Guide

How fab incentives work in India: the money mechanics, gate by gate

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

The sentence the government is paying half is repeated constantly and examined almost never, so this page takes it apart into the instrument, the gates it passes through and the arithmetic it produces.

The claim this page checks

Ask anyone who follows Indian industry why a fab is being built at Dholera and the answer arrives in one clause: the government is paying half. It is close enough to true that it never gets questioned, and vague enough that almost every important detail hides inside it. Half of what, paid to whom, when, against what evidence, and with what added by the state government are five separate questions with five separate documents behind them. This page answers them in order and marks the places where the public record simply stops.

The instrument, stated precisely

The Semicon India Programme was approved by the Union Cabinet in December 2021 with an incentive framework of Rs 76,000 crore. Under it, fiscal support of up to 50 percent is offered for silicon fabs, compound semiconductor facilities, assembly and testing units and chip design, per the Press Information Bureau backgrounder on India Semiconductor Mission 2.0 dated 7 February 2026 DURABLE. The programme runs as a family of schemes rather than one cheque: a fabs scheme, a display fabs scheme, a compound semiconductor and assembly, testing, marking and packaging scheme, and the Design Linked Incentive scheme, which carries its own outlay of Rs 1,000 crore and supports design companies with incentives of up to Rs 15 crore each. That Rs 76,000 crore figure is the first phase of the programme; a second phase carrying its own and considerably larger outlay was approved in July 2026 and is set out further down this page.

Two words in that sentence carry most of the weight. Up to means the approved share can be lower than half, decided project by project. Eligible means the base is not the headline project cost that appears in press releases but a defined subset of capital expenditure agreed with the government. Neither the eligible base for any approved Indian project nor the approved percentage for any individual project has been published as far as this desk can establish, which means every calculation of what a specific company will receive is an upper bound and should be presented as one.

The three gates the money passes through

An approved fab does not receive support because a minister announced it. It receives support because it has passed a sequence of documented gates, and the sequence is visible in the Dholera record.

• Gate one, Cabinet approval. The Union Cabinet approved the Tata Electronics fab at Dholera on 29 February 2024 under the India Semiconductor Mission DURABLE. Approval establishes that the project is inside the scheme and fixes its headline investment figure, recorded by PIB as Rs 91,526 crore with a stated capacity of about 50,000 wafer starts per month.
• Gate two, the fiscal support agreement. The agreement for the Dholera fab was signed on 5 March 2025, a year after approval DURABLE. This is the contract that converts a Cabinet decision into an obligation with conditions attached. Its text and payment schedule are not public.
• Gate three, disbursement. Money moves as the project moves. The clearest public evidence of that structure is the programme's own annual arithmetic, discussed in the next section, and the American parallel, where the Commerce Department reimburses awardees against completed milestones. This desk has not seen the Dholera disbursement schedule and does not know how much has been paid to date. That is a negative finding, and it is stated rather than filled in.

The gates matter because they separate real support from announced support. Our policy stack guide makes the same point structurally: a claim that names a dated instrument is a different class of claim from one that says government backing and moves on.

Where the money actually is, year by year

The Rs 76,000 crore headline is a framework, not a bank balance. What the government actually budgets in a given year is a much smaller number, and it is published. For 2026-27, the modified programme for development of the semiconductor and display manufacturing ecosystem carries a total financial outlay of Rs 8,000 crore, per the 7 February 2026 backgrounder DURABLE. The same document sets out what that year of money is expected to produce: under the fabs scheme, with one fab to be supported, investment during the year of Rs 4,000 crore and 1,500 jobs generated; under the compound semiconductor and assembly and test scheme, with nine units supported, investment of Rs 11,000 crore and 3,000 jobs; and under the Design Linked Incentive scheme, 30 design companies supported, ten semiconductor intellectual property cores developed and 200 design personnel employed TARGET.

Read those numbers as the shape of the flow rather than as a scorecard. A five-year framework of Rs 76,000 crore against an annual programme outlay of Rs 8,000 crore tells you that support is released in slices across the life of a project, which is the correct design for plants that take years to build and which is also why headline commitments and cash outflow will never match in any single year. It is the same structure that produces the persistent gap between committed and disbursed money in the American programme, discussed in our chips acts comparison.

That framework is no longer the whole of the commitment. On 15 July 2026 the Union Cabinet approved Semicon 2.0 for the development of India's semiconductor design and manufacturing ecosystem, with a total budget outlay of Rs 1,27,500 crore, per Press Information Bureau release 2284784 of that date, the decision also being published on the Prime Minister's Office website DURABLE. The release presents the programme as sustained long-term support building on the momentum of Semicon 1.0, organised on six pillars with design named first, and it restates the position on the ground: twelve manufacturing units approved with cumulative investment of over Rs 1.64 lakh crore, comprising one silicon fab, one silicon carbide fab, an integrated gallium nitride micro LED display fab and nine packaging units DURABLE.

Everything this page has said about gates applies to that figure with more force rather than less. Rs 1,27,500 crore is an approved outlay, which is the government granting itself the authority to commit money, and it sits several steps earlier in the sequence than the fiscal support agreement signed for the Dholera fab on 5 March 2025, which is a contract with conditions attached to a named plant. Nothing moves under a budget outlay until schemes are notified, applications are received and approved, and agreements are signed. This desk has not seen a scheme-wise split of the second-phase outlay, a list of its instruments or any disbursement schedule published for it, so the honest description as of 25 August 2026 is a large authorised headroom whose mechanics are still to come. Anyone reading it as spending to date, or as capital committed to a particular factory, is reading it wrong, and the distinction is the same one this record keeps between announced, committed and delivered money in the capital ledger.

The Dholera arithmetic, done in the open

Here is the calculation everyone does silently, done out loud with its assumptions labelled. The Tata Electronics project at Dholera is recorded by PIB at Rs 91,526 crore of investment. If the eligible cost were the whole of that figure, and if the approved share were the full 50 percent, central support would be about Rs 45,763 crore. Both conditions are assumptions, not facts, and the first is unlikely to hold in full, because eligible cost under such schemes typically excludes categories of spend. So the honest statement is that central support for the Dholera fab is bounded above by roughly Rs 45,700 crore on the published investment figure, and that the actual approved amount has not been published.

Now add the state layer. Under the Gujarat Semiconductor Policy 2022-27, the state provides 40 percent of the capital expenditure assistance approved by the Government of India, stacking on top of the central share rather than replacing it, per the Gujarat State Electronics Mission DURABLE. Applying that rate to the same illustrative ceiling gives about Rs 18,300 crore of state assistance, and a combined public share approaching 70 percent of eligible cost in principle. That is arithmetic on published rates, not a claim about any payment made, and the difference between those two things is the whole discipline of this record.

What Gujarat adds beyond the cash share

The state instruments are unusually specific, and several of them are written for Dholera by name. The land subsidy runs at 75 percent on the first 200 acres required for a fab project in Dholera Semicon City and 50 percent on additional land. Water is supplied at Rs 12 per cubic metre for five years with a 10 percent annual increase thereafter. Power carries a subsidy of Rs 2 per unit for ten years plus exemption from electricity duty. A desalination plant built within the project's first five years attracts a 50 percent capital subsidy. All of these sit in the Gujarat Semiconductor Policy 2022-27 and are set out line by line in our file on what Gujarat actually pays a semiconductor company DURABLE.

Operating-cost instruments deserve more attention than they get. A capital subsidy is paid once and shows up in the headline. A fixed water price for five years and a power subsidy for ten years change the plant's cost per wafer for a decade, which is the period during which a first-of-kind fab either becomes competitive or does not. In pure economics, the second set may matter more than the first.

The same instrument across very different projects

The up-to-50-percent rule is not written for fabs alone, and seeing the spread of approved project sizes is the quickest way to understand what the government has actually taken on. The 7 February 2026 backgrounder lists each approved project with its investment figure, and the range is wider than the coverage implies DURABLE.

• Micron Technology, Gujarat, Rs 22,516 crore, assembly and test for DRAM and NAND, capacity around 14 million units per week.
• Tata Electronics, Gujarat, Rs 91,526 crore, the wafer fab with PSMC of Taiwan, around 50,000 wafer starts per month.
• Tata Electronics, Assam, Rs 27,120 crore, packaging using indigenous technologies, capacity 48 million units per day.
• CG Power and Industrial Solutions, Gujarat, Rs 7,584 crore, a joint venture with Renesas Electronics America and STARS Microelectronics of Thailand.
• Kaynes Technology India, Gujarat, Rs 3,307 crore, more than 6.33 million chips per day.
• Vama Sundari Investments, Uttar Pradesh, Rs 3,706 crore, about 20,000 wafers per month and 36 million chips per month.
• 3D Glass Solutions, Odisha, Rs 1,943 crore, glass panel substrates and heterogeneous integration.
• SiCSem, Odisha, Rs 2,066 crore, 5,000 wafers per month with packaging capacity of 8 million units per month.
• Continental Device India, Punjab, Rs 117 crore, an expansion making high-power discrete devices.
• Advanced System in Package Technologies, Andhra Pradesh, Rs 480 crore, around 96 million units per year.

Two readings follow. First, the exchequer's exposure is concentrated: a single project accounts for more than half of the approved investment in that list, which is why the Dholera schedule carries political weight far beyond its share of the project count. Second, the same instrument that supports a Rs 91,526 crore fab also supports a Rs 117 crore expansion of an existing device maker, and the administrative effort of the two is nothing like proportionate to their sizes. A scheme designed around anchor projects tends to under-serve the small end, and whether the small end gets served is one of the quieter tests of the programme.

What taxpayers get, honestly

The case for these instruments is not that fabs are profitable investments for governments. It is that a country without domestic fabrication capacity has no leverage in a shortage, no upgrade path for its electronics assembly base, and no domestic customer for the equipment and materials industries it hopes to build later. Every large programme discussed on this site makes that argument in some form, and the Indian version adds a second one: the country already designs a great deal of silicon and fabricates almost none of it, so the subsidy is buying the missing half of an industry it already half owns.

The case against is equally straightforward and should be stated with the same clarity. Capital moves to whoever pays most, so incentive competition transfers public money to firms without necessarily changing where plants would eventually have been built. Support of this size to a single project creates a political stake in that project's success, which can degrade the quality of later official reporting on it. And the payback is slow: a fab that starts commercial production in the late 2020s will take years beyond that to reach the yields and customer base that justify the outlay. Anyone who tells you the return is obvious in either direction is not reading the same documents.

What the milestone structure does buy the public is protection against the worst failure mode. Money released against verified progress means an abandoned project costs the exchequer far less than a project paid for in advance. That is why the disbursement question, dull as it looks, is the most important unanswered question in Indian semiconductor policy, and why this desk keeps asking for the schedule rather than estimating it.

What is not public, and what to watch

Five gaps, each of which would sharpen every number on this page if it closed. The eligible cost base and approved percentage for each sanctioned project. The disbursement schedule and amounts paid to date under each fiscal support agreement, including the Dholera one signed on 5 March 2025. The state-level equivalent, meaning what Gujarat has actually released against its 40 percent commitment. The conditions attached to support, including any clawback provisions if capacity or employment targets are missed. And the operating guidelines for Semicon 2.0: how the Rs 1,27,500 crore outlay divides across the six stated pillars, at what support rates, and on what timetable applications open.

Until those are published, the defensible statements are the ones on this page: an instrument of up to 50 percent, a first-phase framework of Rs 76,000 crore, a second-phase budget outlay of Rs 1,27,500 crore approved by the Union Cabinet on 15 July 2026, an annual programme outlay of Rs 8,000 crore for 2026-27, a state stack of 40 percent of the central assistance plus land, power and water terms, and an unpublished payment record. The corporate side of the same question, meaning who is putting in the equity and what remains undisclosed, is handled at who funds the Dholera fab, and the entity building it has its own file at Tata Electronics.