How do you go about building self-reliance in semiconductors?

At the ongoing IESA Vision Summit 2020, there was a fireside chat on the topic: Building self-reliance in semiconductors. The participants were Dr. V.K. Saraswat, Member, NITI Aayog and Dr. Satya Gupta, Chairman, IESA. The panel was moderated by Ajai Chowdhry, Founder, HCL.

Ajai Chowdhry, started the discussion saying, should we look at hardware and software? Dr. V.K. Saraswat, NITI Aayog said that a few companies are going to import from China, for strategic sectors, such as railways, etc. Today, the basic digital platforms have security as a major issue. It is important for India to have design and manufacturing capabilities. A certain level of self-reliance is required. The pandemic has made this happen. Imagine, what will happen, if we do not have these capabilities! It is important for us to set up a system in the country.

Dr. Satya Gupta, IESA added that semiconductor assembly-test-mark-pack (ATMP) is very crucial for India. We have to work on the labs. The EMS players can consume the outputs from ATMPs.

Ecosystem focus
Chowdhry noted that the modern warfare is generally fought with semiconductors. What will the ecosystem be like?

Dr. Gupta said that there are two important ecosystems. One is electronics, and the other is semiconductors. R&D is very important. So is manufacturing, and you will need logic fab, etc. In the short term, we need to go fabless, in the mid-term, we needt to look at a fab and ATMP, and for the long-term, R&D. "We need to bring the sector to that level of autonomy. For R&D, there are excellent models. We can emulate them, for India. We need to work on all three fronts -- electronics, products and R&D. We have to take a long-term view of about 15 years. Self reliance is of utmost importance for any country."

Dr. Saraswat felt that this is a catch-22 situation. "A lot of electronics are being manufactured here. We need to translate ourselves by designing components. At least, the designs we are looking for the domestic market, will be available. Let us select 10-15 products that can be designed and manufactured."

He added that the demand for semiconductors should be met by the foundries. "The breakeven points for foundries come after 7-10 years. The same thing will happen here! Fabless foundries will make use of the foundry infrastructure. At least, for 7-8 years, one should not look at making gains from fabs."

In that case, should we get some fab in the country, or should we go for 28nm fab?

Dr. Gupta suggested that we don't decide what components are being used. "Our estimate is that we are consuming $1 billion in 28nm. We are going to see that stay for at least 5 years. We need to learn to develop the technology and business integrations. We should later catch up with the latest technology. 28nm still has lot of potential, and so has ATMP. There is also the importance of fabless. We are working on a plan for 100 fabless companies in India over the next 5 years. India did not have any major investment in hardware. We can get the fabless, and later, fab and the ecosystem."

So, can India get a 28nm fab? Dr. Saraswat added: "If we can lay hand on a 28nm fab, it will be a jumpstart for the nation. We should try and see through the 28nm, and follow the 7nm and 5nm direction. We should try with what we can get, and then jumpstart."

Chowdhry felt that at least, we can have something started. The ATMP can also start. The R&D will also start. Let's get the ecosystem started.

Dr. Saraswat noted: "One of the options was to acquire an existing foundry. The investment in a fabless foundry is interesting. Eg., Imec. They acquired from a variety of sources and then gave them to different designers. The fabless foundry system has to be supported by the government."

To this, Dr. Gupta turned the attention to the SFAL or Semiconductor Fabless Accelerator Lab, in Bangalore. An IESA initiative, it is meant to incubate early-stage startups and existing SMEs in the fabless semicondutor segment. "We are beginning to build the ecosystem and scale it up. We have relationships with Imec, Samsung and Globalfoundries."

Chowdhry noted that there have been discussions about digital security and semiconductors. We need to have much higher valuation.

There is 6-8% PLI scheme. We need to support the scheme and have a large strength. We need to support the system. That will have much better impact on the ecosystem. We need immediate action for specialty fabs. We need fabs for power electronics, as well. We should even extend the PLI scheme for fabs. Also, he wondered, what will SCL do next?

Regarding SCL, according to Dr. Gupta, over the last 5 years, they have done well. Reliability is more important. We need to move to modern MPUs. We should upgrade SCL to at least 65nm.

Dr. Saraswat concluded: "There are other countries having fabs, besides the USA and China. They are manufacturing all kinds of electronics and products. We hardly manufacture any equipment in the country! Companies are not using indigenous equipment. One other weakness in the country is the understanding of semiconductors as a subject. We need to build intelligence for all this. There is a need for R&D for semiconductor materials and manufacturing, as well.

Dr. Gupta agreed, saying that Apple controls where they manufacture, who are their suppliers, etc. We must try and do something similar at the earliest.

Chowdhry concluded that we need a whole ecosystem. We also need to have a timeframe for doing this.