The answer desk

How many semiconductor projects are approved in India?

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

Twelve. As of 25 August 2026, twelve semiconductor manufacturing projects hold approval under the India Semiconductor Mission, carrying about Rs 1.64 lakh crore of approved investment. The Press Information Bureau put the count at twelve when the Union Cabinet cleared Crystal Matrix and Suchi Semicon on 5 May 2026, and the mission's own site, last updated 19 August 2026, shows no approval after that pair. The twelve break down as one silicon wafer fab, one silicon carbide fab, one gallium nitride display fab and nine assembly, test and packaging plants, spread across six states.

The count, and the date it is true as of

Twelve. As of 25 August 2026, twelve semiconductor manufacturing projects hold approval under the India Semiconductor Mission, and the approved investment behind them comes to about Rs 1.64 lakh crore DURABLE. That is not this desk's estimate. It is the government's own count, printed by the Press Information Bureau on 5 May 2026 in the release announcing the Union Cabinet's clearance of two further units, which states that the total number of approved projects under the India Semiconductor Mission reaches twelve, with cumulative investments of around Rs 1.64 lakh crore.

A count like this decays, so the date stamp carries more weight than the number. This record checked the India Semiconductor Mission's own site on 25 August 2026. The page carried a last-updated stamp of 19 August 2026, and it showed no approval later than the pair cleared on 5 May 2026 DURABLE. On that evidence the count has held at twelve for more than three months. Where a figure below carries no separate attribution, it is taken from the annexure to a Lok Sabha reply submitted by the Minister of State for Electronics and Information Technology on 1 April 2026 and published by the Press Information Bureau, which lists the ten projects approved up to that date with investment and capacity for each.

The twelve, one line each

The list below is the approved portfolio in full. Investment figures are the approved commitments as printed by the government, not money spent.

The capacities attached to those approvals are worth reading alongside the money, because they say what each plant is for. The Dholera fab is approved at around 50,000 wafer starts per month DURABLE. The Assam plant is approved at 48 million units a day using packaging technologies described in the approval as indigenous. Micron's Sanand plant is approved at around 14 million units a week for DRAM and NAND products. CG Power's Sanand plant is approved at around 15.07 million units a day, with technology from Renesas Electronics of Japan and STARS Microelectronics of Thailand. Kaynes is approved at more than 6.33 million chips a day for wire bond interconnect and substrate based packages. The Jewar unit is approved at around 20,000 wafers a month, equivalent to 36 million chips a month, for display driver integrated circuits using gold bump technology with chip probing and die processing, in a joint venture with Foxconn and with technology from Hon Hai of Taiwan. SiCSem is approved at 5,000 wafers a month of silicon carbide with a packaging line of 8 million units a month. 3D Glass Solutions is approved at around 5,800 glass panels a month, with assembly at 4.20 million units a month. Continental Device India is approved at around 158.38 million units a year of high power discretes including MOSFETs, IGBTs and Schottky diodes in both silicon and silicon carbide. Advanced System in Package Technologies is approved at around 96 million units a year, with technology from APACT of South Korea.

The two most recent approvals were published in more detail because they are newer. Crystal Matrix Limited was cleared for an integrated compound semiconductor fabrication and assembly, test, marking and packaging facility at Dholera making mini and micro LED display modules, with gallium nitride foundry services including epitaxy on six inch wafers, at an annual capacity of 72,000 square metres of display panels and 24,000 sets of red, green and blue epitaxy wafers. Suchi Semicon Private Limited was cleared for an outsourced assembly and test facility at Surat making discrete semiconductors at a proposed 1,033.20 million chips a year, aimed at power electronics, analog integrated circuits and industrial systems. The two carry a combined approved investment of around Rs 3,936 crore DURABLE, and the release puts expected employment at 2,230 skilled professionals TARGET. The government did not publish the split between the two projects, so this record does not invent one.

What approval actually means, and the four gates after it

The word approved does a lot of work in headlines and very little in engineering. Under this programme, approval means the Union Cabinet has accepted a proposal into the scheme and committed fiscal support, which for manufacturing projects runs at 50 percent of eligible project cost. It does not mean money has moved, ground has broken, or a single wafer exists. Four gates sit between a Cabinet approval and a chip.

• The fiscal support agreement. The formal contract that turns an approval into a disbursement schedule. For the Dholera fab, Cabinet approval came in February 2024 and the agreement was signed on 5 March 2025 DURABLE, an interval of about twelve months that is itself a useful benchmark for reading any fresh approval.
• Financial close and construction start. Ground breaking, civil works, and the structural shell.
• Tool move-in and hookup. The point at which equipment orders become visible in supplier disclosures, and the point at which a project becomes very hard to reverse.
• Qualification and commercial shipment. Yield, customer qualification, and the first invoice. This is the only gate that produces revenue.

Counting approvals therefore measures policy throughput, not industrial output. A record that wants to be useful has to carry both numbers, which is why this desk keeps the approval count and the production count in separate columns on the capital ledger.

The arithmetic, checked

Headline totals are worth testing against their own components, because that is where reporting errors surface. The ten projects listed in the April 2026 annexure sum as follows: Rs 22,516 crore plus Rs 91,526 crore plus Rs 27,120 crore plus Rs 7,584 crore plus Rs 3,307 crore plus Rs 3,706 crore plus Rs 1,943 crore plus Rs 2,066 crore plus Rs 117 crore plus Rs 480 crore, which comes to Rs 1,60,365 crore. That ten project subtotal matches the headline the same reply used, about Rs 1.6 lakh crore. Add the Rs 3,936 crore approved for the two Gujarat units in May 2026 and the total becomes Rs 1,64,301 crore, which matches the headline used in the May release, about Rs 1.64 lakh crore. Both totals reconcile with their own components, which is a small but real confirmation that the published parts and the published sums describe the same portfolio rather than two different tallies that happen to be quoted together.

The same arithmetic exposes how concentrated the book is. The Dholera fab alone accounts for about 56 percent of all approved semiconductor investment in India. The three largest projects, the Dholera fab, the Assam assembly plant and Micron's Sanand plant, together account for about 86 percent. The remaining nine projects share roughly 14 percent between them. Anyone describing India's semiconductor programme as broad-based should be asked which of those three they mean, because the portfolio is one very large front end project, two large back end plants, and a long tail of specialist units.

What the portfolio is made of, and what it is not

Sorted by what the plants actually do, the twelve fall into four groups. There is one silicon wafer fab, the Tata Electronics project at Dholera built in technology partnership with Powerchip Semiconductor Manufacturing Corporation of Taiwan. There is one silicon carbide compound fab, SiCSem in Odisha, described in the Cabinet record as the first commercial compound fab in the country. There is one gallium nitride mini and micro LED display fab, Crystal Matrix at Dholera. The other nine are assembly, test and packaging plants of various kinds, from conventional wire bond lines to glass substrate work.

Two clarifications matter because both are routinely garbled. First, the Jewar unit is widely called a display driver fab, but the approval describes gold bump technology with chip probing and die processing services, which is wafer level back end work rather than front end wafer fabrication. Second, an assembly and test plant is not a fab and does not become one by being large, a distinction set out at what an OSAT is. Micron's Sanand plant packages memory that is fabricated elsewhere; no memory fab exists in India.

What the portfolio does not contain is equally load bearing. There is no approved logic fab below 28 nanometres, no extreme ultraviolet lithography, and no silicon wafer manufacturing plant. The programme also runs a separate design track, and confusing the two tracks is the most common way the count goes wrong.

Why other trackers print a different number

Search results for this question return counts from ten to thirteen, and every one of them can be defended if you accept its counting rule. Ten is the correct count for any statement made between the 2025 clearances and 5 May 2026, and it is the number the government itself used in its April 2026 parliamentary reply. Thirteen appears on several aggregator sites, and this desk found no Cabinet approval after 5 May 2026 in either the Press Information Bureau record or the mission's own site as read on 25 August 2026, so it prints twelve and will correct the moment a thirteenth approval is published.

Larger numbers usually fold in the design track. The same April 2026 reply records 24 approved chip design projects and design tools supplied to 315 academic institutions at no cost, and the May 2026 Cabinet release adds design infrastructure support to 104 startups. Those are real, they are part of the same programme, and they are not manufacturing plants. The April reply separately records that a scheme for electronics components had received 249 applications by that date, which is an application count rather than an approval count. Whenever a headline number looks too high, the usual explanation is that manufacturing approvals, design approvals and applications have been added together.

The execution scoreboard as of 25 August 2026

Approvals are the easy half. The mission's own site records that Micron's Sanand plant commenced commercial production in February 2026, that Kaynes Semicon commenced commercial production in March 2026, and that CG Semi was inaugurated in July 2026 DURABLE. It also records a ground breaking at the Jewar unit in February 2026 and at the 3D Glass Solutions plant in Bhubaneswar in April 2026 DURABLE. On 5 May 2026 the Cabinet release described two projects as already making commercial shipments with two more expected soon, which is consistent with the sequence above.

The one plant that has not shipped and will not ship for some time is the largest. Commercial production at the Dholera fab is guided to mid-2028 TARGET, a figure given by the Union Minister for Electronics and Information Technology on 17 July 2026. No source publishes a supported first silicon date for that plant, and this record does not print one. The older talk of first chips by December 2026 traces to ministerial statements rather than any company filing, and it is not treated here as fact.

Dholera's share, stated precisely

Two of the twelve approved projects sit at Dholera: the Tata Electronics silicon fab and the Crystal Matrix compound fab and packaging unit. That is the whole of Dholera's presence in the national count, and the national total of about Rs 1.64 lakh crore is a country figure that must never be described as Dholera investment. What is true is that the single largest line in the national portfolio is at Dholera, and that the site therefore carries an outsized share of the programme's delivery risk. If the Dholera fab slips badly, roughly 56 percent of the approved book slips with it.

Four of the other approved plants sit elsewhere in Gujarat, three at Sanand and one at Surat, which is why the Gujarat total and the Dholera total differ so widely and are so often merged in error. The distinction between the two geographies is set out in Sanand and Dholera compared, and the wider state picture in the Gujarat semiconductor ecosystem.

What is not public, and what to watch

Several things a reader would reasonably expect to be published are not. There is no public project by project disbursement schedule, so the gap between approved investment and released fiscal support cannot be calculated from open sources. The split of the Rs 3,936 crore between Crystal Matrix and Suchi Semicon has not been published. Customer names for the Dholera fab have not been announced. Equipment purchase orders, which are the most reliable early indicator that a fab is real, are not disclosed by either buyer or seller in this case.

Three signals would move this page. A thirteenth Cabinet approval, which would change the count and, depending on its type, the composition. A disclosed equipment order at Dholera, which would date the tool move-in gate. And the first customer qualification announcement from any Indian plant making chips rather than packaging them, which would move India from an approvals story to a production story. The programme's own second phase, covered at the ISM 2.0 file, points at equipment, materials and design rather than at more fabs, so the next approvals may not look like the first twelve. How this record treats the Dholera specific parts of the programme is set out at Dholera and the India Semiconductor Mission.