The CAD-less Semiconductor Company
There is an excellent post by Paul McLlelan Crushing fixed costs that describes very elegantly semiconductor economics as it relates to operating and capital costs of doing business. He notes that…
“There is a trend that the current downturn is only going to accelerate: to turn fixed costs into variable costs. Often this is what is behind outsourcing of some capability.”
While the main thrust of the discussion was driving the conclusion that semiconductor design outsourcing will accelerate, I would look in a different direction for greater savings - the data centers within semiconductor vendors. This actually ties together a perhaps unintentional thread of Paul’s. He discusses Amazon and Cloud Computing as an example of a trend to increase capital efficiencies in other industries.
On this blog I have argued that these two threads should be combined. Specifically, in a gradual move of “well-matched” parts of the EDA flow to a SaaS (Software-as-a-Service) model, semiconductor vendors are able to take a very large capital expense and turn it into a pay-as-you-use operational expense. This will enable companies to redeploy savings to functions that drive forward their value proposition and differentiation. ie more designers designing better chips faster.
Back in the 90’s there was the start of the move to the “Fabless” Semiconductor company (that was ridiculed in the 80’s and up until Broadcom started showing success). Using this current economic environment as a catalyst, I predict that in the next decade we’ll see a move to the “CAD-less” semiconductor company.
Intel Data Center Utilization and Growth
As an illustration of the imperative, the above chart was presented by Intel at the Intel Developer’s Conference a while back. It has been calculated that the Net Present Value of this data center is somewhere between $200 million and $300 million.
The move to the cloud will not be a step function, as the hurdles are many, but a gradual transition. And just as Broadcom showed the way in the 90’s, the company to best take advantage of the opportunity may not be one of the current leaders in the industry. Someone from the middle of the pack or a cash-strapped startup could use this new business structure as a competitive advantage to take the lead.
For those companies looking to discuss how Xuropa can help facilitate this transition, please contact me.
Connect with me:
Posted under Xuropa, business, industry, marketing
This post was written by James Colgan on March 26, 2009
Tags: CADless Semiconductor, cloud, economics, Fabless Semiconductor, iaas, PaaS, saas
7 Comments so far
-
Jeremy Ralph March 27, 2009 10:00 am
-
James Colgan March 30, 2009 11:42 am
Thanks for the comment Jeremy. It is in the nature of distruptive technologies that there will be some feathers that will be ruffled. This is another reason why this will be a transition rather than a step function. By the way, the need for CAD experts will not go away. Not all of the flow will transition and there may be some “oscillation” as new technologies emerge. Something else to keep in mind as we look further out: generalists could move to CAD infrastructure providers and niche experts will still be highly valuable inside resources with very deep knowledge.
-
Gonzalo Picun March 31, 2009 9:29 am
I think that some time ago I have seen a site supported by Silvaco with web-based pay-per-use CAD-EDA tools
Is there any already established company offering pay-per-use CAD-EDA for the semiconductors design market (simulators like ELDO, layout Tanner & verification tools)?
Thanks.
-
David Neiman April 14, 2009 6:32 am
I would think it would be the TSMC types that would start to offer the CAD flow and the compute resource as a service.
-
Sumontro December 4, 2009 5:20 pm
Outsourcing can only take you so far. At the rate the semiconductors are shrinking, it seems to me that the bigger cost driver is making clean spaces to manufacture the products. I came across an interesting blog (www.horizons-s414.blogspot.com) advocating a unique method of reducing costs for manufacturing. It talks about shifting manufacturing into space to take advantage of the clean conditions. If there is the low cost transportation to support this, I think this is an out of the box alternative that semiconductor companies could look at to both reduce costs and increase profits.
-
Sean Murphy April 13, 2010 8:20 pm
I think the move is well underway.
Olivier Coudert calculates in www.ocoudert.com/blog/2009/12/11/why-service-companies-will-eat-up-eda/ that there is $9B in service revenue associated with the $5B EDA market. As EDA is outsourced so that it becomes a service revenue instead of a license revenue this is the pattern you would expect to see.
-
Gary Dare June 14, 2010 11:47 am
SaaS may help to rescue the value of EDA tools which major suppliers have massively discounted in large corporate packages. Firms may only need 50 licenses in constant use on the cloud rather than 400 in restricted, internal circulation on various subnets. But it is easier to defend the price on each of the 50 licenses, than the 400!
Leave a Comment
Name (required)
Email (required)
Website
Comments
Getting rid of the CAD department is an interesting idea that could really reduce costs. It would need to come from the high-ups though and would ruffle a lot of feather within the incumbent CAD departments.
One thing we observe at PDTi in engaging with chip-dev cos. that may replace their in-house register-map solutions with SpectaReg is that we often get connected with the person who built the in-house solution. For whatever reason management appoints them to investigate the commercial tool and make the decision. For that person, the in-house solution is their baby, it’s what makes them valuable within the company, so they feel threatenend and disrupted by a commercial solution.
To replace the central CAD teams with a multi-tenant CAD solution that serves many different companies, decision makers must think in terms of the greater good of the company, rather than thinking in terms of what makes them more valuable. Hence, the first companies to achieve this will be the ones that have employees that think this way, as James C. Collins describes in his epic book, Good to Great.