India’s best near-term opportunity in semiconductors is not the fabrication plant. It is the stages around it: chip design, assembly, test and packaging, and the materials and equipment that feed them. A fab matters, and the country should build one. But treating it as the first test of seriousness sets up a long wait and a high price of failure. Capability that compounds sooner will also make the fab easier when it comes.
A value chain with very uneven difficulty
The chain has distinct parts. Design is the work of specifying a chip, done with software tools by engineers, and it needs talent and tool licences more than factories. Fabrication turns wafers into working circuits in clean rooms, using some of the most expensive machinery ever built. Assembly and test, usually called OSAT (outsourced semiconductor assembly and test), cuts the wafer into dies, packages them and checks that they work. Advanced packaging increasingly decides performance, because chips are now combined inside a single package. Underneath sit materials, such as wafers, specialty gases and chemicals, and equipment, from lithography tools to testers, supplied by a small number of firms worldwide.
India already has a large presence at one end. A large share of the world’s chip design engineering is done by engineers in India, mostly in the design centres of multinational firms. What it lacks is the middle and the base: fabrication and assembly at scale, and most of the supply chain behind them.
Why the fab is the slowest piece
A fab is a bet on three things at once: capital in very large amounts, a process technology that must be licensed or developed, and a dense local supply of engineers, water, power, gases and chemicals. Mature-node fabs are less demanding than leading-edge ones, but they still need all three, and they compete on cost with plants that were paid for years ago. Yield, the share of chips on a wafer that work, takes time to learn, and it is guarded by whoever has learned it.
Utilities deserve more attention than they get. A fab needs very large volumes of ultrapure water and power with almost no interruption, since a brief outage can scrap wafers in process. That is a site and state-government question as much as a company one. Industrial parks that can promise water, power and effluent handling at fab grade matter as much as the size of the subsidy.
A semiconductor industry is built in the order its dependencies allow, not the order that makes the best headline.
Where capability can compound sooner
Assembly, test and packaging are the natural first step. They need less capital than a fab, they can sit near existing electronics manufacturing, and they teach a supplier base to meet the quality standards of the sector. The India Semiconductor Mission, set up under the Digital India Corporation, is designed around this idea. It offers separate schemes for semiconductor fabs, display fabs, compound semiconductors and assembly and test, so that different parts of the chain can start at different speeds. Projects in assembly and test have been approved under it. Whether they reach volume production, and whether suppliers gather around them, is the thing to watch.
Design has its own instrument. The Design Linked Incentive scheme supports start-ups and small firms building chips and related intellectual property. This matters because design is where India’s existing talent can turn into ownership. Engineers who now work on a multinational’s product could own products instead. We would rather see many Indian firms with modest chips in production than a single celebrated design that never ships.
Materials and equipment are the least visible gap and probably the most strategic. An assembly plant that imports every gas, chemical, substrate and spare part is exposed to the same shocks as one that imports its chips. Local production of the less advanced inputs is a manufacturing opportunity in its own right, and it builds the industrial base that a fab will eventually need.
Talent, and the order of operations
Talent splits in two. Design engineers exist in numbers. Process engineers, who run fabs and packaging lines, are scarce, and most were trained abroad or inside a few firms. Training programmes such as the government’s Chips to Startup programme, and closer links with universities, help. But the fastest teacher is a working plant. Assembly and test lines produce process engineers as a by-product, and a fab can later hire them.
Our position is to sequence the work. Build assembly, test and packaging at volume. Back design firms until they own products. Localise the inputs. Let the case for a fab strengthen as the demand and the workforce around it grow, rather than asking one plant to carry the whole strategy.
What we are watching is the supply base. The day a gas, chemical or substrate plant opens because a nearby assembly line gave it a first customer is the day this approach is working. Until then, the open question is whether state and central agencies can deliver fab-grade water and power to more than one or two sites.