Semiconductor manufacturing in India – its time to get on the fast lane
Date: Sunday , February 04, 2007
Success in IT and even chip design has created eagerness among the software-heavy IT community in India to move up the food–chain viz, manufacture chips in–country by setting up Wafer Fabs. Over the last two years several promoters have been doing the rounds proposing to set up state of the art multi - billion dollar Wafer Fabs, supposedly with technology supplied by leading IDMs. Quite predictably nothing has come out of these schemes except swaying the Indian media and even the Government to the irrational exuberance of wanting to start semiconductor manufacturing right at the most complex level, regardless of the enormous financial and technical challenges! Unlike common garden-variety circuit design done using standard software packages with canned device models, semiconductor device development and processing require expertise in the hard sciences like Physics and Materials Science. Then there are the challenges of maintaining high efficiency in a complex manufacturing operation, where unlike IT, mistakes cannot be detected or fixed at a key-stroke!
To leapfrog the technology hurdles the IT and Finance Ministries of India have been considering subsidies to attract established semiconductor companies to set up Fabs - however stipulating that the technology transferred be of the latest generation. The annual production of a state of the art Fab—with very narrow process recipes in the interest of efficiency - and therefore narrow product types as well - has to be at least $2.5 billion per year. So long as the Indian demand for a specific high end product (e.g. Memory) does not exceed at least 60 percent of this threshold, there may not be any takers among the handful of IDMs that really matter - regardless of the subsidy. It is hoped that a stick and carrot approach is taken by the Ministries to boost semiconductor consumption in India, as otherwise it may not cross the threshold for at least 3 years.
Why MNCs are not exactly beating down the door to set up Fabs in India?
Semiconductor consumption in India is still too small (per In - Stat only about 1 percent of the world production) to tempt established Semiconductor houses to part with their proprietary technologies, and is expected to remain so for the next 3 to 4 years even if it grows at a CAGR of 30 percent. Current semiconductor usage in India is some $1.5 billion per year – distributed over a wide variety of applications (analog power devices, amplifiers, ICs, digital logic, processors, memory etc) that require a range of incompatible Fab technologies (CMOS, BiCMOS etc). From a strictly business standpoint no single state of the art Fab (95 nm, 300 mm, min. 300 k wafers per yr) can be operated economically at these levels.
By any standard the Mobile phone and PC sectors in India are already quite big and growing at a healthy rate. Approximately 25 million handsets and 5 million PCs were sold in India last year. A CAGR in excess of 30 percent can be expected for these systems. Since 2005, Nokia, among others, is assembling Mobile Phone handsets in – country, at 20 million units per year.
Yet the MNCs are still not locating their supply chains within India. The reason behind why is — the China Factor!
All components for the handsets assembled in India including semiconductors are still being sourced from large outside suppliers, especially those in China. As a precondition for doing business in China and avoid the infamous 17 percent Chinese tariff, the MNCs are committed to transfer technology and buy certain amount of components from China - hence their reluctance to locate Fabs or purchase components within India.
The role that the Government of India should play:
The IT ministry of the Union Government of India has been rather energetic in directing MNCs like Nokia to locate their new assembly plants in India. One wishes that they would continue this pro-active approach and help the commercial semiconductor manufacturing industry get off the ground. To do this in a professional and competent manner the Ministry needs to first study what China did in this regard (e.g. as far back as 1992 arm-twisting a leading MNC engaged in both communications and semiconductor to set up the first Fab there) and then negotiate with the MNCs assembling systems in - country for a minimum level of sourcing from India.
Assembly & Test as a popular starting point to get into Wafer Fabs:
The area of Advanced Assembly will be an easier baby-step for the Indian semiconductor industry to start with. Taiwan had adopted a similar model to get into the game (for a discussion on strategies followed by various developing countries to launch their semiconductor manufacturing industries refer to the Sept `06 article).
It is gratifying to note that some of the groups who had been trying for some time to promote state of the art Wafer Fabs in India have recently down-shifted their focus to Assembly & Test (A&T). However even for A&T, access to production qualified process technology is still an issue, especially if high – margin niche technologies are desired. Though Government supported semi – academic Labs in Europe (e.g. IMEC) can provide some generic technology modules, they have no production experience or expertise in yield management, critical for end of line operations like A&T when each almost completed wafer can easily cost over $5000. The established outsourced A&T houses in the Far East on the other hand can hardly be expected to license technologies to potential future competitors in India and would rather keep the know-how and business to themselves.
Great opportunities for Start–Ups in Semiconductors – provided the right niches are chosen:
Promoters of semiconductor manufacturing in India, be it in the public or private sector, seem to be oblivious of the lucrative business opportunities in the Advanced end of Assembly operations. Advanced assembly is key to Portable products from Phones to iPods etc. Compared to Wafer Fabs, Advanced Assembly provides a much lower – cost route to integration of functions and mixed technologies, thus opening up greater revenue opportunities for Design houses in India. Unlike the earlier generations of packaging technologies (e.g. chip on board) the current generation (viz. flip – chip and wafer level packaging, much of which were developed by the author while at Motorola and Intel) have not yet been commoditized. Therefore profit margins of 20 percent are possible for technology leaders and the high end of the business is projected to keep growing at 30 percent for the next 5 years. For the high – end of Advanced Assembly technologies, as it relates to various chip-like modules, the Chinese are not yet up to speed and a great window of opportunity still exists here for serious investors in India. But quick decision and adequate support by the Indian Government is needed before the Chinese catch on.
A top of the line Fab for Advanced Assembly & Testing of Chips and Modules using dense interconnect technologies would cost approx. $150 to 220 million and take about 2 years to qualify for volume production. However, with many of the specific tools and all the technologies (wafer bumping, robotic assembly ) supplied by the authors’ company the investment required could be reduced to approx. $110 million and a ROI of 25 percent could be reached within 1.5 years.
Production of organic substrates (another technology originally developed by the author while at Intel) used to assemble micro – processor packages and very high-end mobile phones like the new Apple iPhone can be started with an investment of another $100 million. These facilities can be as good as the captive Assembly & Test facilities of the major IDM that recently located its A&T facility in Vietnam instead of India.
With competent guidance the Indian A&T facility can best the capabilities of those in Taiwan or Singapore and thus attract business from the Fabless houses in the US and even the Foundries of Taiwan. It can also broaden its portfolio beyond Assembly by exploiting synergies with low – cost Design / Driver Software capabilities in India to integrate stock chips into novel OEM modules for RF, Video etc. for export.
So “lets get serious”
It is hoped that sensible investors interested in starting profitable semiconductor manufacturing in India will not remain fixated with only the most visible high – end digital processors but broaden their horizon regarding semiconductors and associated components. Rather than wait indefinitely for the proverbial “knight on a white horse” they must themselves quickly go after the low-hanging fruits e.g. Advanced Assembly.
The Government of India should formulate a policy to (a) induce MNCs already assembling systems within the country to source more from within India (along the lines of what those MNCs have been doing for China all along), as well as (b) actively fund semiconductor ventures e.g. Advanced Assembly that are most likely to succeed within two to three years and (c) by means of well—planned and adequately financed R&D position India to intercept future growth areas in semiconductors, especially those driven by killer-apps specific to India such as distributed power generation ( solar, power management ), delivery of enabling services (literacy, healthcare, governance, self - help, e-commerce) to rural areas by cellular based internet. These diverse applications (high efficiency photo-voltaics, solid state batteries, LEDs for efficient lighting as well as low – noise RF amplifiers for Cellular nets) use Compound Semiconductors. Responsible Ministries should pool their resources to launch an integrated and aggressive R&D / Manufacturing program for these Semiconductors to place India in a position of commercial leadership in these new industries!
After chasing multi – billion dollar Wafer Fabs for 2 years the time to pick the right goals and go after them is NOW!!
The author is Chief Technical Officer, APSTL LLC, of Scottsdale, AZ, USA