You are here: Home » Companies » News
Business Standard

Should an Indian fab use an older process technology?

Rajiv Rao  |  New Delhi 

The recent Budget waiver of the five per cent import duty on chip-making equipment, as well as another tax deduction of 15 per cent on investments above Rs 100 crore for two years, is the latest smoke signal from the government that it is serious about bringing a world-class fabrication facility (also known as a 'fab') to India. "We want to make a dent against the global supply chain," says Ajay Kumar, joint secretary, ministry of communications and information technology, which is shepherding the

Will a 65-nm fab help?

Exactly what kind of a fab this is going to be is unclear, but sources say it will most probably be in the 65-nanometre (nm) node. It was roughly the technology used to make the Intel Pentium 4 processor in 2006 - seemingly a light year away from the devices and chips we use today. So, the $3-billion question - roughly the sticker price of building a 65-nm fab - is whether this 2006-era technology can service a Rs 3,000 smartphone that a farmer in rural India might be using five years from now to track his crop prices, leave alone competing with the best of breed phones around at that time.

and are the fastest-growing segments in the electronics industry. Smaller nodes give birth to smaller chips, which, in turn, breed smaller, faster and less power-consuming devices, thanks to Moore's Law that says chip size halves roughly every 18 months. Essentially, miniaturisation means tinier transistors, more of those on the chip and, therefore, quicker switching speeds, which lead to lower power consumption and a faster, cooler chip. Most cutting-edge fabs will be upgraded to 20-nm by 2013-end.

However, choosing a 65-nm node might seem more disastrous than it actually is. According to a McKinsey report, leading edge nodes comprise only 14 per cent of total demand for logic chips and micro-components. Industry observers say at least half the customers of global fabs use chips of a previous generation, 40 per cent use earlier nodes and only around 10 per cent require the best chips money can buy to power their latest device offerings.

Apparently, design innovation can further mitigate the need for a cutting-edge fab. "Leading nodes are more about generating volumes at stable conditions," says Jaswinder Ahuja, managing director of Cadence Design Systems, India. "Globally, the maximum consumption is around 180 nm," says P V G Menon, president, India Electronics and Semiconductor Association (IESA).

Still, using outdated technology can pose problems. According to industry experts, today's chips shrink four billion transistors into a space smaller than a postage stamp, thanks to the method of tiny writing called lithography that etches complex patterns onto a wafer using electromagnetic radiation. In another four years - by the time the purported India fab is up and running - Moore's Law would be at the end of its reign and the industry would be poised to embrace the next big leap in technology and design.

As it is, smaller 'gates,' in chips and larger wafers (which house chips), continue to increase the investments in cutting-edge nodes by 20 per cent (of the cost of the fab) for every upgrade. Also, all the leading-edge fabs are beginning to invest in transitioning from 300-mm to 450-mm sized wafers. According to the Semiconductor Equipment and Materials International (SEMI), research and development for this transition alone is estimated to be between $8 billion and $40 billion, with expenses for the first generation of high-volume manufacturing going past $25 billion in 2016-17. Ana Hunter, vice-president of foundry services at Samsung, has suggested building a factory to make the next generation of chips will cost upwards of $10 billion.

A 'deep pockets' business
This is why entering the world of semi-conductors is only for those with deep pockets. "Every day a fab is not loaded, you lose $10 million," says Cadence's Ahuja. Today, the big guns in the industry are only a handful such as Taiwan Semiconductor Manufacturing Co and United Microelectronics Corp, both contract suppliers; and Toshiba, Samsung and Intel, all integrated device makers (IDMs) who design and build chips for their own products, as well as for select others.

A large number of firms such as Freescale or Texas Instruments are becoming 'fab-lite' - they do initial work on the chip, but get them churned out in large volumes by foundries. The crippling costs of building a fab and the volatile economic climate have pushed long-time US chip-maker AMD to spin off its manufacturing operations into a joint venture with cash-rich Abu Dhabi, called Global Foundries.

So, will the fab in
India make money? "Probably not, especially in the short run" says IESA'a Menon. Israeli fabricator Tower Jazz, for instance, which is apparently one of the contenders for the fab contract, has not made a profit in three years, an indication of how tough this business can be.

It doesn't help that the world has a 15 per cent overcapacity in chips at the moment.

"Can India compete on price, drop prices by 25 per cent, or get better yields?" asks Cadence's Ahuja. "It's probably unlikely in all of these scenarios," he says. However, both Ahuja and Menon feel the fab, in general, is a step in the right direction. Ahuja thinks real traction will be achieved when the government becomes a lead investor in the project -something that it has so far stayed away from addressing.

The real opportunity
Based on these ground realities, the real opportunity for an India fabricator is not necessarily in competing for a slice of the frenzied and rapidly-evolving market for the latest gadgets and gizmos, but in another area entirely - one which makes indigenous products as solutions for indigenous problems. But even here, product companies, which the government hopes will organically mushroom, will have to undertake a giant leap of faith in investing in manufacturing facilities today for a fab that will start working only in three or four years. Still, many years ago, a little known company called CDOT started a communications revolution by modifying an existing technology for the local terrain in a novel way.

Given an intelligently-orchestrated push, it could happen again.

This is the first of a three-part series. Tomorrow: Focus on Indianisation

First Published: Thu, March 21 2013. 00:49 IST