Setting up a semiconductor fab is easier said than done. There are a lot of aspects to look into. This includes, sourcing lithography tools from global manufacturers, having suitable infrastructure such as a vast area of land, copious amounts of water, water recycling system, electricity lines, and labour power, among other things.
Analytics India Magazine spoke to Arun Mampazhy, a veteran semiconductor analyst, to gauge the cost estimate of setting up a fabrication plant. At the outset, Mampazhy addressed the government scheme surrounding semiconductor operations in India saying, “The bigger picture, if you look at it [the policy], covers a variety of spectrum”.
To begin with, a compound semiconductor or silicon photonics fabrication, for instance, is a smaller investment. A compound semiconductor fabrication will typically consist of three phases—growing wafers, making chips, and packaging—and would total up to $40 million. RuttonSha International Rectifier Ltd, a big player in the power semiconductor industry, is one of the applicants for a compound semiconductor plant.
However, more than 80 percent of the world’s semiconductors are still made of silicon substrate. Hence, silicon fabrication becomes a crucial area to discuss, with regard to the government’s semiconductor mission. The cost of setting up a silicon fabrication plant is hugely contingent on the technology node the buyers will aim for. This is because you need lithography equipment for manufacturing semiconductor chips, and the kind of tool used decides the range of technology nodes that you can work with.
Currently, there are three applications in the race for silicon fabrication. They are targeting the following technology node:
(1) IGSS Ventures, which is looking to set up a fabrication unit in Tamil Nadu, is aiming at the range of 28nm, 45nm, and 65nm technology.
(2) ISMC has proposed a fabrication unit designing 65nm technology.
(3) Lastly, the Foxconn-Vedanta joint venture will be looking at a 12-inch (300mm) wafer carrying 28nm technology.
It is important to set the background straight with what the applicants are looking for to determine the cost of the lithography tool that will be needed for the process. Mampazhy said, “The industry typically uses a 200mm wafer for more than 90nm process, whereas less than 90nm process generally requires a 300mm wafer.” Considering the requirements of the buyers, we are looking at a process technology etched on a 300mm wafer size.
But, within the 300mm wafer size as well, there is a cost bifurcation. Up to 65nm or 55nm technology can be made with 193 Argon Flouride (ArF) Laser, which will cost about $40 million a piece. And if we go further below to the 45nm, 28nm all the way up to 16nm range, a 93-nanometer immersion will be required which will cost about $100 million a piece.
Mampazhy added that the current policy requires the manufacturers to build a 40,000 per month wafer capacity. To get that kind of production running, he said, you need about 15 to 20 of these lithography tools. This means that the lithography tool you use impacts the overall cost by a huge difference. According to Mampazhy, 65nm to 55nm technology is currently the sweet spot, since you could fully utilise the resources with the best technology available. Hence, 15-20 of 193 ArF lasers would take the cost tally up to $2 billion.
One of the ways the cost could be controlled is by using refurbished equipment. But, according to Mampazhy, “The government of India is not supportive of this, because they’re afraid that India will become a dumping ground of old equipment.” Besides, there could also be instances where the applicant might quote an overpriced number for the refurbished tool, and get the subsidy for the same.
The $2-billion figure is however just the equipment cost. Adding to that, the infrastructural cost of setting up a fab will be about $200 million to $300 million. Plus, there will also be a technology transfer fee levied by the companies issuing the lithographic tools to the fabrication plants. The 65nm-55nm range will have a licensing fee of about $300million to $400million, and as we go to the narrower range, the transfer cost increases to about $1 billion for a 28nm range and more if we go further down.
Together, the three components, Mampazhy said, will add up to $3 billion for the 65nm range. Likewise, we can estimate the total cost at the lower end of the spectrum.
Govt subsidiary to the rescue
In September 2022, the Ministry of Electronics & Information Technology (MeitY), modified the existing semiconductor policy. The new programme had additional incentives to attract investments from companies/consortia to create semiconductor and display fabrication ecosystem in India. These incentives include an outlay of ₹76,000 crores ($10 billion), where the government will provide the eligible applicants 50% of financial support of the project cost on a pari-passu basis.
The incentive package will also apply to setting up compound semiconductors/silicon photonics/sensors (including MEMS) fabs/discrete semiconductor fabs and semiconductor ATMP/OSAT units. The government intends to establish at least 20 such units under this scheme.
The announcement of the scheme has since sparked interest of quite a few buyers, who wish to avail the scheme. Of late, Reliance and HCL bid to acquire 26-51% stake in ISMC Analog after their proposal to establish a semiconductor wafer fabrication facility in Mysore.
Beyond the central government subsidy, there are also several state-sponsored schemes pushing India’s semiconductor mission ahead. Gujarat government, for example, proposed additional assistance on the total capital expenditure of selected proposals. The incentives will leave the investors with much lesser investment than they initially had to put in, when kept in proportion to the stake they will have in the production.
What next?
India has not been able to attract much foreign interest until now—the reason being it is yet to prove itself as a fertile ground for semiconductor manufacturing. But, once the machine gets running, we can expect India to host semiconductor plants for a wide range of technology nodes catering to a large scale of industries and consumer electronics.
Therefore, the modified scheme deserves attention particularly because the previous edition of the scheme had an incremental approach to allocating cost for different ranges depending on the technology node. The 28nm or lower range was eligible for 50 percent of the project cost, 28nm to 45nm for up to 40 percent, and above 45nm up to 65nm were eligible for up to 30 percent of the project cost. Thus, the new model has levelled the playfield for everyone.
Speaking about the road ahead for India in its semiconductor ambitions, Mampazhy stressed that India can have lofty ambitions for the future but at present taking the first step is important. He added, “Be it 65nm or 28nm, for the next 2 to 3 years you build a path…that market will be there for these chips for next at least 10 to 15 years and by that time you can decide what is the next way whether you want to go for advanced nodes that have silicon itself or perhaps in a different direction.”
