Guide

Semiconductor Manufacturing in India

Published 25 August 2026. Facts verified to 25 August 2026 unless dated otherwise.

Twelve approved plants, one commercial wafer fab under construction, two units already shipping: the state of Indian semiconductor manufacturing, counted rather than celebrated.

What the phrase covers, and what it does not

Semiconductor manufacturing in India is asked as one question and answered as three different ones, which is why published counts of Indian chip plants run anywhere from one to a dozen depending on who is doing the counting. The three activities are separate industries with separate economics. Wafer fabrication builds transistors into silicon across hundreds of process steps in a cleanroom held to particle counts a surgical theatre would not attempt, and the capital bill runs to tens of thousands of crores. Assembly, test and packaging takes finished wafers, separates the dies, connects each one to a package and proves it works, at roughly an order of magnitude less capital and in a fraction of the construction time. Design draws the circuit and never touches a machine.

India now has real activity in all three. It also has exactly one commercial silicon wafer fab under construction and nine approved packaging plants, and any sentence that flattens those into a single count of chip factories has already given up the precision a reader came for. This record keeps the categories apart on every page, because the difference between them is the difference between a plant that can make a chip and a plant that can finish one.

The rest of this guide sets out the approved portfolio with its money and its dates, separates what is producing from what is building, and states plainly which layers of the chain India still does not have. Where a figure comes from a government release, the release and its date are named in the sentence.

The scoreboard as of 25 August 2026

Twelve manufacturing units stand approved under the India Semiconductor Mission with cumulative investment of over Rs 1.64 lakh crore. The Union Cabinet release of 15 July 2026, issued alongside the approval of Semicon 2.0, breaks that portfolio down as one silicon fab, one silicon carbide fab, one integrated gallium nitride micro-LED display fab and nine packaging units DURABLE. That is the government describing its own portfolio, and it is the cleanest official categorisation published to date.

The arithmetic underneath it holds. The annexure to the Ministry of Electronics and Information Technology reply in the Lok Sabha on 1 April 2026, given by Minister of State Jitin Prasada, lists ten approved projects with individual investment values that sum to about Rs 1,60,365 crore. The Cabinet then approved two further units in Gujarat on 5 May 2026 with a combined investment of around Rs 3,936 crore. Add them and the total reaches about Rs 1,64,301 crore, which is what the Cabinet's own rounded figure of about Rs 1.64 lakh crore describes. Two independent government documents agree to within rounding, and that is worth stating because a great deal of published commentary quotes totals no annexure supports.

Those twelve units sit in six states: Gujarat, Assam, Uttar Pradesh, Odisha, Punjab and Andhra Pradesh. Six of the twelve are in Gujarat, which is half the national portfolio in one state, and the reason Gujarat carries the densest semiconductor geography in the country. Investment is far more concentrated than unit count suggests: the single Dholera wafer fab accounts for about Rs 91,526 crore of the Rs 1.64 lakh crore, which is roughly fifty six paise of every rupee approved across all twelve projects.

The approved portfolio, unit by unit

The table below is the annexure record, with the two May 2026 approvals appended. Investment figures are the values stated in the government documents, not estimates, and capacity is quoted in the unit the document used.

Read down the investment column and the shape of the programme becomes obvious. One project is a wafer fab and costs more than the other eleven combined. The smallest entry, an expansion at Continental Device India in Punjab, is approved at Rs 117 crore, which is about one eight-hundredth of the Dholera figure and still counts as a unit in every headline that says twelve. Counting projects tells you about policy activity. Counting rupees tells you where the industry is actually being built, which is why this record maintains a capital ledger rather than a project tally.

What is actually producing

Approval is not production, and the government has been unusually specific about the gap. The reply of 1 April 2026 states that commercial production has commenced from two plants, Micron and Kaynes DURABLE, and that two more were expected to begin commercial production during the same year TARGET. The Cabinet release of 15 July 2026 repeats the position in shipment language: two projects have started commercial shipments from India and two more are expected to start soon.

So the honest summary in August 2026 is that India has two semiconductor plants shipping commercially, both of them back-end operations in Gujarat, and ten approved units at various stages of construction or fit-out. Micron's Sanand facility does assembly and test for DRAM and NAND products at a stated capacity of around 14 million units per week. Kaynes Technology India, also in Gujarat, runs wire bond interconnect and substrate-based packages with technology provided by ISO Technology Sdn Bhd and AOI Electronics, at a stated capacity above 6.33 million chips per day.

One older plant sits outside the mission portfolio entirely and is routinely forgotten. The Semi-Conductor Laboratory at Mohali fabricates wafers today, and the same 1 April 2026 reply records that of 211 chips taped out by 75 Indian institutions using government-provided design tools, 149 were fabricated at 180nm at SCL Mohali and 62 at overseas foundries. India has been fabricating chips for years. What it has not had is a commercial fab at scale, which is the thing under construction at Dholera.

The front end: the Dholera wafer fab

The Union Cabinet approved the Tata Electronics fab at Dholera on 29 February 2024 DURABLE, and the Fiscal Support Agreement, which is the instrument that converts an approval into an enforceable commitment, was signed on 5 March 2025 with the central government committing fifty percent fiscal support. The approved investment is Rs 91,526 crore, the plant is designed for up to 50,000 wafer starts per month on 300mm wafers, and the technology partner is Powerchip Semiconductor Manufacturing Corporation of Taiwan.

Construction evidence is more useful than announcement volume. Civil work was at about 50 percent as of mid-2026 with cleanroom fit-out under way REPORTED. Fugro, Cengrs and Geo Dynamics were engaged to redesign the main fab foundations after soil testing found the ground unsuitable, reported by The Economic Times on 28 November 2025 REPORTED. The Ministry of Commerce and Industry notified a special economic zone for Tata Semiconductor Manufacturing Private Limited at Dholera on or about 16 April 2026 covering 66.16 hectares. ASML and Tata Electronics both announced a memorandum of understanding on 16 May 2026 naming a node range of 28, 40, 55, 90 and 110nm REPORTED.

Two things are commonly stated about this fab that the sourced record does not support. The first is a first-silicon date. No primary source publishes one. What exists is commercial production guided to mid-2028 by the union minister on 17 July 2026 TARGET, and this record scores that date rather than any earlier claim. The second is the node. Tata's official range remains 28 to 110nm and the Tata Sons chairman's FY25 letter described starting at 28nm, while Bloomberg on 17 July 2026 and TrendForce on 20 July 2026 reported the plant opening mostly at 90nm. Both statements are on the record and neither has been withdrawn, which is why the node contradiction is tracked as an open item rather than resolved in favour of the more flattering version.

The back end: why packaging came first

Nine of the twelve approved units are packaging plants, and that ratio is a rational answer to a capital problem rather than a lack of ambition. A packaging plant can be built for a few thousand crore, staffed with engineers trained in months rather than years, and connected to customers who already buy assembly services on the open market. A wafer fab requires process knowledge accumulated over years, a technology partner willing to transfer it, and a customer base prepared to qualify a new supplier. Assembly and test is the rung most industrialising economies climb first, because it converts capital into shipped product on a timescale a private board can underwrite.

The stated capacities show what that layer will actually do. Tata Electronics at Jagiroad in Assam is approved at Rs 27,120 crore for 48 million units per day using indigenous packaging technologies. CG Power's Sanand joint venture with Renesas Electronics America and STARS Microelectronics of Thailand is approved at Rs 7,584 crore for about 15.07 million units per day, with technology supplied by Renesas of Japan and STARS. Advanced System in Package Technologies in Andhra Pradesh, working with APACT of South Korea, is approved at Rs 480 crore for about 96 million units per year. Suchi Semicon at Surat will assemble and test discrete semiconductors at 1,033.20 million chips per year for power electronics, analog and industrial customers.

Not all of it is conventional wire bonding. 3D Glass Solutions in Odisha is approved for flip chip ball grid array assembly, radio frequency and antenna-in-package system-in-package products, glass interposers with passives and silicon bridges, and three-dimensional heterogeneous integration modules. That is genuinely advanced packaging work, and it belongs on the record as such. If you want the engineering rather than the portfolio, the packaging and test explainer covers what these plants physically do.

The design layer, and the layers that are missing

Design is the one part of the chain India did not have to build from zero. The 1 April 2026 reply records chip design tools from eight companies provided free to 315 universities, with usage exceeding 200 lakh hours, 24 approved design projects covering video surveillance, drone detection, energy metering, microprocessors, satellite communications and broadband and internet-of-things systems on chip, fourteen of those companies raising more than Rs 650 crore in venture funding, and seven chips fabricated including at nodes as advanced as 12nm. The Cabinet release of 15 July 2026 counts 105 startups developing chips.

The gaps are equally documented, and stating them is not pessimism. No logic node below 28nm is approved anywhere in India. No extreme ultraviolet lithography tool is in the country, and none is needed for the mature-node work approved so far, a point the mature node guide argues from demand rather than from consolation. There is no memory wafer fab: Micron packages memory at Sanand, which is a different activity from fabricating it. Silicon wafers, photoresists, electronic-grade specialty chemicals and effectively all lithography, etch and deposition tools are imported today.

The early moves against those gaps are visible but small and should be tiered accordingly. INOX Air Products has a reported Rs 500 crore electronic specialty gas hub with a 200 tonnes per day air separation unit at Dholera REPORTED. Fujifilm India signed a memorandum with the Gujarat State Electronics Mission on 30 June 2026 to explore a semiconductor materials base at Dholera, which is exploratory by its own wording. Neither is a supply contract, and no supply agreement between any of these suppliers and Tata Electronics is in the public domain.

The policy machinery behind the numbers

The Semicon India Programme was approved by the Union Cabinet in December 2021 and launched in January 2022 with an incentive framework of Rs 76,000 crore, offering fiscal support of up to fifty percent of project cost for silicon fabs, compound semiconductor facilities, assembly and testing units and chip design. That fifty percent is the single most consequential number in Indian semiconductor policy, because it is what makes a Rs 91,526 crore fab financeable for a private balance sheet.

Central support is not the whole stack. Gujarat's semiconductor policy adds forty percent of the capital assistance approved by the centre, on top of land and utility instruments, which is why unit siting has clustered the way it has. Disbursal is milestone-linked rather than upfront, so money follows verified construction rather than press releases, and the twelve months between the Dholera Cabinet approval and its Fiscal Support Agreement is a fair indication of how long the paperwork between decision and contract actually takes.

The second phase is now approved. The Union Cabinet cleared Semicon 2.0 on 15 July 2026 with a total budget outlay of Rs 1,27,500 crore, built on six pillars including design, machines and materials, and manufacturing DURABLE. The emphasis on equipment and materials matters more than the headline number, because that is precisely the layer the current portfolio does not touch. Separately, the Union Budget for 2026-27 carried a provision of Rs 1,000 crore for industry-led research and training centres under the new mission, which is a research line item rather than a manufacturing subsidy and should not be quoted as the programme's size.

What this record watches next

Four signals would change the picture materially, and none of them is a ribbon cutting. The first is a customer qualification announcement from the Dholera fab, because a fab without qualified customers is a building. The second is tool ordering and move-in, since equipment purchase orders are the hardest evidence of schedule that exists in this industry and they precede production by a long, measurable interval. The third is the first approvals under Semicon 2.0 in equipment and materials, which would begin closing the import dependency described above. The fourth is confirmation of the two additional plants the government expects to reach commercial shipments, which would take India from two producing units to four.

Everything else on the public record is either approved and building, in which case this record tracks its construction evidence, or announced and unbuilt, in which case it stays at reported tier until something physical happens. That distinction is the whole discipline. The first commercial fab guide carries the Dholera story in full, and the twelve-unit portfolio above is the national frame it sits inside.