Will India ever get its own semiconductor fab? Should it? In a conversation with members of the EB Bureau, P.V.G. Menon, the founder and CEO of VANN Consulting, shares his views on these and other complex matters of concern for the Indian electronics industry. With a deep insight into all aspects of the ESDM domain, Menon is well qualified to address these questions.
Menon was earlier the president of the India Electronics & Semiconductor Association (IESA) and firmly advocates an ‘India First’ policy for the Indian ESDM industry. He passionately advocates an innovation and IPR strategy, which he sums up as ‘Innovation-led design. Design-led manufacturing’.
The only role in the ESDM sector that he’s not handled so far is that of the policy maker. And perhaps that’s why India does not have a fab, yet.
EB: Can you give us an overview of the semiconductor ecosystem and how a fab fits in there?
As an industry leader once told me, “Semiconductors are the tailwind behind the entire electronics industry.” Simply put, semiconductors are to the electronics (products) industry what engines are to the automotive industry.
In my view, the semiconductor chip ecosystem consists of the following five main components:
- Integrated ODMs that design, produce and sell their own chips – like Intel, Samsung, Micron, etc.
- Fabless semiconductor companies that design and sell semiconductor chips, but use a commercial fab to manufacture them–like Qualcomm, Mediatek, NVIDIA, etc.
- Fabs or fabrication plants that make these semiconductor devices or chips for fabless companies – like TSMC, Global Foundries, etc.
- Assembly, test, mark and packaging (ATMP) companies that take the output from a fab (called a wafer) and cut it, package it and deliver working chips to the fabless company that designed them and got them manufactured at a fab.
- Advanced equipment makers that supply highly complex and specialised equipment to the fabs or ATMP players—like Applied Materials, KLA Tencor, LAM Research, etc.
India has considerable strength in R&D and VLSI design and verification, as well as embedded software. Most of the top semiconductor product and equipment companies have R&D facilities in India, either directly or via third-party partnerships.
We do not have any company in the first category. But almost all the global integrated ODMs have captive R&D centres in India.
In the category of fabless companies, we have players like Saankhya Labs (www.saankhyalabs.com) and Accord Systems (http://accord-soft.com/semiconductor.html).
Then there are SoCs developed by academic institutions like IIT-Madras (the Shakti processor— https://shakti.org.in/ ), which have also worked with silicon. We also had other fabless companies that were acquired either for their IPR, their team or clients.
A subset here is a large number of services companies offering varying degrees of sophistication and skill levels. They offer services in VLSI design, testing, verification, embedded software development, PCB design, etc.
In Category 3 (fabs), we have a digital fab called the Semiconductor Complex Ltd (SCL) at Chandigarh. This is a captive facility owned by the Department of Space (DoS), and is not open to the industry. SCL makes chips for India’s space and defence programmes.
We also have two prototype fabs at the Centre for Nano Science and Engineering (CeNSE) at IISc-Bangalore and the Centre of Excellence for Nanoengineering (CEN) at IIT-Bombay. Both work with industry and are doing excellent work.
In Category 4 (ATMP), we have two companies – SPEL Semiconductors and Tessolve. But since ATMP is very closely downstream from a fab, there is the eternal debate on whether they should come first, or the fab should come first.
In Category 5 (equipment suppliers), almost all the leading players have development centres in India.
Now let me answer your question on how a semiconductor fab fits in the whole ecosystem.
Think of it as a series of concentric circles, with the fab at the centre. The first outer circle would consist of companies that provide services to the fab – like effluent treatment, pure gases, speciality HVAC, etc. The next circle would consist of companies that work with the output from a fab, like ATMP. As you radiate outwards, successive circles would include companies making PCBs, connectors, wiring harnesses, sub-assemblies, etc. Then would come companies making end products like computers, mobile phones, STBs, etc.
Remember, each circle feeds off the previous one, and into the next one. But many of them have ecosystems of their own – ATMP, PCBs, systems companies (which need EMS), and so on.
EB: What is the benefit of the semiconductor industry ecosystem? Are there larger benefits beyond what just the fab enjoys?
I like to compare the quest to build a fab, to the space programme in India. While the space programme was incredibly hard to put together, and we were fortunate to have some incredibly talented leaders and engineers who together built up a world class institution like ISRO, there are parallels we can draw.
For instance, the space programme requires huge amounts of work for complicated computations, algorithms, thermo-dynamics, etc. Some of the systems to address this were imported and the rest were home-grown.
Similarly, the semiconductor industry needs VLSI design skills, modelling and verification skills, highly specialised tools, knowledge of materials sciences, etc. India has these across domains and a vibrant semiconductor industry would help to harness, hone and scale these.
So is the case with some of the ancillary suppliers to a semiconductor fab. Take waste water (effluent) treatment, for example. The technology needed to treat the effluent from a semiconductor fab is very advanced. Now consider the problems linked to treating the waste water of cities in India. Bengaluru, for example, is struggling with a rapidly depleting water table. The kind of waste water treatment technologies that a fab employs can easily be downscaled and local city municipalities can deploy them to solve their water woes, to a large extent.
This is just one example; there are many more. The point is that many of these ancillaries to a fab use horizontal technologies which have wide applications across industry sectors.
The vision is so crystal clear – but it needs a visionary leader to drive this. That is why I strongly feel that it is time for a National Electronics Mission to be set up directly under the PMO, with a pan-ministry mandate to realise this grand vision. By the way, this is how the US drove an agenda to come back as the leader in semiconductor manufacturing and achieved success. The US and Taiwanese models are ones we have a lot to learn from.
EB: What are the main hurdles in setting up a fab in India, and what could be the solution?
This is a very complex question. There is a long and painful history to it, and there have been many unsuccessful attempts already undertaken.
First of all, let us not forget that building Fabs is a ‘Big Boys’ game. The sheer complexity of designing and building a fab, the amount of capital required, the project management skills needed, etc, are so high, that only large companies should even get into this game. So let us be clear that companies attempting to put up fabs are seasoned players who know how the system works, know how to plan and execute complex mega-projects, and have the ability to raise large amounts of capital. I might add that the kind of highly specialised engineering expertise needed to build and operate a fab is not available in India – so at least the initial teams will have a large number of highly skilled expats.
India has tried several times to build fabs with private sector players – for semiconductors, and solar and display panels. These efforts have been going on since 2006-2007. Very large conglomerates have been involved, with marquee top-notch international collaborations. Yet no one has succeeded.
If one or two had failed, one could have faulted the private sector. When so many attempts have consecutively failed, I submit that it is time for policy makers to take a good hard look at why they are failing to meet the expectations of the private sector.
So many consecutive failures, with such a diverse consortia of companies, clearly points to either serious policy shortcomings, or a massive failure of implementation.
First and foremost, setting up a fab involves a massive investment. A display fab would start at around US$ 2.5 billion. A semiconductor digital fab (300 mm) would require a minimum of around US$ 6-8 billion depending on the technology nodes. And remember that these also require ongoing investments and have huge operating costs – so both capex and opex are high. Factor in regular (expensive!) technology updates, and you realise that only companies with extremely deep pockets can play in this space. The risks are high, but if done right, so are the rewards.
So with the huge risks and long ROI cycles for the company investing in setting up the fab, obviously, the regulatory agencies and policy makers need to work extra hard to give confidence to the investors. Policy measures have to be industry-friendly, transparent, rapidly executed and the investor has to have the confidence that the policies will be consistent and not flip-flop. The fiscal support being extended also has to be generous.
Evaluation of applications needs to be efficient and time-bound, while disbursals need to be handled efficiently, competently and quickly. A delay in disbursals means the companies will incur huge working capital costs.
Also, both the Centre and the state have to work very closely as the policies need to mesh together and support the investors putting huge sums of money into such high-tech mega-projects. It’s also worth remembering that the Centre will have to pull its weight here, since micro-benefits to the state (like employment) are not significant enough to justify large outlays from the state budgets. Though there are huge macro-benefits from a fab, a state government will still ask hard questions about what are the localised benefits accruing to it for the investment from its side.
Second, India needs to learn from countries like Taiwan, USA, South Korea, China, etc, on how to set up, nurture and grow a high tech manufacturing industry like semiconductors. All these countries have supported, nurtured and invested in the building of what can best be described as ‘National technology infrastructure assets’ in partnership with the private industry.
In recent times, the USA has demonstrated to the world what I believe has been one of the most forward-looking policies in this regard. The private sector, the federal government, the states and academia have all worked closely to build up world class technology in this area – and this has been across successive administrations. It is this kind of determined and futuristic policy support that is needed to put into place world class national technology infrastructure assets. It is no wonder that Intel, Texas Instruments, Global Foundries and now even TSMC are calling USA home. And this is something that is driven directly from the office of the president of the United States of America.
Third, a private sector digital fab will never be successful by catering to purely Indian demand. However, there has to be a way to assure the investor about demand aggregation for the domestic market as every effort must be made to help the fab operator to get maximum market share. One of the most important success factors for a fab is about how much ‘loading’ it has – i.e., how much is the capacity utilisation. A fab with a low capacity loading will lose a lot of money very quickly. And vice versa.
This is a complex problem and will require out-of-the-box thinking, as well as implementation of existing statutes in letter and spirit. A pan-ministry approach is necessary to make this happen, as well as to smoothen the way for the private sector partner to succeed in the global market.
EB: Considering the challenges, should India even think about investing in a fab?
The answer is complex, and involves both strategic and commercial considerations.
First and foremost, let us look at the strategic intent. Today, electronics plays a crucial role in the development of India. The pandemic has starkly highlighted just how crucial electronics is
In India, the government is looking at electronics as one of the critical enablers to drive an agenda of inclusive growth. The JAM trilogy, authentication of identity using biometrics, dissemination of information over mobile or television – all of these need electronics to make them happen. Therefore, securing our supply chain for electronics is critical to the growth of the Indian electronics sector.
At the core of the electronics supply chain are three things – compute (chips), display (screens) and memory (RAM and Flash). All three require fabs, albeit of different types.
Second, the global semiconductor industry was worth US$ 481 billion in 2018. India’s electronics consumption is growing at an estimated CAGR of over 16 per cent. This translates to a growing demand for compute, display and memory – all of which are 100 per cent imported today. Our consumption of semiconductors was an estimated US$ 12 billion or so last year, and 100 per cent of that was imported. This exposes our supply chains to huge risks.
Electronics is today the second highest contributor to the current account deficit (CAD) after petroleum. India’s young population, rising aspirations and the government’s mission to use technology to drive an agenda of inclusive growth – all of this means that our national appetite for electronics is not going to diminish anytime soon.
EB: Given the long history of failed attempts, and the huge investments required, do you think India can eventually be successful in attracting private investments for a semiconductor fab?
PVG: Bluntly put, I don’t think so. Too many failed attempts means there is a credibility problem. No fab can be built without collaborations with the best technology partners in the world. Almost all the players have been approached multiple times, and corporate memory is long. So any new attempt from India will be viewed with a lot of scepticism.
Having said that, let me add that I would be the happiest person to be proved wrong, and that a private sector fab does get built in India. If that happens, we are putting in place technology infrastructure which will leapfrog India by decades, and will help future generations of technologists to unleash their animal spirits of innovation.
EB: One keeps hearing from a section of the industry that we should forget our dreams of a fab. Instead, we should focus only on creating fabless companies. What is your view on that?
One thing to keep in mind is that investing in fabs or their ecosystem is NOT a substitute for investing in a fabless ecosystem. BOTH are needed for India.
Investing in a fabless ecosystem is important. Back in 2012, as president of IESA, I had set out a vision that in ten years we should aim to have 50 electronic product companies and 100 fabless semiconductor product companies – in other words, as a country we needed to have our own equivalent of a Sony, LG or Samsung, as well as Indian equivalents of a Qualcomm or Broadcom.
There is a wealth of talent available in India in terms of VLSI design skills, embedded software skills, etc. What is needed is to make that transition to defining our own products. The software industry is trying to make that transition – by leveraging the talent and skills of India’s US$ 190 billion IT-BPM sector to the next logical step of seeing what is needed to build a Microsoft, SAP or Intuit out of India.
India’s VLSI and embedded design industry is around US$ 20-25 billion today. Yet we have only a handful of fabless product companies from India, with none of them at a global scale. And we have a US$ 70 billion plus electronic product manufacturing industry in India, which consumes the semiconductor chips that are made outside India.
So we do need to invest in fabless product companies from India.
We have already spoken extensively on fabs. In the end, let me just say that when India invests in setting up a fab here, we are not making that investment for the next few quarters or years – but for the next generation. That’s where the vision and the dream lies, and that is why I believe we need a National Electronics Mission directly under the Prime Minister’s Office with a pan-ministry mandate, and headed by a visionary leader to drive it.