed in 2D will have to be re-defined and probably re-written.
The EDA guys might still find new innovative ways to kill the value of what they will create but at least this new design style, that is unavoidably coming will give them a fresh opportunity to innovate and grow their industry.

Lavi Monday, November 6, 2006 at 9:55 PM PT

Lavi,
What you mentioned was very good point..but even if EDA tools are 3D capable, market is not ready ..atleast from a business perspective, one EDA company back in 2001/2002 timeframe talked about the distributed computing/or much talked multi-threading in EDA.. for x number of reasons it failed..reason was market was not ready…what you are talking is 2-3 decades from now..It has lot of missing pieces..Analog DA is not mature yet..actually its just getting started…once it is matured enough..both DIC ( Digital IC ) and AIC (Analog IC) has to converge …Well even if all these are possible and done..it will not significantly boost the EDA industry revnue…..The way EDA companies licensing model has to change …it has to change from the traditional single year or perptual or 3 yr time based licenses to more of a revenue sharing model from semiconductor industry it drives..for example per sucessfull chip taped out using their product (atleast initially) to revenue sharing per no of chips sold…. that will give siginicant boost to the EDA industry and will open up new channels and venues..I have already seen and I myself was involved in some deals which are moving into these directions….very few companies recognize the imp of packaging ..

Kiran Monday, November 6, 2006 at 11:29 PM PT

0.000000006 feet under…

What does the firing of Gary Smith  bode for the EDA industry? Not too good it would seem.
I had commented on the EDA industry in my earlier posts. What looked positively heartening to me was that venture funding in the EDA industry looked to be goin…

DAtum Wednesday, November 8, 2006 at 1:29 AM PT

Kiran,

I am not sure that we are decades away from true “from scratch” 3D design, because the consumer demand for “ new supercomputer in your pocket every Christmas” drives companies for integration levels that are very expensive and very risky.
I do agree that the current way the EDA guys conduct their business kill their prospects to break away revenue wise.
I do think that they need to start investing in 3D electronics , also on the system side, in order to enable this design style to happen. It’s a chicken and egg situation here.
The first company that will offer a reasonable flow, will be able to charge significant premium for it, including revenue share. At least until the rest will catch up with them.

On the general model of making money the only way for them is to provide comprehensive solution not at a task/tool base but as a whole product base.
Flows that create a cell phone, cable modem etc. From chip to board to box , the whole thing.
For this EDA companies need to have an open agreed upon data base, to enable designers to mix and match tools easily; EDA need to be able to “talk” to equipment Fab, Test, etc. which will require the players in these fields to agree on a common nomenclatures.
When this type of a flow is there, one can ask for revenue sharing and maybe they will get it.
Given that the focus of most EDA companies and equipment guys is first to kill the competition and secondly to focus on the customer, we can be assured that the convergence mentioned above is not very likely to take place.

Lavi Monday, December 4, 2006 at 9:12 AM PT

[…] a running series Jump to part: Magma?, LBO?, Buyout Rumors PersistI have regularly commented on the EDA industry and discussed specific companies in the past, including Cadence, Mentor, Magma, and Synopsys. Today […]

Cadence: Buyout Rumors Persist - Sramana Mitra on Strategy Wednesday, December 26, 2007 at 4:22 AM PT

“If EDA, as an industry, collapses, or becomes a monopoly / duopoly, the innovation in semiconductors will also stall.”

I like this statement. Yes without foundry/EDA, there won’t be so many exiting ASICs, so many consumer electronics that make our life better.
I feel better as a worker(not sales guy) in this industry: as an industry, we are not vampires, we provide possibilities, we are really helping you.

Ric Thursday, February 21, 2008 at 8:07 PM PT

[…] Target for SilverLake?, Future of EDA, DFM Vision, RTL Hand-off and Predictive Prototyping, Future of EDA: Addendum, I discussed the future of the EDA industry and its structural dysfunctions at […]

Cadence Takes a Page Out of Microsoft’s PlayBook - Sramana Mitra on Strategy Tuesday, June 17, 2008 at 10:50 AM PT
`