It’s a simple enough question: Is there now, or will there be at some point going forward, an India-based EDA industry? The reason for asking the question seems obvious. The answer is not so obvious.
The Republic of India is the second-most populous nation and, per Wikipedia, the largest democracy on earth, with 1.1 billion people, 28 states, 7 territories, 22 official languages, and 1600 minor languages and dialects arrayed across 3.3 million square kilometers. Hindi and English are considered by many to be the “principal” languages in India; Hindi is spoken by over 200 million people, and English by over 400 million.
India is home to the second-fastest growing economy in the world (after China), and that circumstance is the sub-context of this article. Per The Economist (February 3, 2007), India’s economy expanded by 9.2% in 2006 and is on track to outpace that rate in 2007. Whether that growth rate is good or bad for the country, or the world, is yet to be determined, but the immediate impact is clear -- India is hot. Business opportunities are coming online in India at a far faster rate than most could have predicted even 5 years ago, and the semiconductor sector in India is part of that landscape.
Last month, the newly formed India Semiconductor Association hosted the ISA Vision Summit 2007 in Hyderabad. Reading from the briefing document associated with the event:
“The Indian semiconductor design industry, comprised of VLSI design, board design, and embedded software companies, has design companies across Bangalore, Delhi & Noida, Hyderabad, Chennai, Pune, Ahmedabad and Goa. All of the global top-ten fabless design companies have India operations, and 19 of the top 25 semiconductor companies have a strong presence here.”
“The semiconductor design industry in India had [gross revenues] of US $3.2 billion in 2005, with an engineering workforce of around 75,000. It is estimated to reach US $43 billion by the year 2015 and [to] provide jobs to 780,000 professionals with a CAGR of around 30% for this period.”
“With its growing middle class population of nearly 400 million people, which will only increase over time, India’s electronic equipment consumption, which was estimated at around US $28.2 billion in 2005, is expected to reach US $363 billion by 2015 growing at a CAGR of nearly 30%. Indian electronics equipment domestic production was US $10.99 billion in 2005, and [is projected to hit] US $155 billion in 2015.”
If your eyes have not glazed over at these predicted red-hot rates of growth, you can see that India, both as design community and end-product market, is emerging as a force to be reckoned with. Hence the question: Will an India-based EDA industry emerge to meet the demands of India-based designers? Not “off-shore” R&D facilities for the currently established players in EDA, but new, locally-based EDA companies built on Indian venture capital expressly supplying the growing needs of the Indian design community.
EDA vendors constantly argue that they need to be “close to their customers.” If the design community in India is set to increase 10x over the next 10 years, will India become one of the global epicenters for emerging EDA companies at the same time?
To answer these questions, I spoke by phone with five people involved in the EDA industry, and exchanged email with a sixth, all of whom have ties to India either by birth and education, or through business, or both:
Kamal Aggarwal, Vice President of Marketing & Strategy at SoftJin Technologies (Bangalore)
I was surprised by the fairly consistent answers I received from the first five on the list, and the contrasting assertions put forth from the sixth. Five appear to say that something’s coming, but it’s not here yet. One appears to say that something’s already here. Perhaps only time will tell who’s right.
There is a lot of tool development going on in India, but [it is] predominantly for U.S.-based EDA companies. Depending on the company, they may have 10-to-20 percent of their staff based in India working in tools development, and activities related to development. At the same time, there have been a few India-based startups. For instance, SoftJin started in 2001, and since that time we’ve also seen a couple of other startups in India.
Our take on EDA is slightly different from the predominant product model. We do customized EDA. We develop EDA tools for specific requirements of semiconductor companies [who] ask us to develop a tool not available on the market that they would like to have for in-house use. Primarily, we partner with either in-house CAD groups who’s mandate it is to develop the tools, or we work as tools developers for other EDA product companies who want to make use of our services or offerings.
There are always challenges for new players [trying] to make a dent in a mature market [like EDA], but we’ve been growing at a reasonably healthy clip. In 2004 we had 14 engineers, and [today] have about 100 engineers [providing] post-layout tools and tools targeted at programmable platforms.
Our marketing approach in other geographies is to work with customers in the U.S. and Japan -- a challenge when we are headquartered here in India. [However], having India as the place where we do tool development has not been an issue for our customers, apart from the business issues related to manpower or IP protection that any prospective customer has to deal with [working across] large geographies.
The EDA market in India is growing. It started from a small base, but is growing as semiconductor companies set up large facilities in India. We are not selling to semiconductor end users, but to tool developers, so to the extent that tool development has moved to India, for us the EDA market is right here in India and right now. Three or four years ago it would have been difficult to [support an India-based EDA industry] without a critical mass of designers to develop tools locally. Now we have local developers and are able to find local, experienced designers who we can test ideas on, which is very useful.
From the supply side, there is the dynamic today that a lot of EDA tool developers with long experience working in U.S.-based companies are looking to move back to India to be part of Indian companies. By offering a challenging opportunity, we are offering an invitation to come back to India where [the engineers] were born and brought up.
And experienced engineers are not just coming back in EDA, but in VLSI design as well, to work for Indian companies or in the Indian centers of foreign companies. For instance, Vinod Malhotra, our VP of Engineering, has long years of experience managing R&D post-layout tools. He recently came back to India [after working at Synopsys in the U.S.] and has joined SoftJin because he wanted to be part of an Indian EDA company generating new products.
It’s never an easy decision to make [this move]. Silicon Valley has a very unique blend of cultures and provides a huge opportunity for intellectual people to grow and work on challenging problems. The one thing we have seen, however, is that gradually the conditions are emerging in India where similar opportunities are arising here and giving an inducement to return.
[In addition], our universities are becoming increasingly sophisticated. In the past, a good proportion of our top students went abroad for higher studies or to work. That proportion is coming down as a lot of engineers here are looking to get absorbed into local companies.
Will there eventually be a complete semiconductor ecosystem in India? I believe we are gradually moving to a point where there is valuation in all points of the ecosystem -- tools development, usage, design, and manufacturing. India is becoming a growing market for semiconductor chips, so it’s natural that this could lead to an integrated India semiconductor ecosystem. That’s [certainly] the vision of the India Semiconductor Association -- to move towards a point where more and more of the parts cooperate with each other.
I have been encountering, however, a debate as to whether or not India should be putting up a fab, or a latest-technology fab. I think the decision will mainly be driven by market forces, and is best left to the people trying to make these investments based on the financial viability of the projects. They see growing consumption in India and may want [to bring] the manufacturing closer to the market.
It’s difficult for me to say whether [India should] invest in semiconductor manufacturing or not, but there are arguments on both side. Manufacturing is a volume game, and we’re a little late getting into that game. One the other side, in order to capture a larger portion of the semiconductor value chain, you’ll have to move manufacturing to India. Some companies in India are putting their resources into this [idea]. It’s a major decision because of the investments involved.
I have been associated with India in one way or another since 1995. That’s when I did my first entrepreneurial venture, moving from the U.S. to Bangalore to start Synopsys India. In those days -- beside Cadence -- there wasn’t anything in India [by way of EDA]. I spent 3 years starting the Synopsys organization and development team.
In 1998, I left the EDA space and went into the mainstream fabless IC space with a couple of startups before ArchPro. These startups both had a significant presence in India in varying degrees. Of course, in those days I was a user of EDA tools not trying to develop and sell tools. With ArchPro, I came back to the EDA space to focus on certain power management problems.
ArchPro has a unique model. Clearly we are headquartered here in California -- most of our executive management is here, plus engineering, marketing, and a couple of architects. But predominantly our development is in Bangalore. Also, we have a significant applications engineering force in India, mainly to cater to the local customers and also as application sales providers in Japan, Korea, and Europe.
[You have to look closely at the reasons the EDA companies first went to India.] When I approached Synopsys in 1994 with the idea of having an off-shore Synopsys India, it was the golden years of EDA -- Synopsys and logic synthesis were on a tear. There was not really any economic pressure at all to make the move to India. The primary purpose of proposing and starting the center in India was that most of the EDA players at the time -- Synopsys, Cadence, and Mentor -- were predominantly a workforce with software-oriented people. Even though the SOC and chip designers [were using the tools], there was very little hardware talent in the companies.
My main idea was to go off-shore to [have access] to chip and system-level designers, to take advantage of having those designers in-house, to develop domain-specific knowledge, and to have real chip designers using the tools and creating IP. That was the motivation for the move the India.
At the time, it took a while for people to comprehend all of it. [Of course], now it looks like a solid proposition. India now has a number of design centers and a number of [different companies] doing ICs there. It’s an entirely different story today -- the challenges that the EDA guys are facing, both as sellers and buyers of tools.
So, broadly, there are two categories [of topics] related to doing EDA business in India. Clearly, EDA companies that are founded here [in North America] have off-shore organizations in India. ArchPro’s in this first category. But the second category is having EDA organizations in India itself, trying to do business there. If you really look at this “flat” world, things have changed significantly -- there’s not a huge wall [between geographies] like in the old days.
If you look at the exciting players in EDA -- Synopsys, Cadence, Mentor, Magma -- they do all have R&D centers in India, the first category, but it’s only a certain percentage [of their effort]. Things get done there, and certain activities are supported, but today it’s more [because] of the cost issue.
However, within the second category where people are thinking of doing EDA work in India, it’s possibly a lost cause. Even here in Silicon Valley, EDA is such a small space. It’s only a $4 billion or $5 billion market. Yes, anybody can start a company, but it’s not really viable to have something in India under the current business models or at the current stage in the industry -- unless, of course, there’s a sweeping change in the whole design chain.
For instance, with new process geometries, or things like power, that we’re addressing, we’re talking about very innovative stuff. Can that happen in India? Possibly, but you have to look at the big market. If it’s there, perhaps a startup can work in India.
Another [concern] -- as a vendor, if you’re trying to sell your products in India, most of the decision making today for the multinational organizations happens outside of India. That means the selling process has to happen in both places. You’re selling value to the design community in India, but [handling] the legal part and contracts, etc., outside of India. Most of the design centers out there [in India] are focusing on the engineering. There are no EDA managers [on site], no contract person to [handle the details]. All the big companies make purchasing deals for the whole company [at a central location]. That may change over time, but right now it’s quite natural
The other interesting thing is that if you look at most of the companies in India, they have an STP organization, and most people don’t want to break away from that. If we sell into India, we have to use a different infrastructure to do the selling there versus developing relationships with the customer here. Synopsys, for example, has 2 distinct entities -- one for R&D and one for selling. They can still be a STP, but they are different tax entities and the people who handle the legal processes -- the kind that originate in India -- actually do it outside of India.
[Having said that], there is a little bit of EDA emerging in India. Several of the large design services and IP companies are finding a lucrative market in India, but they continue to service customers worldwide.
Technically, you could start a company in India, but I would not do it. In a startup, you need access to the people [who can help you do things], and fundamentally I don’t believe that’s possible in India. But again, if there were a fundamental change in the design flow -- the first in 25 years -- a new set of products would be required and there might be new opportunities.
EDA is a very difficult niche for investment and new startups, but there are huge opportunities as well. ArchPro is an example. There are changes that are going to have a profound impact on the flow, whether the change is being driven from here or from India. I’m very bullish on EDA for the short and long term.
The recent ISA Vision Summit was a great opportunity to learn what’s going on in India, with over 250 CXOs and VPs of engineering in attendance. While I was there -- I was one of the speakers at the conference -- I was surprised to learn that there are now 12 or so EDA startups in India.
The total EDA workforce in India today is in the range of 2000 engineers working in Noida, Hyderabad, Bangalore, and Pune. Also, [my understanding of the approximate numbers of employees by company are] Cadence with 800, Mentor with 300, Synopsys with 200, Magma with 120, Atrenta with 230, CoWare with 30, Sequence with 35, ArchPro with 10, Synfora and Bluespec with 20 each, Calypto with 30, Apache with 10, and SoftJin with 100 or so. Of course, these numbers are approximate and can change, but it gives you an idea of the size of the EDA community there today.
When I was in Hyderabad at the Summit, I found quite a few companies’ hiring managers. They say they’re getting tons of resumes from people who want to go back. People who are not from India are moving there as well, learning Hindi, and finding it to be a fascinating culture.
As recently as 4 years ago, I didn’t go to India that often because there was no compelling business reason to do so. Now, there are compelling reasons to go. It’s all about globalization and things happening in China and India. There are over 125 companies who are potential EDA customers in Bangalore, Hyderabad, Dehli, and so on -- wi-max companies, mobile computing companies, and so forth.
Also, the engineering colleges in India are producing more and more advanced-technology engineers, and successfully measuring up to international standards. The IITs are, of course, excellent. But there are also lower-level colleges turning out quality engineers. [In fact], there’s been a sea change in engineering education, and not just in India. Many universities are moving [away from the idea] of a campus and going “worldwide” with web-based training. The IITs are seeing the change and understanding that branding is important, even for universities. There will be even more shifts [in this area] over the next 5 years.
The only thing holding back EDA in India is the training issue, and questions of how you train the resources you need in chip design itself. I’ve seen tremendous progress over the last several years -- government training of VLSI designers with investments from companies like Cadence, Synopsys, and Intel. Training institutes themselves have become profit centers. People are really hungry to learn chip design.
While I was at the Summit, I also talked with [various government officials and professors] and learned that there is a move to make Calcutta a leading-chip design center [to rival] Bangalore. They have millions of dollars for the project and I was amazed -- I have never seen politicians speak so elegantly about technical education. I was really shocked to see such savvy. They are not old-fashioned bureaucrats, but next-generation leaders. Jody Shelton, the executive director of the FSA, was at the Summit as well. She told me that she had never seen politicians anywhere speak so well about technology either.
Brian Lewis from Dataquest was also there, on a panel talking about a key controversy going on in India. Does the country need a 90-nanometer state-of-the-art fab? The controversy actually was a little bit emotional. People were saying, “You guys always told us not to create manufacturing -- first in cars, and then in chips. But we wouldn’t be behind in those things if we had started back then.”
The India Semiconductor Association is raising the capital to create a fab to compete in the market. [However], it’s my opinion that they should not go and chase a fab for the sake of a fab. There are already expert fabs elsewhere -- really, how can you compete with TSMC? Instead, they should focus on applications in India that don’t require 90 nanometers.
For instance, products like the Texas Instruments’ Speak & Spell chip from many years ago was a type of application that had a huge market, but didn’t need next-generation technology. They should let the design community try to find these applications. Let the community find the next killer app, whether it’s solar, or handheld, or an Internet application for farmers, which in itself is a market of 600 million people in India.
One note in all of this discussion about India does worry me, however. Intel is saying to us that they have 2000 designers in India, but they’re not paid on U.S. salaries, so why are companies like Cadence, Mentor, Synopsys, etc. using U.S. pricing for their tools. They’re asking if there shouldn’t be a local cost-of-living adjustment for the tools. That’s definitely a trend we all need to watch out for -- pricing today is based on a worldwide LAN, not on local pricing.
So yes, EDA will continue to grow in India, but the creativity is still in Silicon Valley. Of course, Silicon Valley doesn’t have a lock on creativity. If the developers in China and India want to create their own markets for EDA tools, they can do it.
An India-based EDA industry is far from happening -- the domain expertise is just not there to do it yet. And even if they have the domain expertise because some of the customers are there, they have little of the necessary business expertise. It’s one thing to build the technology, and something else altogether to build the business. It will be at least 3 years before I could see it happening, and when it does people will see early-stage companies fail.
Right now there are some companies like SoftJin in Bangalore which do consulting, and there are some companies doing QA, but these companies are providing design services to EDA companies to make their products more stable, better, or more tested. Those types of businesses I can see [succeeding]. But in terms of doing really good EDA in India, not yet. The management talent in India is still extremely weak, with little experience managing customer requirements.
Will that change going forward? Yes, in 15-to-20 years we’ll see as many startups in India as we see in Silicon Valley around anything having to do with IC design. That’s the case already in India in IT. But it’s a pretty aggressive business model in EDA, and [for now] the VCs are only just starting to appear in India.
The EDA industry is one of those places where the “Me, too” [factor] has not changed. You have to be significantly better than the other guys to succeed. You have to know the customers and the business side of the industry, and the execution has to be really good. You have to have productivity and understand the number of people doing chips. This is a business that’s really cut-throat. An EDA startup only has 6-to-9 months to make a dent, or you will become the living dead. If you don’t succeed in that time you will fold. The sad thing is that this is an evolutionary market, not a revolutionary market -- it takes a lot more venture capital to succeed.
Will there be manufacturing in India? Speaking as a person from India and an entrepreneur living here in Silicon Valley -- from an expat’s point of view, India should not get into manufacturing. The only reason to do it would be if the costs would be cheaper than manufacturing in China. [I don’t believe] India can compete with China with a 300 mm, 65-nanometer fab.
Between the amount of investment it would take [to build the fab], the amount it would take [to operate the fab], and the discipline it would takes -- it would require a lot of time to succeed. Even China has not had big successes so far, so what makes India think they can start 10 years later and succeed? I know there’s a lot of patriotic reasons for building a fab, but purely from a technical and business perspective, it doesn’t make sense.
Most of the work being done in EDA in India has been support and R&D, starting with the multi-nationals opening centers in India. But that’s a more recent development that went hand-in-hand with support -- field application support engineers followed the design centers into India as they were established by the semiconductor companies.
Now we see that even the foundries are getting in. There was a small news item recently from TSMC announcing a new sales office in Bangalore. It’s actually surprising that it took so long for TSMC [to make this move] as they’re one of the more popular fabs in India.
From a design-tools and support-engineering point of view, EDA has been gathering steam in India since about 2003. The EDA commercial tools market is growing really fast in India. For instance, SoftJin is one example of a locally-funded EDA tools company. They started out as a service provider, and then commercialized a product. That’s a strong path to follow. [Of course], if there is going to be more and more electronic product design in India, they will need to find the tools somewhere, so we could see some homegrown tools companies coming online in India to support local fabrication plants.
Perhaps in 15 years or so, we will see India-based EDA companies -- or maybe even in 10 years. But a lot of the research that supports EDA development comes out of the universities, and I don’t think the Asian and Chinese universities are yet able to develop breakthrough technologies that can be commercialized. That will be the gating factor that will determine the timeframe within which an Asia-based EDA industry will emerge.
However, globalization is happening. You can fight it kicking and screaming, [but you cant stop it]. I live in North America, but I have a foot in Asia as well through my work covering both markets. We see that software has revolutionized globalization. There used to be a disaggregated industry that was regionally specific, but now there is design going on in North America, in Japan, and everywhere else as well.
As the semiconductor industry evolves, so will its supplier companies. The EDA companies are well advised to consider that they, too, are participants in this globalization movement. There will always be a need for local specialists, but more and more of the work they do will affect a global audience. In 15 years, it won’t matter in the least where an EDA company is based, where the R&D is done, where the software is done on the server. It simply won’t matter.
Meanwhile, with regards to semiconductor manufacturing in India, a new policy statement has just been released by the India Semiconductor Association. I haven’t seen the document in detail as yet (as of late February), but in the past the government of India has shown that they really understood how to support the software market. It took them a while, but they set up special industrialized zones that had tax benefits for setting up software companies in those zones. That enabled the software market to take off, in terms of services, and to a lesser extent, products.
Now the government is trying to do something similar for the electronics industry. One of the things generating a lot of interest is the idea of building a foundry. There are a couple of engagements underway -- one is to set up a fab in Hyderabad. However, there is a debate over whether this makes sense or not.
From my perspective, it’s really a cost-versus-benefits issue. The costs are infrastructure-type costs. The Indian infrastructure is really nascent at this point. It needs to be at a more mature level to support a leading-edge fab. However, it can make sense to have a local foundry because local companies [in India] may not always have the volume, or the muscle, to be able to get what they want from the independent foundries. They might get a better deal from a local fab in terms of negotiating smaller volume devices, but you don’t need a leading-edge fab to do that. A mainstream fab is good enough.
There are fabs today in India, of course, but these are all older technologies in captive fabs for the defense electronics sector. There was just one public/private consortium with a fab at the 1-micron technology level. They brought it down to .35 micron, and then brought it down again. Does that make sense? Not really when you can use Singapore in the region.
So, although it may not make sense to have a fab from an economic or technology perspective, however, it may make sense from a strategic point of view.
Raju Pudota, Managing Director for Denali Design Systems India, is quoted in the email I received from Denali addressing my questions about EDA in India: "The EDA industry is seeing an upswing in the number of startups innovating for solutions to deliver sub-90 nanometer development, be it in the area of implementation or in the area of enabling simulation of low-power design, multi-voltage solutions, etc."
“Given the success the current EDA industry is enjoying in India, and the availability of the skilled resources developed over the many years the industry has matured in India, India is an ideal choice of location for design centers to take advantage of the readily available skilled labor at a very competitive costs."
"With a focus on providing packaged solutions to its customers, Denali is working towards transitioning the delivery boundary to the customer, where the delivered product will be a combination of IP and embedded software. With an aim to enabling seamless integration of its IP into the customers' systems, Denali is investing in a specialized embedded software team, and building resources in India for research and development activities pertaining to system architecture, flash-file systems, complex algorithms for data management, and portable memory subsystems."
The Denali email also noted: “With the design teams in India moving up the value chain in terms of the number and the complexity of designs, with the manufacturing industry all set to get rolling, and the support ecosystem beginning to mushroom, India provides an excellent location for EDA industry to set up R&D centers.”
“Today all leading EDA companies have design centers in India. They are clustered around two locations, namely New Delhi and Bangalore, also known as the Silicon Valley of India. Denali has set up a design center in India to focus on providing leading edge EDA and IP solutions to its customers worldwide.”
“[Meanwhile], with cost pressures on the rise, India is continuing to be a hotbed for investment in the EDA space by startups and small EDA companies. This last year has seen a lot of new companies investing in India. ArchPro and Sierra are a couple of examples.”
“Given the worldwide trend pertaining to EDA startups, India will also see an upsurge in the number of EDA startups over the next couple of years. These startups will be funded both by global venture funds (with an India-specific focus), as well as local venture funds. The new semiconductor policy, released by the India government in February 2007, will provide the much needed impetus to fuel this growth.”