Published: October 9, 2026 | Category: Technology | By Mahesh
The India Semiconductor Mission has reached the stage where its numbers sound like victory. At SEMICON India in September 2026, the government said five chip plants were in commercial production, up from none in 2021.[2][3] That progress is real and it deserves credit. It also needs a careful reading, because every one of those five plants sits at the back end of chipmaking. The country does not yet operate a single wafer fab.[3]
So what does the count measure, and what does it skip? This piece separates the three things a headline can mean by “made in India”: finishing chips, fabricating them and designing them. The distinction matters to founders choosing where to build, engineers choosing where to work and investors deciding which announcement deserves a premium. It also connects to our earlier reporting on Volantis and its light-based AI chips, an Indian design story that sits at the opposite end of the chain from a packaging line.
What “Five Plants in Production” Counts
Start with the arithmetic of the announcement itself. When the Union Cabinet approved the second phase of the mission on 15 July 2026, the official note said twelve manufacturing units had been approved and three of them, Micron, Kaynes Semicon and CG Semi, had begun commercial production, with one more expected in 2026.[1] By September the count read five. Two further lines were inaugurated by video at the SEMICON India event. CDIL Semiconductor in Mohali makes discrete devices. Suchi Semicon in Surat handles packaging.[3] The jump from three to five is therefore a matter of new lines opening, not of a changed definition.
Now the part that rarely makes the headline. By the government’s own account, all five are back-end units, meaning assembly, test, marking and packaging lines rather than wafer fabs.[3] A back-end plant receives a finished silicon wafer made somewhere else. It cuts the wafer into individual dies, mounts each die in a package, wires it up, seals it and tests it. The chip leaves the plant ready to be soldered onto a board, but the expensive and technically hardest step, printing billions of transistors onto the wafer, happened in another country.
None of this makes the plants fake or the milestone small. Going from no commercial chip plant in 2021 to five in 2026 is a fast change for any country.[2] The point is narrower. When a politician says India now makes chips, an engineer hears “fabricates” while the factory floor in Sanand means “finishes”. Both statements can be true on the same day. The gap between them is where the next five years of the programme will be decided.
The Arithmetic of the Value Chain
Why does the distinction carry so much weight? Because the stages of chipmaking do not share the profits equally. A widely used analysis by the Semiconductor Industry Association and Boston Consulting Group attributed roughly half of the industry’s value added in 2019 to chip design, about a quarter to wafer fabrication and only around 6 percent to assembly, packaging and testing.[5] The OECD, drawing on the same work, adds that the back end absorbs 10 to 15 percent of the industry’s physical capital spending.[6] In other words, packaging is cheap to build and modest in what it earns.
Why start there anyway? Several reasons hold up. A back-end plant can be built far faster than a fab. It can bring in an anchor customer, employ thousands of technicians and test engineers and teach a supplier base how to meet semiconductor quality standards. It also fits a hole in the world map. A BCG and SIA study of the supply chain noted that mainland China and Taiwan are expected to keep the largest share of back-end capacity, which leaves global buyers looking for a credible second location.[7] India can sell exactly that.
The counter-argument is the arithmetic above. A country that only packages chips imported as wafers keeps a thin slice of each chip’s price. It stays exposed to whoever controls the wafer supply. Packaging is also getting more sophisticated. Our report on the advanced packaging bottleneck in AI chips shows that the best back-end work can be strategic rather than routine. India’s current lines serve memory, power and automotive demand, not the cutting edge of that story.
Micron and the Back-End Bet
The anchor of the whole effort is Micron’s plant at Sanand in Gujarat. Micron’s chief executive, Sanjay Mehrotra, told the SEMICON India audience that commercial DRAM and NAND production was underway there.[3] The American memory maker is a sensible first customer for a country testing its back-end capabilities, because memory packaging is high-volume and unforgiving of defects. If the Sanand line holds yield and schedule, it becomes the reference every later entrant is judged against.
Read the sentence carefully, though. Production at Sanand means assembling and testing memory chips, not fabricating the wafers they come from. Those wafers are made in Micron’s fabs elsewhere and travel to Gujarat. The benefit to India is real but specific: jobs, exports of packaged goods, a trained workforce and a local supplier network. The silicon IP and the fab know-how stay abroad.
That is not a flaw so much as a stage. Several Asian economies built their electronics industries by doing back-end work for foreign firms first and climbing the chain later. The risk is complacency. A plant that works well can create the comfortable impression that the hard part is finished. Kaynes and CG Semi, both Indian-led ventures, will test a different question. Can they run these lines profitably once the inauguration cameras leave?
Dholera Is the Project That Changes the Category
If the five back-end plants are the opening move, the Tata Electronics fab at Dholera is the one that changes what India can claim. Tata is building the country’s first commercial 300 mm wafer fab there with Taiwan’s PSMC as technology partner.[8][9] In May 2026 Tata Electronics and the Dutch lithography supplier ASML announced a partnership under which ASML will supply lithography tools and support the fab’s ramp-up, with joint work on local talent, supply chain resilience and research infrastructure.[8] Lithography is the step that prints circuit patterns on the wafer, so a signed supplier is a prerequisite for any serious production date.
At the SEMICON India event, Tata also signed supply memoranda with materials and gas firms including Fujifilm, JSR, Sumitomo Chemical and INOX Air Products. Another covers a 363-acre vendor park near the fab.[3] Tata put the combined investment in Dholera and its assembly site in Assam at $14 billion and said construction at both was on track.[3] The announced design is a fab of about 50,000 wafer starts a month aimed at mature process nodes. Those suit automotive, power and industrial chips better than smartphone processors or AI accelerators.
One caution deserves space. Public timelines for the first wafer disagree with each other. Some trackers cite trial production in late 2026 while others point to commercial output around 2028. The September coverage still describes the fab as under construction.[3] A date that moves depending on who is quoting it is a reason to treat any first-silicon claim as unconfirmed until Tata states it. Fabs are famously difficult to bring to yield. The stretch between the first wafer and a profitable one is where projects usually stumble.
The mature-node focus is also a strategic choice rather than a shortfall. Cars, appliances, power electronics and industrial controllers consume enormous volumes of chips that do not need the newest transistors. Shortages of exactly those parts have halted car lines before. As our coverage of chips for robots and physical AI suggests, demand outside the data centre is broadening. A reliable 28 nanometre supply source in India serves that demand without competing with Taiwan for the frontier.
Semicon 2.0 Pays for the Parts Nobody Photographs
The second phase of the mission is where policy admits that plants alone are not enough. The Union Cabinet approved Semicon 2.0 on 15 July 2026 with an outlay of Rs 1,27,500 crore, about US$13.25 billion, against the Rs 76,000 crore approved for the first phase in December 2021.[1][2][4] It organises the money around six pillars. They cover chip design, equipment and materials, fabrication, assembly and packaging and research and development, with talent development as the sixth.[2][4] At SEMICON India the Prime Minister described the second phase as a larger commitment than the first, at roughly $13.5 billion against about $8 billion.[3]
The equipment and materials pillar deserves more attention than it gets. Chip plants consume ultra-pure gases, photoresists, specialty chemicals, substrates and thousands of precision parts, and most of those inputs are supplied by a handful of foreign firms. A plant in Gujarat that imports every input is a very expensive assembly point. Tata’s supplier memoranda and its planned vendor park are early attempts to pull some of that supply chain onshore. The second phase also offers incentives to companies in equipment, materials, chemicals and gases.
The design pillar matters for a different reason. India has long hosted large chip design teams inside multinational design centres, yet much of the intellectual property they create belongs to foreign parents. Funding that favours Indian design startups, such as the memory-focused photonics company we profiled in our Volantis funding report, is how a country converts its engineering talent into owned products. Design is also the cheapest part of the chain to start, so it makes sense for the state to push it alongside the expensive fab.
Money at this scale also follows demand. The AI boom is pulling capital into chips on a scale that dwarfs national programmes, as the $40 billion arrangement in our report on SpaceX’s Nvidia chip financing shows. India’s incentives compete for factory investment against that gravity. A budget of roughly US$13 billion is meaningful but not overwhelming.
Risks the Headline Count Hides
The first risk is dependence. Because all five operating plants finish wafers made elsewhere, their output rises and falls with foreign wafer supply and with trade rules that India does not control. A disruption in Taiwan or a licensing dispute in the United States could slow Sanand within weeks. The Dholera fab reduces that exposure only for the mature-node parts it will make. It will itself depend on imported lithography tools and chemicals.
The second is credibility of timelines. Programmes of this kind live on announcements, and announcements run ahead of fabs. Approved investment and projected capacity are not the same thing as realised output. The mixed first-wafer dates for Dholera are a small example. A reader who tracks only approvals will overestimate progress by a wide margin.
The third is utilisation. A packaging plant earns money only when it is full. Micron brings its own demand, but newer entrants have to win customers who already have established suppliers in Malaysia, Taiwan and China. Price, reliability and delivery speed will decide whether the lines fill up. Governments can subsidise construction, yet no scheme can compel a procurement manager at a global electronics firm to switch suppliers.
The fourth is people. A fab needs process engineers, equipment technicians and yield specialists in numbers the country has not produced before. The ASML partnership includes training commitments for exactly this reason.[8] Talent takes years to build. A fab that opens before its workforce is ready can lose its first year to avoidable mistakes. The talent pillar of Semicon 2.0 recognises that.
What This Means for You
If you are an engineer. Back-end plants create test, reliability and packaging jobs now across Gujarat, Punjab and Assam. They teach skills that transfer to fab work later. Process and equipment roles will grow around Dholera as it approaches production. Design roles remain the highest-paid and the most portable, so keep an eye on which Indian design houses win funding.
If you are a founder building hardware. Domestic packaging and test capacity could shorten supply lines for power electronics, automotive and industrial parts. Ask suppliers where the die is fabricated and where the package is finished, because those two answers are no longer the same country. For design startups, Semicon 2.0 is the funding window to study before the next cycle.
If you are an investor. Judge announcements by utilisation and yield, not by approvals. A plant that has been “inaugurated” tells you little, while quarterly volume, customer wins and exports tell you much more. Equipment, materials and gas suppliers could prove steadier beneficiaries than the plant operators themselves, but that depends on how quickly the second phase actually awards incentives.
If you buy electronics in India. Do not expect Indian-made chips in your phone next year. Locally packaged memory and power parts will reach domestic manufacturers first. Prices will follow global supply conditions rather than fall because of a new factory.
Four events will show whether the count turns into capability. Watch for the first wafer from Dholera, a sixth plant entering production, the first awards under the equipment and materials pillar of Semicon 2.0 and sustained monthly volumes at Sanand. Until those arrive, India’s achievement is genuine and incomplete. That is a respectable place to stand four years into a decade-long build.
Common Questions
How many semiconductor plants are operating in India in 2026?
The government says five of the 12 approved manufacturing units were in commercial production by September 2026. They include Micron, Kaynes Semicon and CG Semi, plus new lines at CDIL Semiconductor in Mohali and Suchi Semicon in Surat. All five are back-end units that assemble, test and package chips or make discrete devices. None is a wafer fab.
What is the difference between a chip fab and an ATMP or OSAT plant?
A fab, or wafer fabrication plant, prints circuits onto silicon wafers and is the front end of chipmaking. An ATMP or OSAT plant sits at the back end, where finished wafers are cut into chips, mounted, wired, sealed and tested. Back-end plants cost less and start faster but capture a much smaller share of the value in a chip.
When will India's first semiconductor fab start production?
Tata Electronics is building India's first commercial 300 mm fab at Dholera in Gujarat with Taiwan's PSMC. It was still under construction in September 2026. Public timelines conflict, with some trackers citing trial production in late 2026 and others pointing to commercial output around 2028. Treat any first-wafer date as unconfirmed until the company announces it.
What is Semicon 2.0 and how much money does it carry?
Semicon 2.0 is the second phase of the India Semiconductor Mission, approved by the Union Cabinet on 15 July 2026 with an outlay of Rs 1,27,500 crore, roughly US$13.25 billion. It is built on six pillars covering chip design, equipment and materials, fabs, assembly and packaging, research and development plus talent development.
Does Micron make chips in India?
Micron runs a commercial assembly, test and packaging plant at Sanand in Gujarat, where its chief executive said DRAM and NAND production is underway. The plant packages and tests memory chips. It does not fabricate the silicon wafers, which come from Micron fabs outside India.
Why does India start with packaging instead of a fab?
Back-end plants need a fraction of the capital that a fab demands, with assembly, test and packaging taking an estimated 10 to 15 percent of industry capital spending. They can also be built and staffed faster, so a country can bring in customers and skills early. The trade-off is that the stage adds only around 6 percent of the industry's value.
This analysis is editorial commentary based on the public sources listed below. It is not investment advice. Project timelines and official counts change often, so confirm current figures with the Press Information Bureau and company announcements before acting.
Sources
- Press Information Bureau, "Cabinet approves Semicon 2.0," July 15, 2026. Link
- Press Information Bureau, "India Building Semiconductor Future," backgrounder, September 2026. Link
- EE Times, "SEMICON INDIA 2026: India Starts Packaging Chips as Ecosystem Takes Shape," September 2026 (trade press report of the event). Link
- India Brand Equity Foundation, "Cabinet approves India Semiconductor Mission 2.0 with outlay of US$ 13.25 billion," July 16, 2026. Link
- Semiconductor Industry Association and Boston Consulting Group, "Strengthening the Global Semiconductor Value Chain in an Uncertain Era," April 2021. Link
- OECD, "Vulnerabilities in the semiconductor supply chain," Science, Technology and Industry Working Papers, 2023. Link
- Boston Consulting Group and Semiconductor Industry Association, "Emerging Resilience in the Semiconductor Supply Chain," May 2024. Link
- ASML, "Tata Electronics and ASML announce strategic partnership to advance the semiconductor manufacturing ecosystem in India," May 2026. Link
- Tata Group, "Tata Electronics and ASML Announce Strategic Partnership," press release, May 17, 2026. Link
Read More on Depth Grid
- Volantis $88M Funding: Light-Based AI Chips Explained
- The AI Chip Market Everyone Forgot: Robots That Don’t Talk
- SpaceX’s $40B Nvidia Chip Financing: What We Know
- San Francisco Hits Pause on New Data Centers
Article by Mahesh | Depth Grid
Discussion (0)
Comments are disabled for this post.