The answer desk
Which Indian states have semiconductor projects?
Published 25 August 2026. Facts verified to 25 August 2026 unless dated otherwise.
Six. Every one of the twelve semiconductor manufacturing projects approved under the India Semiconductor Mission as of 25 August 2026 sits in Gujarat, Assam, Uttar Pradesh, Odisha, Andhra Pradesh or Punjab. Gujarat holds six of the twelve and, on this desk's arithmetic over the investment figures published by the Press Information Bureau, about 78 percent of the approved money. Assam holds the second largest single commitment. Odisha holds two projects, and Uttar Pradesh, Andhra Pradesh and Punjab hold one each. Other states have semiconductor policies but no approved manufacturing project.
The six states, and the two ways to count them
Six states hold approved semiconductor manufacturing projects in India: Gujarat, Assam, Uttar Pradesh, Odisha, Andhra Pradesh and Punjab. That covers all twelve projects approved under the India Semiconductor Mission as of 25 August 2026, a count confirmed by the Press Information Bureau release of 5 May 2026 and unchanged on the mission's own site as read on 25 August 2026, which carried a last-updated stamp of 19 August 2026 DURABLE. Project locations and investment figures used throughout this page come from the annexure to a Lok Sabha reply of 1 April 2026 published by the Press Information Bureau, and from that May 2026 Cabinet release.
There are two honest ways to answer the question, and they give different rankings. Counting projects, Gujarat holds six, Odisha two, and Assam, Uttar Pradesh, Andhra Pradesh and Punjab one each. Counting money, Gujarat holds about Rs 1,28,869 crore, Assam Rs 27,120 crore, Odisha Rs 4,009 crore, Uttar Pradesh Rs 3,706 crore, Andhra Pradesh Rs 480 crore and Punjab Rs 117 crore. Those state totals are this desk's arithmetic over the government's published per project figures, and they sum to Rs 1,64,301 crore, which reconciles with the official headline of about Rs 1.64 lakh crore. On the money count Gujarat holds about 78 percent of the national book and Assam about 16.5 percent, leaving roughly 5 percent for the other four states combined.
Note that some aggregator sites report seven states. This record counts six, from the government's own project list, and will correct the moment a seventh appears in the approval record. A state having a semiconductor policy, a memorandum, or an announced intention is not the same as holding an approved project, and the difference between those two categories is where most of the confusion in this subject lives.
Gujarat: six projects and most of the money
Gujarat holds six of the twelve: the Tata Electronics silicon wafer fab at Dholera at Rs 91,526 crore, Micron's assembly and test plant at Sanand at Rs 22,516 crore, the CG Power joint venture with Renesas and STARS Microelectronics at Sanand at Rs 7,584 crore, Kaynes Technology India at Sanand at Rs 3,307 crore, and the two units cleared on 5 May 2026, Crystal Matrix at Dholera and Suchi Semicon at Surat, which share a combined approval of about Rs 3,936 crore DURABLE.
Those six sit in three distinct places, and merging them is the most common factual error in coverage of this subject. Sanand is a mature industrial belt in Ahmedabad's orbit with three operating or near operating back end plants. Dholera is a greenfield build carrying the front end fab and, since May 2026, a compound semiconductor and packaging unit. Surat is a fourth location entirely, in the south of the state. The comparison this record keeps between the first two is at Sanand and Dholera compared, and the fuller state picture at the Gujarat semiconductor ecosystem.
Gujarat is also the one state whose incentive stack this desk has read line by line. Under the Gujarat Semiconductor Policy 2022-27, announced on 27 July 2022 and the first state semiconductor policy in India, the state contributes 40 percent of the capital assistance approved by the Centre, offers a 75 percent land subsidy on the first 200 acres in Dholera Semicon City, prices water at Rs 12 per cubic metre, and adds a power subsidy of Rs 2 per unit for ten years with electricity duty exemption DURABLE. Stacked on the central support of 50 percent of eligible project cost, that is the arithmetic behind Gujarat's position on this list. The itemised terms are set out at Gujarat's semiconductor incentives.
Assam: one project, the second largest cheque
Assam holds a single approved project and it is the second largest in the country: the Tata Electronics assembly and test plant at Jagiroad, approved at Rs 27,120 crore with a stated capacity of 48 million units a day, using packaging technologies the approval describes as indigenous DURABLE. It was cleared by the Union Cabinet in February 2024, in the same round as the Dholera fab.
The Assam plant matters to the Dholera story for a structural reason rather than a geographic one. Tata Electronics is building a front end in Gujarat and a back end in Assam, at opposite ends of the country, which means wafers made at Dholera would travel most of the width of India to be turned into packaged chips. That is not unusual by industry standards, where front end and back end routinely sit in different countries, but it does put weight on logistics and on scheduling discipline between two plants ramping at once. It also involves customs procedure rather than simple domestic freight, because the Dholera fab sits inside a special economic zone notified on 66.16 hectares at about April 2026 DURABLE, and movement from a zone into the domestic tariff area is treated as an import. It also means the two states are complements rather than competitors in this particular corporate plan.
Uttar Pradesh: one project, and what it actually does
Uttar Pradesh holds one approved project, the Vama Sundari Investments unit at Jewar, a joint venture with Foxconn approved at Rs 3,706 crore. The approval describes display driver integrated circuits made using gold bump technology, with chip probing facilities and die processing services, at a capacity of around 20,000 wafers a month or 36 million chips a month, with technology provided by Hon Hai of Taiwan DURABLE. The mission's site records a ground breaking in February 2026.
This unit is commonly described as a display driver fab, and the description is loose. Gold bumping, probing and die processing are wafer level operations performed on wafers that have already been fabricated elsewhere. The unit is a specialised back end plant, not a front end wafer fab, and the difference matters for anyone counting fabs. Its location at Jewar, beside the new airport, reflects the same logic that puts high value electronics near air cargo everywhere: chips move by air, and a plant's distance from a runway is a real cost line.
Odisha: two projects, and the compound bet
Odisha holds two approved projects, both at Info Valley in Bhubaneswar. SiCSem Private Limited is approved at Rs 2,066 crore for an integrated silicon carbide facility with Clas-SiC Wafer Fab of the United Kingdom as fab technology partner and Continental Device India as packaging partner, at 5,000 wafers a month with a packaging capacity of 8 million units a month DURABLE. The Cabinet record describes it as the first commercial compound fab in the country. 3D Glass Solutions is approved at Rs 1,943 crore for a vertically integrated advanced packaging and embedded glass substrate unit, covering flip chip ball grid array assembly, radio frequency and antenna in package work, glass interposers and three dimensional heterogeneous integration modules, at around 5,800 panels a month. The mission's site records its ground breaking in April 2026.
Odisha's two projects carry less than 3 percent of the national money and a disproportionate share of the national novelty. Silicon carbide is a different material system from the silicon complementary metal oxide semiconductor process at Dholera, aimed at power electronics rather than logic, and glass substrate packaging is a frontier the established packaging industry is still working through. A state with no prior semiconductor base bought into two technically ambitious projects at modest cost, which is a defensible strategy and a risky one, and this record will score both on delivery rather than on intent.
Andhra Pradesh and Punjab: the small end of the ladder
Andhra Pradesh holds one approved project, Advanced System in Package Technologies, at Rs 480 crore with technology from APACT of South Korea and a capacity of around 96 million units a year. Punjab holds one, an expansion by Continental Device India at Mohali at Rs 117 crore, making high power discrete devices including MOSFETs, IGBTs, Schottky diodes and transistors in both silicon and silicon carbide, at around 158.38 million units a year DURABLE.
The Punjab entry is the smallest approval in the portfolio by a wide margin, and it is also the oldest industrial base on the list. Continental Device India has manufactured discretes at Mohali for decades, which makes this an expansion of a running business rather than a greenfield bet. Punjab is also where India's existing government fab sits, the Semi-Conductor Laboratory at Mohali, which is not part of this scheme and is not counted among the twelve. Anyone assembling a state map from press coverage will trip over that, because Mohali appears twice in the story for two entirely different reasons.
The state policy layer, and what this desk will not print
Several states on this list have published electronics or semiconductor policies, and so have several states that hold no approved project at all. This record has read and verified the Gujarat policy in detail, and its terms are given above. For every other state named on this page, this desk has not read the notified policy document against a primary source, and it does not print incentive line items it has not read. What can be said without a document in hand is structural rather than numeric: state support in this sector generally takes the form of a top up on the central capital subsidy, land at a discount or on long lease, power tariff support and duty exemptions, and single window clearance handling. Which state offers which band, at what ceiling, is exactly the kind of detail that gets garbled in secondary coverage, so the honest position is a blank rather than a plausible table.
What interstate competition does to the national build-out is a fair question and the answer is mixed. Competition has clearly pulled more states into offering serviced land and power at credible terms, which lowers the cost of the next project. It has also spread twelve projects across six states in a sector whose entire economics depend on clustering, since a fab wants its gas supplier, its chemical logistics and its trained technicians within an hour, not a time zone. The counter argument is that back end plants cluster far less tightly than front end plants, and nine of the twelve are back end. On that reading the current spread is less costly than it looks, and the real test is whether the two front end sites, Dholera and Bhubaneswar, each attract a supplier layer around them.
What a state actually supplies, and why it decides siting
Reduced to essentials, a state brings four things to a semiconductor project, and money is the least distinctive of them. It brings land at scale with clean title and the right zoning, which is why greenfield industrial regions with pre-assembled land banks keep winning these projects. It brings power that is both plentiful and stable, because a fab treats a voltage dip as a scrapped wafer lot rather than an inconvenience. It brings water at volume with the treatment and recycling capacity to match, since ultrapure water consumption is one of the defining utility loads of a wafer fab. And it brings administrative speed, which in practice means environmental clearance, building approvals and customs handling moving in months rather than years.
Central money is broadly uniform across states at 50 percent of eligible project cost, so it does not differentiate. State top ups differ, but by amounts that are small against a project of this size. The four items above are what actually decide where a plant goes, and they are also the items that are hardest to fake in a press release. When reading the next state announcement, the useful questions are how many contiguous acres are already in the state's hands, what the firm power position is, where the water comes from, and how many months the last comparable clearance took. The state by state detail this record keeps is being built out at the semiconductor state map, and the incentive mechanics that sit under all of it at the Dholera semiconductor policy guide.
What to watch
Three developments would redraw this map. A thirteenth approval in a seventh state, which would break the six state pattern and signal that the programme is deliberately spreading rather than clustering. A second front end fab approval anywhere, which would matter far more than another packaging plant because front end projects pull supplier ecosystems behind them. And the first published evidence that a supplier layer is forming around a site rather than around a state capital, which is the difference between a cluster and a set of unrelated factories that happen to share a border.
The counter signal to watch is a state announcement that is not accompanied by a Cabinet approval. Memoranda are cheap, and this record has already documented several in adjacent sectors that were reported as investment and never became projects. Until a proposal appears in the approval record with a rupee figure and a capacity attached to it, it belongs in the reported column, not on the map.