There is no doubt that India has huge potential in terms of the market for semiconductors. In 2013 it imported $33.5 billion worth of electronics, from semiconductors to smartphones. The country’s demand for electronics products is estimated to rise by more than an order of magnitude to reach $400 billion by 2020. This has caused the country’s policy makers to worry that the value of electronics imports, with no major local manufacturing whatsoever, could exceed that of oil imports.
For at least ten years now, India has been chasing the dream of setting up semiconductor wafer fabs. In 2005, there was a false start involving SemIndia, a consortium of Indians living overseas and having a collaboration agreement with AMD. It faltered and died a slow death.
After a long delay, an empowered committee headed by Dr. M.J. Zarabi was formed two years ago to work on the feasibility of wafer manufacturing in India. In February, the committee’s efforts resulted in a long-awaited announcement and the issuance of letters of intent for two consortia wanting to set up semiconductor wafer fabs with a total investment of $10.17 billion. One consortium is made up of India's Jaiprakash Associates Ltd., Israel’s TowerJazz, and IBM. The second comprises HSMC Technologies India Private Ltd, Malaysia's Silterra, and STMicroelectronics.
But the burning question on everyone’s mind is will India be able to sustain this effort and make a success out of it? Len Jelinek, director & chief analyst, Semiconductor Manufacturing, IHS, says “Although this is not a clear cut yes or no, I am leaning toward yes. If a fab is heavily sponsored it will meet the objective of the government, that being to create high value-added jobs. And if there are domestic incentives provided, then designers in India may actually use the fab and make a success out of it. Other than those actions there is no reason for a company to transfer manufacturing to a new manufacturing venture other than cost.”
He added that if the Indian government is planning to contribute funds for a long period of time, then the venture may be reasonably successful. “The government needs to look at what has taken place in Malaysia (Silterra) or even at the business model used initially in Singapore (Chartered) and decide whether it is willing to support this venture for the long term,” he noted.
But there is another section of the Indian industry that feels that although there is a tremendous sense of patriotism in saying India has a couple of fabs of its own, there is an equally huge dose of skepticism in the marketability and eventual success of these fabs.
First off, there is hardly any chip design activity done in India. Companies like Wipro and HCL perform chip design in India, but the designs go to the companies’ clients, located outside of India. There are very few examples of small successes. If there is a successful startup, it invariably gets acquired by a global company and becomes the global company’s R&D center. Examples are Beeceem’s acquisition by Broadcom, Cosmic Circuits’ acquisition by Cadence, and Spike Technologies’ acquisition by Qualcomm. So far, there are no examples of free-standing Indian companies that have been successful in chip design and that have not been acquired.
“If you don’t have chip designers in India, the only value a fab can add is to become a global fab. You are not servicing any local buyers because there are no local buyers, unless you magically create local chip design buyers,” says an industry expert who prefers to be unnamed.
“It would happen eventually when all the dots get connected. The chip design industry should develop here so that they buy the local wafers and then sell the electronics to those who will work on design in India, who in turn will sell those electronics to Indian buyers who are going to consume in India. Yes, there is a path to it but it is a long drawn-out path,” he added.
Others are a bit more optimistic. “But a differentiation strategy is vital if India plans to be successful, as it is starting out late, nearly 30 years after Taiwan, Singapore and South Korea. While Taiwan, South Korea and China will remain key players, the semiconductor paradigm changes can create viable opportunities for potential fab nations like India to capitalize and become a successful semiconductor fab nation,” says Raj Kumar, CEO, Innovative Global Solution and Services Pte Ltd (IGSS), Singapore. Incidentally, he has been working with global companies like National Semiconductor, Chartered Semiconductor and Global Foundries and is exploring fab opportunities across Asia, including India.
“Indirectly, [having fabs in India] would put India on the map of high technology, it would create jobs, it would clear the path for peripheral industries to start happening, like testing and packaging and others that need to be co-located with the fabs,” he added.
Still, there are serious challenges. Says IHS’ Jelinek , “Let’s start with the obvious one – technology. No one is clear yet on which technology will be used in the fabs. Clearly they cannot compete with the market leaders – TSMC, Globalfoundries , UMC and SMIC. This means they will be installing legacy technology which is available at many other facilities. When you cannot differentiate yourself in terms of technology, you only have price.”
According to Rajiv Kapur, Managing Director, Broadcom India, “A fabless company like ours would be happy to see successful choices around the globe. Now what an Indian fab will specifically offer is still not very clear. Is it cutting edge tech or a cheaper trailing edge? If it is a cheaper trailing edge technology, we have to see their success in quality, yields and price points. But it has to be as good as TSMC or the other fabs we are currently using. Our corporate office makes the decision about which fab we ought to be using, and if the India fab offers something unique we would certainly look at it.”
Infrastructure is another major challenge. The facilities, power, water and transportation required to support a fab are critical and very expensive. If the government cannot get all of these basic services upgraded, then the project will most likely fail.
The list of challenges goes on. The long gestation period and the technology mandates of the government have also raised questions about the usefulness of the projects when they finally come up. The government has mandated vendors to produce chips at the 90, 65 and 45 nm process nodes. The chips would go into devices such as medical equipment and low-end handsets. Chips manufactured at 28 nm and 22 nm and that go into high-end products are likely to come up in the second phase.
“The best case scenario is that in three years the backers of these ventures will be able to convince a multinational IDM to transfer manufacturing. Most probable scenario is that the facility is built and partially equipped. The facility will then operate at significantly sub-optimal run rates and the government will look to either sell it, pressure the backers to meet the original goals, or close the facility,” Jelinek noted.
But striking a more optimistic note, Broadcom’s Kapur says that there could be approaches to “Indianization” where the government would step in and say that if the fab products are eventually imported back into India, there could be certain economically significant benefits. “These benefits can then be passed along to the entire supply chain, so that if the chip becomes cheaper, the electronics manufacturer can get a cheaper component, and they in turn can sell the end product cheaper. At the end of the day, India can benefit”