Over the past week we heard good news from Xilinx and Altera, both raising their revenue targets for Q4CY09 (Q3FY10 and Q4FY09 respectively). Both of the FPGA giants are doing fine, and are poised to grow twice as fast as the semiconductor industry. The semiconductors companies are doing well too, with TI upping its Q4CY09 guidance, National leading the forecast in industrial demand, UMC and TSMC reporting a year-to-year sales increase of 52% in November, and the overall chip sales growing 14% year-to-year in October.

It is good to learn that the customers of the EDA industry are doing better –if they do badly, EDA will do too. But will that seemingly economic improvement of the semiconductor industry translate into better days for EDA? Nothing is less certain. The recent quarterly reports of Synopsys, Mentor, Cadence, and Magma, although slightly above guidance, show a bleak outlook. Most of the book-to-bill ratios decreased, and they all carefully announcing a lean year ahead.

I recently ran into some acquaintance working for a leading semiconductor company (in the top 15), who told me that they are reaching out to services companies to get more values out of them. The numbers speak for themselves: they will put $12 millions down for an evaluation project that will encompass the full backend part of the design cycle –about 8 months project. That is only for an evaluation! When was the last time any EDA company was given that amount of cash for a real-life trial?

More numbers? Let us only look at the VLSI service companies in India, i.e., in no specific order: HCL Technologies, KPIT Cummins Infosystems Ltd, MindTree Ltd, Sasken Communication Technologies, Tata Consultancy Services, Wipro Technologies. According to the India Semiconductor Association, VLSI design service revenues in India could hit $1.13 billion in 2009, while hardware and board design could reach $560 million and embedded design and services about $7.29 billion. Yes, that’s nearly $9 billion overall, nearly twice the EDA market, and China is not even in the picture yet. Despite the dramatic downturn in 2009, some of these services companies did quite well, and most expect an uptick with a recovery in the semi industry next year.

The truth is that EDA companies have been providing software solutions that are more and more seen as commodities. The license renewal rate is dropping and its volume is decreasing.  In a flat, if not slowly shrinking market, the EDA firms have to eat their competitors’ share if they want to grow or just survive. They drop their prices and fork free AE support to sweet the deal for the customer. The vast majority of the designs can be done with last year’s generation suite, thus there is no urgency to buy new design tools. Then semiconductor companies might indeed be better off with a dedicated service company, which provides hands-on design expertise, and will be judged on results, i.e., the final tapeout. This is a win-win situation: the customer can fully rely on the service company, and since this business model commands a much higher fee than for a software license, the service company can expand and further invest to be an intimate part of their customers’ flows.

EDA has better look around and see what is happening. Semiconductors companies will more and more rely on service companies, tailored to their needs. Chip design and verification looks more and more like an IP assembly that requires an expertise that EDA tools do no longer deliver. The value-added is in that expertise, not in the tools that are becoming more and more push-process.

It is true that VLSI service companies buy tools to EDA companies, but the service companies factorize the license usage between several customers, which means that overall, less licenses are needed. Today, TCS can easily rent any EDA tool from the big 3 by the week or by the month. Imagine tomorrow Synopsys, Cadence, and Mentor dealing only with the top 6 hardware design service companies, themselves servicing the top 20-30 in the semi industry: EDA will loose a lot of leverage in the sale negotiation process. If Magma ends up as a cheap provider of IC implementation solutions, all developed in India, it will lower the bar even more.

EDA has to evolve quickly if it does not want to be sidelined as just an enabler. The EDA industry must be part of the design expertise, and work closely with its customers, even if it means its solution is no longer generic. And yes, as I said in the past, the value-added is in the system-level software, and this is where resides the growth of hardware designs. So the EDA industry must go into chip software design and verification if it wants to be relevant in five years from now.

Tags: , , , ,

24 Comments on Why service companies will eat up EDA

  1. sol gradman says:

    Hi Olivier,

    Nice article! I enjoied reading it.

    Sol

  2. Hi Sol!

    Hope you’re doing fine in Israel. Are you seeing the same trend I am seeing, i.e., more and more service companies (whether for software or hardware design expertise) are taking over the customers’ flows?

  3. Jason Landry says:

    Actually, I’m not seeing such thing except when I read blogs. I wonder whether it’s just a blogging trend since there are a couple of EDA bloggers who are just blindly writing things not relevant to real life situations.

  4. […] This post was Twitted by 0x9900 […]

  5. Without violating confidences, what are some examples of EDA service companies that are thriving or that you believe will benefit from this emerging shift in semiconductor preference for service over software? I am not trying to challenge the accuracy of your statement, I am trying understand more specifically which companies you are talking about. In particular is it:
    1. Service firms using commercial EDA tools to deliver a result (whether it’s a component of a design, an analysis for a design or design block, or a transformation of one aspect of a design into another), which means that someone is still buying tools and adding a service margin on top.
    2. Service firms using internally developed tools, but selling services instead of tools.
    3. services firms pirating tools and selling services based on operating them.
    4. other: one or more that don’t fit into the first three categories.

  6. I am seeing two leading semiconductor companies in Europe increasingly relying on 3rd parties for some part of the flow. They use the services of an external company (not an EDA vendor) that provides expertise (not a tool) to implement a specific process. Imagine the top semi doing their chip implementation with 3rd parties. Then the services companies would become EDA’s primary customers. As said above, instead of having 20 main customers, EDA would have 5-6 large services companies as lead customers. EDA would be squeezed even more.

  7. Sean, thank you for your incisive questions. It motivated a clear answer from me, which eventually I moved to the post itself –the part where I discuss the top 6 Indian VLSI service companies. You made this post better for everybody to read.

    There are a few small companies in the #2 space (i.e., developing some EDA tools, but selling the service). A few companies in timing and physical characterization, some into had-crafted low power architecture. It is still small, but if they deliver value that meets the customers’ needs, there is no reason why they will not grow.

  8. Hi Olivier: I would agree with a trend towards more flow automation and a higher level of abstraction with designers focusing more on the high leverage aspects of the design: architecture, software, verification and validation. As an example a very large chip designer/EDA customer feels that the RTL2GDSII process is a solved problem, and thus it supports a direction of flow automation. Yet this trend can take several directions.

    One alternative path is for EDA vendors to provide services and design realization on their own. They own the tools, have the best insight on how to generate results and it would bring them to a more intimate role with their customers. They will need to work on cost efficiency to compete with design houses, yet they have an unfair advantage of infinite tool access and ability to work closely internally with the EDA product teams so they can optimize the flow/tool integration and also make any tool releases transparent to the end designer.

    Yet another alternatives is for foundries to absorb this function and just increase the abstraction level for hand-off. Foundries offer RTL hand-off services at a premium, and this premium is likely highly flexible for star high volume accounts. The FPGA vendor’s do their own tool development and give away the software. A good example of this is the Apple/Samsung relationship in which Samsung provides a lot of the upfront NRE of the back-end for free, yet shares great success on iphone semi sales.

    In order to control the flow service market any player would need to have a unique differentiated value in technology or a leveraged business position, or both. EDA’s have the tools expertise, foundries have the customer relationship and foundry data and control. It’s not clear what advantage if any service companies provide, other than low cost talent in remote regions.

    So in summary the trend towards implementation and DFM outsourcing is an accurate prediction, yet I see the result will be more flow automation and business ultimately shared between semi foundries and EDA.

  9. Oliver thanks for the update, it looks like the EDA Industry’s economic center of gravity has shifted to India (about 20 years faster than Peggy Aycinena predicted in 2006 in http://www10.edacafe.com/nbc/a.....eid=321147

    Here were her concluding paragraphs

    Here’s how the common wisdom goes. Cal is about real people. People who don’t come from money. People who are honest, urban, authentic, rebels, thinkers, innovators, agitators, and fun. Stanford is about the Bubble People. People who come from money, live in a Bubble, and go to school on The Farm. Stanford people are polite, well-dressed, pleasant, tidy and smart. Sometimes they can play a good game of football, but more likely they’re busy winning Nobel Prizes. Cal people wear jeans. Stanford people wear slacks. Cal people wear sweatshirts or t-shirts. Stanford people wear cardigan sweaters. Cal people drive clunkers, if they own a car at all. Stanford people drive BMWs. I think you get the idea.

    But you know what the other idea is that you should also get? Cal and Stanford people, at least the ones who were swapping barbs over dinner tonight at the Marriott Hotel, are both equally capable of losing sight of the fact that not everybody in EDA gets it. Not everybody in EDA thinks that the center of gravity of the world is somewhere on the straight line that connects Hoover Tower to the Campanile. Not everybody in EDA thinks that Northern California, or EDAC in particular, is actually the center of the world.

    And I’m guessing the majority of those people – the ones who don’t get it – are working away somewhere on the other side of the International Date Line. They’re not thinking about the frat boys – be they from Cal or Stanford – at the back of the room, or the presentations and power brokers – be they from Cal or Stanford – at the front of the room. In fact the people on the other side of the International Date Line, probably aren’t thinking about the Kaufman Award dinner at all.

    Because, in the opinion of this bemused school marm-turned-industry freeloader, those people are the Real People. For them, the differences between Cal and Stanford are totally negligible – smaller even than 45 nanometers – when compared to the differences between the entire well-heeled, well-fed, highly motivated, incredibly creative, innovate, and self-obsessed Bay Area population of techno-successes, and those folks in other geographies who are working hard to wrest control of the technical momentum in this century.

    Whatever was going on tonight at the Marriott Hotel (and I can promise there was lots!), whatever politics caused the Head Honcho of the Numero Uno company in EDA to be sitting at the back, and caused the other two CEOs to be sitting at the front, whatever caused the next Executive Director of EDAC not to be named tonight, whatever caused the CEO of the Numero Cuatro company in EDA not to be sitting in the room at all – all of those politics, and more, pale in comparison to the real sea change that’s underway in the industry and the world.

    The center of gravity for the entire universe of high-tech is moving. Profoundly, steadily, and undeniably. Ten years from now, twenty years from now, 30 years from now — when all of the very well dressed (even those from Cal), very well fed, very well paid people who were in the Marriott tonight have either retired or gone to their great reward – this whole event will be happening somewhere else. Where that will be, is anybody’s guess.

    It won’t be in Northern California. It may not even be in North America. But it will happen. Because people who are real, authentic, hard working, and don’t come from money today are working hard to see that their children are well fed and well paid tomorrow. Those people are going to beat tonight’s crowd at their own game. Creating CAD tools that serve their local markets, and then serve global markets. It’s inevitable. And they’ll give awards to those who are most successful at contributing to the game.

  10. Sean, I unfortunately believe that you are right, and I don’t see much effort by the EDA industry to reverse the trend.

    The fact that high-tech industries establish themselves in India or China is not new, and I personally cannot see anything wrong with it as long as you agree on the rules of free-market, competition, and transparency. There are plenty of high-tech companies with huge facilities in India and China (Google, Microsoft, Yahoo, SAP, Cisco, etc) that are firmly anchored in the US as far as strategy and innovation is concerned. It looks like EDA gave up and its promises some time ago, for instance when Synopsys and Cadence ramped up services and consulting business units that eventually staled or regressed. The big 3 have no more momentum left. The VLSI service units in India (and much smaller ones in the US and Europe) are gaining into the customer flows. If a Tata consulting can buy a few ICC licenses to cover 20% of the next ASIC designs, the biggest looser will be Synopsys (take your tool and substitute your company’s name). That is exactly the scenario that is looming unless EDA companies can have a consistent strategy to address customers’ needs, and work with them as partners, not as enablers. Else it will not be long before the bulk of the design and implementation can be fully covered by services, making an independent EDA industry a thing of the past.

  11. Nick says:

    Hi Olivier,
    I agree with stephen’s comments.

    The value add of most of these service companies is not great and using a lot of jargon on their websites doesnt definitely differentiate them from one another!

    Most semi companies outsource those projects which they feel are not their bread and butter product lines (for example: TI wouldnt outsource OMAP to OSI) and neither would samsung outsource it’s APPLE applications processor.

    Most service companies really are a “me too”.
    They get business which can be done by anybody else (product lines which need to be maintained, products which are not differentiating enough for the market!) albeit at a lower cost.

    If we look at another scenario, a startup company might have difficulty building a physical design team for their first tapeout in 45nm technology. It might outsource the project first time around to get a foot hold in the market for its chips.

    But as the company faces competetion going forward the company will gradually build its own layout team so that their silicon is a true differentiator not just in terms of functionality, but also in timing/power and area.

    EDA I guess should not be threatened by this phenomena. They should be willing to provide services to a company if needed and factor it in as cost for services and not support!

    EDA has the upper hand in understanding the technology and the nuts and bolts of a CAD system and they are the ones who can truly differentiate. I am sure the semiconductor companies understand this!.

    -Nick

  12. Hi Stephen,

    I like the analogy with Apple/Samsung. But I would not underestimate the value that service companies provide. It is clear that some of these service companies benefit from, as you said, low cost talent in remote regions. But that talent is obviously something that the customers look for, and the fact that the service companies can put revenue numbers a few times larger that the whole EDA industry shows that this value is required for a lot of designs. Now we can argue why semiconductor companies are willing to pay for these services. Is it because they cannot find that expertise within their ranks or their country? Or is it because EDA tools are lacking automation? Likely some mix of both.

    We both agree that the IC implementation and DFM parts will become a commodity that is going be outsourced to very cost-efficient service companies in India (and possibly China in the future). I am not sold yet on more flow automation, which will result in more business shared between semis and the EDA industry. My concern is that if the service companies walk up the food chain (and why wouldn’t they?), they will cover the complete system-to-GDSII flow, and therefore become the primary customers of EDA for its full tool suite. In that context, and don’t know how EDA will be able to increase its business if the service companies can optimize the license usage among their semi customers.

  13. Hi Nick,

    I am not as optimistic as you here. First of all, the added value of a ASIC startup is more in the system than the actual silicon. I don’t think a newcomer that wants to design the next great consumer electronics good will go through the burden of building up its own physical design team. It is more cost-efficient and less riskier to work with IPs and service companies to support the IC design and IC implementation flow.

    Second, if a company wants to differentiate at the silicon level, then I have to say that getting an edge on power is not something EDA tools do very well. Power optimization is still a very manual process, which requires quite a lot of expertise. This is definitely an aspect of the QoR where a service company can provide a value that EDA tools cannot (yet).

    EDA does have the upper hand in understanding the technology and the nuts and bolts of a CAD system, but that does not mean that this is a healthy business. Growth in EDA has been pretty much flat for the last few years. Analysts predict that EDA will follow their customer’s growth in the next 5 years, but as the number of ASIC design starts goes down, this means you have to convince the customers to pay more for tools that still have to meet their promises in terms of scalability and performances –which, for the most part, they did not entirely fulfilled.

    I don’t see EDA having the upper hand when it comes to working out a business partnership with the semiconductor industry. At best, EDA growth will be aligned to the semi’s, while EDA is already hurting. At worst, EDA will provide the software tools that will be the utilities of a much larger service industry. If EDA can establish a real partnership with its customers, and if the customers truly appreciate the value of a vigorous EDA industry, then we might see innovative business models that would benefit both side. Until then, I don’t see how EDA can improve its business foothold.

  14. Nick says:

    Hi Olivier,

    Looking at the list of VLSI services companies above and their revenues that you mentioned.. a major contributor to their revenue is embedded design and services space. None of the major EDA players seem to have a big presence here?

    Where they offer true differentiation is in the silicon design phase. (Cadence, synopsys, Mentor and Magma).

    Most semiconductor companies outsource their WinCE support/Linux/Android work on their chips to services companies. If you think about it this is not an impossible task but semiconductor companies are averse to doing it themselves as they can get it for cheap some where else rather than doing it in house. They offer real differentiation at the silicon level and thats where their core competence has been so far!.

    This is just like application software outsourcing and it has been happening for 20 years now. This is not software that can truely differentiate itself like the EDA software.

    Do you think EDA can add value by getting into embedded space? I think they will start developing all kinds of software like by doing this to add to their revenues and thats where they will lose their identity.

    -Nick

  15. Kanai says:

    Thanks Olivier for opening up this highly relevant topic. I agree with you that its time for EDA to adapt with the market dynamics or to acceptt the fate …

  16. Hi Nick,

    Actually, I see these VLSI service companies grabbing a lot of business on the implementation phase of the IC design (i.e., from gate to GDSII). Basically, the place-and-route space. This is becoming more and more of a commodity. And a few of these companies are trying to go to the next obvious step, which is the RTL-to-GDSII flow, some with with notable success.

    I’m not sure about how specific embedded designs are for service companies. Usually it requires more IP and software design and testing/verification, and I do not have any information about whether they can provide any significant added value there.

    I think EDA should focus more on high (or soon-high-to-be) added-value tools: low power design & optimization, test + simulation + verification, and system-level design (IP+software). These are difficult tasks to automate or/and to perform quickly, and EDA could reduce TAT drastically here. These will command a better business than P&R, which is becoming trivialized for most of the ASIC designs (either the QoR can be met automatically, or lots of know-how is needed to meet the targets, therefore out of EDA’s scope).

  17. […] Why service companies will eat up EDA […]

  18. Roberto says:

    I do not think Indian consulting companies earn that much on hardware services. The most important thing for the chip vendors is IP. When they give a testcase to their vendor for debugging they have a huge nondisclosure agreement. It is highly unlikely that they will trust services companies for most of their IP. Your premise is based on the fact that these services companies will be able to optimize their products across different vendors. Again all these big semis have plenty of design groups and they try to optimize the license usage in-house and so license usage is unlikely to be drastically reduced. All EDA vendors have a big presence in India (they have been there for ages..early to mid 90s). They have consultants there who can work at the same salary as say the other consulting companies. Someone like Synopsys will have unlimited licenses in-house and they do not need to pay anyone for any tools. So any price these competitors offer, the likes of Synopsys, Cadence and Mentor can always beat it. If the flow is to be outsources, the EDA vendors with their huge platforms are going to benefit immensely. There is no way other indian consulting firms can beat them. Those firms have a huge advantage in software services because of cost of labor. The cost of tools is not high for software services. For hardware services, the cost of tools are huge and they would have to buy tools from their competitors for consulting services. So, as Steve mentioned above, the EDA vendors and foundries are more likely to offer these services.

  19. Kanai says:

    Hi Roberto,
    Thanks a lot for sharing your thought with us. I find it difficult to buy your explanation when I see it from a different perspective. See below my comments:

    – The focus of the EDA companies should be to grow in the domain of EDA. If they starts getting into services, they may loose their focus- it may not be good for their future and hence for the health of the semiconductor industry. We need to understand why the design houses do not develop EDA tool by themselves- one of the reasons is that it is not their primary focus, they do not want to become specilazed in EDA, also ROI etc … for the same reason an EDA product company should not get into design/Flow services- if at all they get into it then they should try to develop/sell the design for themselves …

    – It is hard to believe that one EDA company will have all the best quality solutions/tools inhouse. So, it may not make sense to take the design-steps through the EDA company’s inhouse tool(s) which is/are not the best in the industry.

    – The established leaders in the domain of design/flow services is expected to have access to all the best quality EDA tools from different vendors and their experts. This way it will be possible to take a design through the best in class EDA tools in the industry.

    – Regarding your comment on cost of license for the third party service companies vs. free inhouse EDA companies, this is not quite true. This was well explained by Olivier in his first post …

    Keeping the above in mind, I believe that we should go for outsourcing to third party( non EDA-product ) companies to get the best quality design without compromising the cost and also for the betterment of the semiconductor industry.

    Thanks,
    –Kanai

  20. Roberto says:

    Hi Kanai,

    I am not convinced by your arguments. Firstly Oliver says that most of the designs can be done by last year’s tools and hence it is not necessary for any EDA vendor to have all the best in class tools. Some best, some second best and he is there. With the advent of “platforms” more and more companies actually use one company’s tools for the whole P&R. Even if they do not, in P&R for non-bleeding edge designs most companies will do a very decent job. Secondly, the likes of Synopsys and Cadence do have design services and they do not go aggressively because they do not want to be competing with their customers. however if there is an ecosystem where most designs are getting outsourced, they will go after it. Your assertion that EDA companies should sell desing for themselves is already partially true (in terms of IP). Synopsys is the second largest IP vendor after ARM. Thirdly, if cloud computing/SaaS takes over this is likely to be truer.

    Again you talk about established leaders in design/flows have access to best tools and experts, I would like to know what you mean by established leaders. If you say Samsung/TSMC, I agree. If you say TCS/Wipro, I do not.

    In short, you proposition is that no company has all best in class tool. And I am saying that for the type of designs getting outsourced you do not need best in class tool at each step. The likes of Synopsys, Magma and Cadence have good platforms to tape a chip out on a single platform.

  21. Gary Dare says:

    An interesting company that is following the services-first model is Magillem Design Services from France, successor to Prosilog which went into reorganization and came out with an employee and management buyout. (Magillem was originally the name of the EDA platform.) Their entry point comes as a design consultancy, their solution seems to be based on their EDA platform, which supports flows in a vendor-neutral fashion (via IP-XACT make-like generators).

    Their approach seems to work, as they had projected a loss for 2009 early last year but ended up with a modest profit! And on top of that, they executed an IPO and are now trading on EuroNext. It was a surprise to dig up that info, as nothing appeared in the EDA-related trade press (if Gary Smith mentioned it … I’m not a subscriber, sorry!). But their success may be a hopeful light in the darkness of the past couple of years …

  22. Kiran says:

    I feel the author is correct. The service comapnies in a way are damaging not just EDA but the traditional product companies as well.

    I am from india. In the last 5 years there is tremendous change in a way VLSI is taken by the Engineers and services based comapanies. thousands of electronic engineers pass out every year and the companies are just cashing the opportunity by paying a small amount as salaries to them. What this makes is the product companies prefer to outsource ip’s which are done by double the number of resources in service based companies at less cost. All this comes with a massive setback in quality. Unless product companies realise that this is not the way, no one can help.

    Its time for product companies to realise that if you want the cost to come down setup a subsidiary in india and work out.

    Anyways i may a too small person to comment on such a large business

  23. Anirban says:

    Hi Oliver,

    Stumbled upon this article. Interesting article indeed.

    So, in April 2013, 3.5 years down the line, how would you rate your prediction? Do you still see the Indian service companies going stronger in ECE/VLSI domain, compared to EDA companies? Surely, they have tried a lot, lured efficient resources, funds have flown into IP/Patenting.

    But, did they succeed?

    Rgds,
    Anirban

  24. After more than 3 years, I saw Indian service companies going very strong in ECE/VLSI. Most of the IC European companies (not many left, you know who they are…) kept the RTL and system design teams at home, but moved everything else –implementation & verification– to India, either using their own facilities (e.g., ST), or using Indian service companies (Wipro, Mindtree, Tata, etc).

    My feeling is that EDA companies’s strategy is shifting towards verification (logic verification, with simulation and emulation), because that is a market that can only grow. One, it doesn’t depend on whether the HW is ASIC or FPGA: verification will always be a must-have. Two, the systems are growing in complexity, and with mobile applications everywhere, the need for SW/HW verification is ever more pressing. Regardless of whether verification is done by the design houses or 3rd party service companies, they always will need more licenses to sign-on on the functional aspects of their designs and implementations.

Leave a Reply