The answer desk
Who is funding the Dholera fab?
Published 25 August 2026. Facts verified to 25 August 2026 unless dated otherwise.
Three layers pay for it. Tata Electronics is the promoter and carries the project through Tata Semiconductor Manufacturing Private Limited. The India Semiconductor Mission commits central fiscal support of up to 50 percent of eligible project cost, formalised in a fiscal support agreement signed on 5 March 2025. Gujarat adds 40 percent of the central capital assistance plus land, water and power subsidies under its 2022 policy. The debt structure, the tranche schedule and the amounts released so far are not public.
The three layers, named
The Dholera fab is paid for by three identifiable sources of money and one that is frequently miscounted. The promoter is Tata Electronics, which holds the project through Tata Semiconductor Manufacturing Private Limited, the entity named in the government's approval and in the special economic zone notification. The central government commits fiscal support of up to 50 percent of eligible project cost through the India Semiconductor Mission, formalised for this project in a fiscal support agreement signed on 5 March 2025 DURABLE. The Gujarat government adds a state layer on top of the central one under its Semiconductor Policy 2022-27. The fourth item, customs treatment inside the notified zone, is not cash and should never be added to the others as though it were, though it is worth real money.
The frequently miscounted item is Powerchip Semiconductor Manufacturing Corporation of Taiwan. PSMC is the technology partner on this project, and technology partnership is not the same thing as capital. No public document that this desk has seen establishes a PSMC equity contribution to the Dholera plant, and the honest entry is therefore that the partner's financial position is not public. The PSMC file holds what is known about the partner itself.
What Rs 91,000 crore is, and what it is not
The Union Cabinet approved the project on 29 February 2024 at Rs 91,000 crore, a figure of roughly eleven billion dollars DURABLE. The fiscal support agreement recorded by the Press Information Bureau on 5 March 2025 cites Rs 91,526 crore. The two numbers are the same project stated to different precision, and the more precise one belongs to the later and more formal document.
What that number is, is the announced project cost. What it is not, is a sum that has been spent, nor a sum that has been transferred, nor a measure of how much of the plant exists. Civil work at the site stood at about 50 percent as of mid-2026, with cleanroom fit-out under way REPORTED, and the largest single cost item in any fab, the tool set, is spent late in the sequence rather than early. A reader who sees Rs 91,000 crore and pictures that amount of construction on the ground has misread a budget as a bank statement.
It is also not necessarily identical to the base on which support is computed. Central support under the mission is expressed as a share of eligible project cost, and eligibility is defined by scheme rules that determine which categories of expenditure count. The scheme's definition of eligible capital expenditure for this project has not been published, so the exact base is not public. Anyone multiplying the headline by one half and treating the answer as a committed government cheque is performing arithmetic on an assumption.
The central layer: what fifty percent actually means
The Semicon India programme, under which the India Semiconductor Mission operates, was approved by the Union Cabinet in December 2021 with an incentive framework of Rs 76,000 crore, offering fiscal support of up to 50 percent for silicon fabs, compound semiconductor facilities, assembly and testing units and chip design, per the Press Information Bureau backgrounder of 7 February 2026. The full programme machinery is set out at how fab incentives work in India, and the outlay question is handled at what is the ISM budget.
Two dates carry more information than the percentage does. The Cabinet decision of 29 February 2024 created eligibility and fixed a ceiling, and it moved no money. The instrument that moves money is the fiscal support agreement of 5 March 2025, and agreements of this class exist to settle four questions: which categories of spending qualify, what evidence a claim has to carry, which verified milestone releases which tranche, and where the government stands if the plant is never finished.
The interval between those two dates deserves attention on its own. Twelve months and a few days separated a decision announced to the country from the contract that gave it effect, and this was the flagship project of a national mission rather than a marginal file. Timelines drawn afterwards tend to collapse that stretch into a single arrow, which is how the speed of Indian industrial policy gets systematically overstated.
The terms of this specific agreement are not public. What can be said generally is that capital subsidy schemes of this kind reimburse against certified expenditure rather than paying in advance, and that Indian incentive schemes conventionally release such support pari passu with project spending rather than up front REPORTED, a general practice rather than a published term of this agreement. The practical consequence is that the promoter funds first and claims after, which means the promoter carries the working capital and the timing risk of the government's own verification process. No figure for support actually released to this project has been published, and this record will not estimate one.
Gujarat's layer, line by line
Gujarat was the first Indian state with a dedicated semiconductor policy, announced on 27 July 2022, and its structure is additive rather than substitutive: the state contributes 40 percent of the capital assistance approved by the central government, on top of that assistance rather than in place of it DURABLE. The line-by-line reading of that policy covers the instruments in full. The items that bear on the fab's capital stack are these: a land subsidy of 75 percent on the first 200 acres required for a fab project in Dholera Semicon City and 50 percent on additional land; water at Rs 12 per cubic metre for five years, rising 10 percent annually thereafter; a power subsidy of Rs 2 per unit for ten years, with exemption from electricity duty; and a 50 percent capital subsidy for a desalination plant built within the project's first five years DURABLE.
Two of those instruments cut the cost of arriving and two cut the cost of staying, which is a less common balance than it sounds. Discounted land and a shared desalination bill reduce what the promoter has to fund before anything is produced. The water tariff and the power subsidy run for five years and ten years respectively, which means the state has agreed to sit inside the plant's operating cost sheet for a decade, and operating cost is what decides whether a fab is still winning orders in its second process generation.
The desalination line carries a deadline as well as a subsidy, since the plant has to be built inside the project's first five years to qualify. Read as an instrument rather than as generosity, that clause is an official expectation put in writing: the state assumes a fab on this coast will need a water source of its own rather than leaning on the regional supply, and it has priced half of that assumption into the policy.
The zone layer, which is money but not cash
The special economic zone notified for Tata Semiconductor Manufacturing Private Limited at Dholera around April 2026, covering 66.16 hectares DURABLE, adds a customs regime on top of the subsidy regime. Its value lies in the treatment of an import bill that, for a fab, accounts for most of the capital cost. It is genuinely material and it belongs in any honest account of how this project is financed. It is also not a transfer, it does not appear as a receipt, and adding a notional duty saving to the subsidy figures to produce a larger headline is the kind of arithmetic this desk exists to refuse. The regime and its 2025 rule changes are set out at the Dholera SEZ file.
What is not public
The list is longer than the list of what is, and stating it plainly is the point of this page.
- The debt structure. Whether the project carries external borrowing, from which lenders, on what terms and with what security, has not been published.
- The equity split inside the promoter group, and whether any partner or customer holds a stake in the project entity.
- The milestone schedule in the fiscal support agreement, and therefore the trigger points for each tranche.
- How much central or state support has actually been released to date. Approval and agreement are documented; disbursement is not.
- The definition of eligible project cost applied to this project, which sets the base for the 50 percent computation.
- Any customer prepayment, capacity reservation or take-or-pay arrangement, which in this industry is a normal financing instrument and which, if it exists here, has not been disclosed.
None of these absences is evidence of a problem. Project finance is ordinarily confidential and there is no obligation to publish it. The failure mode this record guards against is the opposite one: coverage that treats an undisclosed structure as though it were a known one, and readers who conclude from a large approval figure that a correspondingly large amount of money has moved.
What Semicon 2.0 changes for this project, and what it does not
The Union Cabinet approved the next phase of the Semicon India programme on 15 July 2026, with a total budget outlay of Rs 1,27,500 crore DURABLE. That decision does not retroactively alter the terms under which the Dholera fab was approved. This project sits inside the first phase, was cleared against the Rs 76,000 crore framework, and its support is governed by an agreement signed sixteen months before the second phase existed.
Where the second phase touches Dholera is indirect and slower. Its stated pillars include incentives for companies manufacturing the machines, materials, chemicals and gases that fabs consume, and further strengthening of the assembly and test industry. If those instruments succeed, the fab's suppliers become eligible for support of their own, and the cluster around the plant gets cheaper to build. That is a second-order effect on a multi-year horizon, and it should not be reported as new money for the fab, because it is not.
The annual budget line, which is the closest thing to a cash signal
A scheme outlay is a ceiling spread across years. What a government actually spends in a given year appears in that year's budget provision, and the two are routinely confused by people quoting the larger number. For 2026-27, the Press Information Bureau backgrounder of 7 February 2026 records that the Modified Programme for Development of Semiconductor and Display Manufacturing Ecosystem in India carries a total financial outlay of Rs 8,000 crore DURABLE.
The same document publishes projected targets for the year, and one row of that table is about this project. Under the modified scheme for semiconductor fabs, the number of fabs to be supported during 2026-27 is one, the projected investment during the year is Rs 4,000 crore, and projected employment generated is 1,500 persons TARGET. There is only one silicon fab in the approved portfolio, so the row can be read for what it is. Two adjacent rows give the rest of the programme's shape for the year: nine units under the compound semiconductor, silicon photonics, sensors, discrete fabs and assembly and test scheme, with Rs 11,000 crore of investment by those units during the year and 3,000 persons employed, and thirty design companies supported under the design-linked incentive with 200 design personnel employed.
Two cautions before anyone builds a model on that table. The column is labelled investment during the year, which is investment by the supported unit, not subsidy paid out by the government, so it is not a disbursement figure. And these are projections published at the start of a financial year rather than audited outcomes, which is precisely why they carry a target tier here and why this desk will score them against whatever is eventually reported.
Even with those cautions the row is informative. Rs 4,000 crore of projected spend in a year, against a project whose announced cost is Rs 91,000 crore, is the arithmetic signature of a build that spends most of its money late, when tools arrive. Anyone expecting the capital to flow evenly across the construction period is expecting the wrong shape.
How you would know it is working
Four signals, in rough order of how much they would tell you. A published figure for support actually disbursed against this project, whether through a parliamentary answer, an annual report or a scheme disclosure, would convert the entire central layer from committed to delivered. A promoter disclosure of capital deployed to date at Dholera would do the same for the private half. Equipment orders placed with named toolmakers would show capital converting into the plant's most expensive physical asset, since tools are bought late and paid for early relative to revenue. And any disclosure of external financing would tell us whether this project is being carried on the promoter's balance sheet alone, which for a plant of this size is itself a strategic statement. Until those appear, the correct description of the capital stack is the one given here: three named layers, one of them undisclosed in its mechanics, and a fourth that is real but is not cash.