How many people are fixated on 2nm and 3nm, watching the process race between TSMC and Samsung?
Yet India is pouring billions of dollars into "legacy chips" from 28nm to 180nm, into packaging, testing, and assembly, into specialty semiconductors and design talent. Is India slow? Is India backing down? Is India picking up business others don't want?
In fact, it may not necessarily want to become the next TSMC, because a more realistic goal is to become a backup node, a certification node, and a delivery node in the global chip supply chain.
This is probably why the Modi government repeatedly emphasizes "trusted," and probably also the reason behind the $13.5 billion subsidy.
Making "legacy chips," what is India after?
Mature process nodes may not sound sexy, but they support nearly 70% of global chip output.
Chips from 28nm to 180nm are widely used in automobiles, industrial machinery, telecommunications equipment, home appliances, power grids, power management, and network components. No matter how advanced AI data centers are, they still rely on power management chips, network chips, and a large number of analog devices to operate. Advanced processes determine the upper limit of computing power, while mature processes determine whether systems can be delivered stably.
The two have different divisions of labor in the industrial chain, and different cash flow characteristics.
India's choice, on the surface, is to avoid head-on competition in cutting-edge processes, but in reality, it still fits economic logic very well.
In India, building a cutting-edge 3nm manufacturing plant could cost more than $15 billion, and 2nm production is "too costly." Krishnan, Secretary of India's Ministry of Electronics and Information Technology, has also publicly acknowledged this.
For an economy with limited per capita income and a manufacturing base still catching up, chasing 2nm right from the start means heavy fiscal pressure, a large talent gap, constrained access to equipment and materials, and long customer certification cycles.
Pouring money into the most cutting-edge processes can easily become an expensive vanity project.
The economics of mature processes and back-end steps are completely different.
Back-end packaging and testing have a lower capital threshold, stronger customer certification stickiness, quality systems that can be accumulated step by step, and talent that can be trained on the job. India's first semiconductor plan approved 12 manufacturing projects with planned investment exceeding $17 billion, including one silicon wafer fabrication plant, two specialty semiconductor facilities, and nine assembly, testing, and packaging units.
Five of the nine units have already begun commercial production, including Micron and Indian companies such as Kaynes Semicon, CG Semi, CDIL Semiconductor, and Suchi Semicon.
These projects mainly focus on assembly, packaging, and testing, and have not yet directly manufactured integrated circuits on silicon wafers.
The next step is to build front-end wafer manufacturing capabilities.
This path is quite interesting.
The traditional semiconductor development path usually goes from front-end to back-end: first build wafer fabs, then support them with packaging and testing.
India is doing the opposite: starting with back-end, using the customer stickiness, quality certification, and delivery records of the back-end to drive front-end manufacturing in reverse.
Back-end steps may not look high-margin, but they can accumulate three things: customer trust, quality systems, and capacity management experience.
When a customer hands you chips for packaging and testing, it means they are willing to entrust a segment of their supply chain to you; once certification is passed, the cost of switching suppliers is not low. India uses back-end as the entry point, mature processes as the springboard, and the domestic demand market as the order anchor. This approach is more in line with business logic than "directly chasing 2nm."
Furthermore, one of the products India sells is called "trusted."
At SEMICON India 2026, Modi said, "The world needs new and trusted chip manufacturing locations, and India is preparing for this."
Global supply chains used to pursue Just-in-Time, emphasizing zero inventory, low cost, and fast delivery; now more and more countries and large companies are turning to Just-in-Case, emphasizing backup, redundancy, and security.
Redundancy is insurance, and insurance requires premiums.
What India wants to collect is precisely this "supply chain insurance" premium.
Whether the premium can be collected depends on whether India can turn "trusted" into an institutional product. Intellectual property protection, export control compliance, supply chain transparency, delivery reliability, and geopolitical neutrality may sound abstract, but in the semiconductor industry they are all hard constraints.
An automotive-grade chip must go through lengthy certification, a wafer fab must undergo repeated customer audits, and a single quality incident can lose years of orders. The "trusted" that India wants to sell needs long-term records to support it.
This is also the most easily underestimated part of India's semiconductor strategy: it is using national credit to endorse the industry and using institutional arrangements to reduce the risk premium for foreign investment.
India also has a ready card in hand, and that is design talent.
India accounts for nearly one-fifth of the world's chip design talent. Companies such as Intel, Qualcomm, and Texas Instruments have major R&D centers in India. Indian engineers have participated in the development of a large number of complex chips worldwide, while actual manufacturing is mostly completed overseas.
Imports meet 90% to 95% of India's semiconductor demand. From fiscal year 2017 to fiscal year 2025, India spent nearly $150 billion on semiconductor product imports. This figure is both pressure and a demand anchor.
Domestic demand from automobiles, telecommunications, home appliances, power grids, and data centers can provide initial orders for India's local manufacturing and reduce the risk of capacity ramp-up after projects go into production.
Therefore, what is worth paying attention to in India's making of "legacy chips" should be the demand structure and institutional products.
Mature processes have huge and stable demand, back-end steps have customer stickiness, the domestic demand market has an order anchor, and national credit can be converted into supply chain insurance.
With this combination, India turns semiconductors from a pure technology race into a business about supply chain certainty.
What it pursues is cash flow and industrial capability.
As for where the money comes from, how it is spent, and whether it is worth spending, that depends on the situation in the Indian government's subsidy accounts.
How should the $13.5 billion subsidy be spent without losing money?
India's "Semiconductor 2.0" plan has received $13.5 billion in funding support, and the projects approved under the first plan involve investment exceeding $17 billion.
Looking at the global semiconductor subsidy race, $13.5 billion is not an astronomical figure.
The subsidy scales of the United States, the European Union, and Japan are larger, and the support received by TSMC, Samsung, and Intel is more concentrated. India does not have the conditions to bloom everywhere and can only adopt a "nail strategy," forming irreplaceability in a few niche areas and using limited fiscal funds to leverage more private capital.
Financially speaking, government subsidies are like option premiums, and projects are like underlying assets.
The role of subsidies is to reduce the risk premium for private capital, making consortiums, foreign investors, and equipment makers willing to bet together. Tata Group and Taiwan's Powerchip are building a $11 billion wafer fab in Dholera, Gujarat, with plans to start commercial production in mid-2028, initially using 55nm and 90nm technology, then transitioning to 28nm.
Applied Materials has announced plans to invest $5 billion in India over the next decade, covering R&D, domestic supply chains, and talent development.
Lam Research plans to invest about $1.04 billion to build its first silicon component manufacturing plant in India and expand its R&D operations. These investments are motivated both by locking in India's future capital expenditure and by being close to customers and policy dividends.
Subsidies as premiums have one advantage: if successful, India gains industrial chain and supply chain status; if unsuccessful, the fiscal loss is relatively controllable.
But they also have one disadvantage: options have time value.
The subsidy window will pass, technology will iterate, and the capacity cycle will reverse. If projects are delayed, yields cannot ramp up, and customer certification is not obtained, the options will become sunk costs.
What India's semiconductor plan most needs to guard against now is the "announcement economy" and "MOU market."
Signing contracts, laying foundation stones, and issuing press releases are easy actions; turning factories into capacity, turning capacity into yield, and turning yield into customer orders is the real hard work.
It should be noted that the semiconductor capital expenditure cycle has a clear order of beneficiaries: equipment and materials orders usually lead wafer fab production by one to two years, back-end packaging and testing commercialization leads front-end manufacturing, and design services benefit from talent supply.
Most of India's current projects are in the stage from project groundbreaking to early commercialization. The market narrative easily jumps directly from "policy announcements" to "capacity explosion," yet in between lie several hurdles: yield ramp-up, customer qualification, and supply chain supporting infrastructure. Subsidies can shorten the first leg of the journey, but cannot shorten the legs that follow.
However, the Indian government is also filling in the ecosystem.
The scope of support under the Semiconductor 2.0 plan has expanded to equipment manufacturers, specialty chemicals, raw materials, chip design, R&D, and talent development. Raghav Srinivasan Balakrishnan, director of the Sudit Foundation, pointed out that India currently relies almost entirely on foreign sources for basic raw materials and machinery.
This is the real shortcoming of Indian semiconductors: back-end processes can run first, but front-end manufacturing has many lessons to catch up on. Cleanroom management, ultrapure water, specialty gases, photoresist, silicon wafers, and equipment components—each requires time and capital.
India can use subsidies to attract equipment makers, can use domestic demand orders to feed packaging and testing plants, and can use design talent to incubate fabless companies, but the learning curve of front-end manufacturing still has to be walked step by step.
NITI Aayog's roadmap proposes building a $120 billion to $150 billion semiconductor industry chain by 2035, and making India one of the three main global destinations for outsourced semiconductor packaging, testing, and advanced packaging.
This goal is fairly pragmatic.
Advanced packaging combines different chips into more complex packages, improving performance without fully relying on shrinking transistor dimensions. By finding a position in advanced packaging, India is in effect seeking an ecological niche in the value migration of "reverse Moore's Law."
The economics of Moore's Law are changing: the marginal cost of process scaling is rising, and value is migrating toward packaging, power, analog, and system integration.
India is not hard-chasing transistor density; it is seeking opportunities in system integration and packaging, a path more suited to its resource endowments.
But advanced packaging also requires equipment and materials, and India will still need to import in the short term.
So this is not a feel-good story of "overtaking on a curve," but more like "ecological niche arbitrage." First gain a firm foothold in segmented fields with huge demand where existing advantages can bring commercial advantage, then upgrade toward 28nm front-end and advanced packaging.
The success or failure of India's semiconductor strategy does not depend on how lively the subsidy announcements are, but on whether projects can start production on time, whether yields can meet standards, whether customer qualification can be passed, and whether capacity utilization can climb.
Subsidies are the premium, projects are the options, and the exercise date is around 2028. Before that day arrives, the market will repeatedly trade expectations and repeatedly test execution.
Who will capture India's chip dividend first?
In the game of Indian semiconductors, the time window is very clear.
2026 to 2027 is the back-end delivery period.
India already has five chip production units that have started commercial production, mainly focused on assembly, packaging, and testing. More OSAT and ATMP projects will land, and equipment and materials investment will begin to pay off.
The market narrative will shift from "policy announcements" to "orders and capacity utilization."
In this stage, what is most easily hyped is the "Indian semiconductor concept"; what most needs to be watched is the customer list, qualification progress, and capacity ramp-up data.
2028 is the front-end validation period.
Tata and Powerchip's wafer fab in Dholera plans commercial production in mid-2028, initially at 55nm and 90nm, later transitioning to 28nm.
This is the touchstone for India's front-end manufacturing capability.
If mass production goes smoothly, yields meet standards, and customer qualification is passed, the Indian semiconductor narrative will upgrade, and equipment and materials suppliers and design service providers will all benefit. If there are delays or yields fall short, market confidence will be hurt, and subsidy efficiency will also be re-examined.
2029 to 2031 is the differentiation period.
28nm, advanced packaging, power semiconductors, analog chips, and automotive-grade semiconductors will begin to diverge. If India locks in long-term automotive-grade and industrial orders, it can form differentiation; if it only does cheap legacy chip foundry, it will easily fall into price wars.
TrendForce reported in June 2026 that rising demand for power-related semiconductors, including AI infrastructure, is tightening manufacturing capacity and may sustain price increases until 2027.
This window is favorable for India. After new capacity is released in 2027 to 2028, mature process nodes may diverge.
India must use "trustworthiness" and "automotive-grade certification" to avoid pure cost competition.
2030 to 2035 is the big test of the roadmap.
India's semiconductor demand is expected to reach $87 billion in fiscal 2030 and $206 billion in fiscal 2035. The import substitution space is huge, but the increase in manufacturing share is a slow variable.
India may become an important global destination for OSAT and advanced packaging, while front-end advanced nodes still depend on Taiwan and South Korea.
It may not necessarily be a disruptor, but more likely a complementor and insurer.
As for the stock market and industry situation, it is obvious that equipment and materials suppliers will benefit first. Applied Materials and Lam Research investing in India is about locking in future orders, but India's revenue share is small and should not be exaggerated.
Back-end OSAT and EMS companies are commercializing earlier. Kaynes, Suchi, CDIL, and others have already started production lines. Valuations are high and execution risks are also high; the key is customer qualification and capacity utilization.
Power, analog, and automotive-grade semiconductors benefit from Indian automotive and industrial demand. Certification cycles are long, and once they enter the supply chain, stickiness is strong.
Design services benefit from talent supply, are asset-light with high gross margins, but economies of scale take time.
Wafer foundry is in the last position to deliver, with large capital expenditure and slow yield ramp-up; 2028 is the real checkpoint.
Indian semiconductor targets are scarce, and the theme may spill over into automotive electronics, industrial, and EMS. Investment needs to be wary of the "announcement economy" and "MOU rallies."
Signing and groundbreaking can bring sentiment, but only capacity and cash flow can bring valuation.
The Indian semiconductor theme will go through four stages: "policy announcement—project groundbreaking—commercialization—cash flow." Most projects are currently between the second and third stages, which most tests patience and due diligence.
For Chinese observers, India's semiconductor strategy has several points worth taking seriously.
Mature process nodes are the cash flow and industrial capability foundation, and should not be underestimated just because of the words "legacy chips." India relies on Chinese equipment, materials, and components in the short term, and in the long term will create diversion in OSAT, EMS, and mature process nodes.
China should consolidate its cost and capacity advantages in mature process nodes, promote domestic substitution of equipment and materials, develop advanced packaging, and at the same time can participate in Indian supply chains, while also being alert to the "trustworthy" brand replacing Chinese supply.
India uses national credit to endorse semiconductors; China likewise needs to use institutional certainty and industrial efficiency to stabilize its own position.
I have three judgments.
First, the essence of India's semiconductor strategy is to use limited subsidies to buy a geopolitical call option, and to use the cash flow from back-end packaging and testing and mature process nodes to gradually earn the learning curve of front-end manufacturing. It may not necessarily want to become the next TSMC; its more realistic goal is to become a backup node and insurer in the global supply chain.
Second, the key to India's success is not the amount of subsidies, but yield, talent, equipment and materials, customer qualification, and delivery reliability. The mass production of Tata's wafer fab in 2028 is the touchstone, and the global divergence of mature process capacity after 2027 is the external test.
Third, the real product of Indian semiconductors, besides chips, is the word "trustworthy."
It turns national credit into an institutional product, turns supply chain security into a business, and turns back-end processes into a springboard for front-end manufacturing.
This article comes from the WeChat public account "Dongzhen Shanglue," author: Dongzhen Shanglue
