Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/are-there-monopolies-or-cartels-in-eda.15687/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Are there monopolies or cartels in EDA ?

tooLongInEDA

Moderator
There's been a fair amount of discussion in other forum articles recently about whether there are de facto monopolies or cartels operating in EDA, specifically the "big 3" (Synopsys, Cadence, Siemens [Mentor for us oldies]).

I've opened a new thread to allow us to keep the discussion in one place.

Full disclosure:

1) I don't believe these claims
2) But I'm biased, having spent around 3 decades in and around digital ASIC/SoC EDA (worked at 2 of the big 3 and a startup) - perhaps I've led a sheltered life !
3) As far as I'm concerned these are tightly defined legal terms - but others have broader definitions of "monopoly" and "cartel" and these may vary globally - so some flexibility may be needed

Opinions welcome.
 
Semiconductor is a field where monopoly is the norm. Just look at general-purpose CPU,GPU,FPGA,lithography machine,cutting edge nodes,memorys etc,all of which are monopolized by less than 3 companies.
 
There's been a fair amount of discussion in other forum articles recently about whether there are de facto monopolies or cartels operating in EDA, specifically the "big 3" (Synopsys, Cadence, Siemens [Mentor for us oldies]).
I've opened a new thread to allow us to keep the discussion in one place.
Full disclosure:
1) I don't believe these claims
2) But I'm biased, having spent around 3 decades in and around digital ASIC/SoC EDA (worked at 2 of the big 3 and a startup) - perhaps I've led a sheltered life !
3) As far as I'm concerned these are tightly defined legal terms - but others have broader definitions of "monopoly" and "cartel" and these may vary globally - so some flexibility may be needed
Opinions welcome.

You left out Ansys so there are now four and with the high rate of acquisitions I do not see that changing anytime soon. I don't think the word cartels is appropriate since there is no collusion that I know of, just a healthy competitive spirit. I am concerned with the lack of EDA start-ups. It will be interesting to see how #59thDAC goes this July. Hopefully EDA will be well represented.

It is also interesting to see the ASIC business expanding. Alphawave buys OpenFive and now SEMIFIVE buys Analog Bits? Interesting times.
 
Semiconductor is a field where monopoly is the norm. Just look at general-purpose CPU,GPU,FPGA,lithography machine,cutting edge nodes,memorys etc,all of which are monopolized by less than 3 companies.
I think the technical term for the market conditions of the Semi field is best characterized by the term "Oligopoly". This is where a market is dominated by a small number of firms and there are significant barriers to entry, a level of interdependence between firms, and a large economic moat (large barriers of entry into the market).
 
I think the technical term for the market conditions of the Semi field is best characterized by the term "Oligopoly". This is where a market is dominated by a small number of firms and there are significant barriers to entry, a level of interdependence between firms, and a large economic moat (large barriers of entry into the market).
Thanks. I personally think that's a fairer description.

I would also agree with Dan - and I think most people here - that we'd like to see more EDA startups. My own (rather limited) experience is that they can innovate a lot faster, but they struggle when scaling up. So absorption by the big 3 (or 4) is not necessarily a bad outcome for the EDA customers. Indeed, I believe some EDA customers actively push for promising startups to be acquired.
 
You left out Ansys so there are now four and with the high rate of acquisitions I do not see that changing anytime soon. I don't think the word cartels is appropriate since there is no collusion that I know of, just a healthy competitive spirit. I am concerned with the lack of EDA start-ups. It will be interesting to see how #59thDAC goes this July. Hopefully EDA will be well represented.

It is also interesting to see the ASIC business expanding. Alphawave buys OpenFive and now SEMIFIVE buys Analog Bits? Interesting times.
Thanks for the correction !

One thing I recall from a couple of visits to DAC in the mid 2000s was both how competitive the industry was, yet at the same time there was a definite collective sense that we were all part of a collective EDA effort. So at a personal level things were always polite and friendly and I've never felt any discomfort in saying "I work in EDA". Which perhaps can't be said for all industries. Of course, EDA tools are far from perfect (the ratio of code base to user base almost guarantees that) and we can always do better.

But both in a startup and large EDA companies (in technical roles), contact with other EDA companies was essentially nil (and necessarily so to avoid things like IP contamination).
 
I've worked for big and small design firms and it's pathetic how the big 3 apply economies of scale to EDA licenses. Even our Cadence AE feels bad about it and tries to cut us deals and free layout trials during a tapeout push. A complete toolset for commodity 180nm costs every bit as much as masks and a minimum wafer run.

The way EDA companies use their command of a closed-source toolchain to control IP markets is also incredibly suspect. I think the industry would benefit tremendously by separating out those conflicts of interests, similar to what the fabless model did to IDM. If these tools resembled more of an open-source ecosystem with publicly scrutinizable code, we could actually standardize some stuff and better enable custom optimizations. Cadence would be more like the Red Hat of EDA, offering "the best" to FANG or whoever, while Arch Linux is still out there for the rest and viable for most applications if you put in the work.

It's also telling when you looks at the slides for Google/Skywater's open-source 130nm PDK, when they get to stuff like synthesis/APR they simply throw up their hands and say "HELP". That should terrify the big 3, because Google is one company with resources and ambition to yank the silver spoons right out of their mouths.
 
I've worked for big and small design firms and it's pathetic how the big 3 apply economies of scale to EDA licenses. Even our Cadence AE feels bad about it and tries to cut us deals and free layout trials during a tapeout push. A complete toolset for commodity 180nm costs every bit as much as masks and a minimum wafer run.

The way EDA companies use their command of a closed-source toolchain to control IP markets is also incredibly suspect. I think the industry would benefit tremendously by separating out those conflicts of interests, similar to what the fabless model did to IDM. If these tools resembled more of an open-source ecosystem with publicly scrutinizable code, we could actually standardize some stuff and better enable custom optimizations. Cadence would be more like the Red Hat of EDA, offering "the best" to FANG or whoever, while Arch Linux is still out there for the rest and viable for most applications if you put in the work.

It's also telling when you looks at the slides for Google/Skywater's open-source 130nm PDK, when they get to stuff like synthesis/APR they simply throw up their hands and say "HELP". That should terrify the big 3, because Google is one company with resources and ambition to yank the silver spoons right out of their mouths.
While I agree that there has been a move away from open interfaces between the tools in EDA toolchains (I'm told this is an intentional decision by Synopsys), there are several other factors.

1) The older "open" interfaces between tools were not always perfect
2) Big customer pressure has often acted to stop this happening
3) If customers want less complex "full flow" solutions from EDA vendors (many do), then this is the price they may have to pay for that

Yes, large customers will get lower unit (seat) pricing - i.e. a volume discount. But is rthat "wrong" ? That's no different from ordering components (though the volume discount may be much higher).

The relationship between a very large customer and the EDA supplier is very different to that for a small customer. The support costs (AEs, etc) for the large customer can be very different than for the small customer. And the large customer is far more likely to be working more directly on new technologies and product versions with the EDA vendor - taking more risks and doing something of great value to the EDA vendor in getting products ready for mass market deployment.

The fact that the large customers get the lower prices presumably tells us these are the most prized accounts for EDA vendors and there's real competition going on to win them.
 
I just read the interesting book by MN senator book
on antitrust: "Antitrust: Taking Monopoloy Power from
the Gilded Age to the Digital Age".

I do not know if the only three EDA companies violates US
competition laws. I think the US courts have sent a very
strong message on copyright laws that makes starting a
new EDA company much easier. The court has ruled
that interfaces are covered by Copyright fair use. I think
if one tries to get EDA funding for an EDA startup, the
answer will be the Synopsys versus Atoptech physical
design copyright (in view of court) case. I think it was
view as Synopsys patenting an interface back then.

The court has ruled that any EDA interface interface
and command set is fair use under US copyright law.
I think there are many interfaces that can be used to
connect to and implement better EDA algorithms so
starting companies should be easy.

Here is a legal article see section B where the US courts
explicitly say the Atoptech case was over turned.

Here is legal article see section b for the Atoptech case
discussion.

Harvard Journal of Law & Technology
Volume 31, Special Issue Spring 2018
THE NEW WAVE : COPYRIGHT AND SOFTWARE INTERFACES
IN THE WAKE OF ORACLE V. GOOGLE

 
It's also telling when you looks at the slides for Google/Skywater's open-source 130nm PDK
(sound of record player screeching to a halt)

There's an open-source PDK?!


well, what do you know! Thanks for the reference....

...oh. Now I feel like I've finally found the briefcase from Pulp Fiction and opened it up and there's a battery-powered Lite Brite inside with a secret message written in Chinese. So there's no magic secret, you actually have to learn to be a chip designer and understand how to use a PDK with EDA tools, to learn much from it.
 
Last edited:
Acoustic models were a constant pain requiring moving
the phone handset to get a better seal. A recent US
Appeals court open interface ruling that is probably
more important to Semiwiki readers is a decision
from a few days ago that ruled scraping data
from Linkedin pages is legal. Microsoft claimed
it was violation of user privacy. Here is a story:

This ruling seems correct.

I find Microsoft/LinkedIn's response ludicruous:

"LinkedIn has no intention of letting the case go. In a statement, LinkedIn spokesperson Greg Snapper said, "We're disappointed in the court's decision. This is a preliminary ruling and the case is far from over." LinkedIn argued, "We will continue to fight to protect our members' ability to control the information they make available on LinkedIn. When your data is taken without permission and used in ways you haven't agreed to, that's not OK. On LinkedIn, our members trust us with their information, which is why we prohibit unauthorized scraping on our platform.""

If I make something public on LinkedIn, I know that I have no control of the data. Same with Facebook.

And if the information published on LinkedIn cannot be legally reused, it probably has no value !

The Microsoft blurb should read "We will continue to fight to protect our [DELETE members'] ability to control the information ...".

Honestly, if these internet companies want to acquire the benefits of being publishers (e.g. copyright), then they must also bear the costs (taking legal responsibility for content). As usual, they are trying to have their cake and eat it.

It's arguable that the original sin here is the the ' "free" to users ' model for these internet services.
 
Back
Top