The answer desk
Will India make advanced chips below 28nm?
Published 25 August 2026. Facts verified to 25 August 2026 unless dated otherwise.
Not on any approved project. As of 25 August 2026 the only silicon logic fab approved under the India Semiconductor Mission is Tata Electronics' plant at Dholera, whose stated node range runs 28 to 110 nanometres, and Bloomberg reported on 17 July 2026 that it will open mostly at 90 nanometres. Nothing below 28 nanometres has been approved for fabrication in India. India already designs below that line: the Press Information Bureau recorded on 1 April 2026 that seven chips supported under the design scheme had been fabricated at various nodes including 12 nanometres, at foundries abroad.
The answer, and the line it turns on
No Indian project approved as of 25 August 2026 will fabricate chips below 28 nanometres. That is a statement about the approval record rather than about ambition, and it is checkable. Of the twelve manufacturing projects approved under the India Semiconductor Mission, exactly one is a silicon logic fab: the Tata Electronics plant at Dholera, built in technology partnership with Powerchip Semiconductor Manufacturing Corporation of Taiwan, whose stated node range runs 28 to 110 nanometres and whose approved investment is listed by the Press Information Bureau at Rs 91,526 crore DURABLE. The remaining eleven are assembly, test and packaging plants, one silicon carbide compound fab in Odisha, and one gallium nitride display fab at Dholera. None of them fabricates logic at any node, let alone below 28 nanometres.
The Dholera fab's own opening node is contested, and the dispute runs in the wrong direction for anyone hoping for advanced capability. Bloomberg reported on 17 July 2026 that the plant is expected to open mostly at 90 nanometres rather than the 28 nanometres presented at approval, with TrendForce carrying the same finding on 20 July 2026 REPORTED. A Tata spokesperson's response was that the plan had always been to start with 55 and 90 nanometres before introducing 28. Against that sits the Tata Sons chairman's letter in the annual report for the 2025 financial year, which said the company had chosen to start its chip journey at the 28 nanometre node REPORTED. This record does not resolve that contradiction, because no primary document settles it. The full evidence is at the node question. Either way, the argument is about where inside the mature range the fab opens, not about whether it goes below it.
Designed here, fabricated elsewhere
The question as usually asked conflates two different industrial capabilities, and separating them dissolves most of the argument. India already designs chips well below 28 nanometres, and has for years. What it does not do is fabricate them.
The evidence is in the government's own record. The Lok Sabha reply of 1 April 2026 published by the Press Information Bureau reports that 24 chip design projects have been approved under the design linked incentive track, that seven chips supported by it have been fabricated at various nodes including 12 nanometres, and that 211 chips have been taped out by 75 academic institutions, of which 149 were made at 180 nanometres at the Semi-Conductor Laboratory at Mohali and 62 at overseas foundries DURABLE. That last split is the whole answer in one line. When an Indian team needs an advanced node, the design goes to a foundry abroad. When it needs a legacy node, it can be made at home.
Beyond the incentive scheme sits a much larger captive design industry, where global semiconductor companies run engineering centres in India that contribute to products manufactured at the most advanced nodes in the world. Indian engineers have been doing physical design, verification and validation work on leading edge parts for many years. So the honest formulation is that India participates in advanced chips at the design layer today, and will not manufacture them at any date supported by the current approval record.
What advanced means, and why the target keeps moving
Part of the difficulty with this question is that its subject will not sit still. Advanced is a relative term with no fixed floor. The 28 nanometre generation that now marks the bottom of the advanced band was itself the leading edge when it was introduced early in the 2010s, and every process now described as legacy was once the most sophisticated manufacturing on earth. Whatever the leading edge is on the day this page is read, it sits several generations below 28 nanometres, and it will have moved again by the time any newly approved fab could be designed, built, tooled and qualified.
That has a consequence for how the question should be framed. Comparing a plant that reaches commercial production in 2028 against the most advanced fab operating in 2026 is a category error, because a fab is not built to catch a moving target; it is built to serve a product set. The useful comparison for the Dholera plant is against the other mature node capacity coming online worldwide over the same window, since that is what it will compete with for the same customers. The question of whether India will ever chase the leading edge is a different question with a different answer, and merging the two produces the familiar and unhelpful conclusion that a fab is behind before it opens.
The second consequence is about sequencing. Because the target moves, a country cannot arrive at the leading edge by aiming at where it is now. It arrives, if it arrives, by running progressively more demanding processes at yield until the gap closes, or by acquiring a partner already at the front. Neither path is available to a country that has not yet run a single commercial logic line at volume, which is the position India occupies as of 25 August 2026.
What an advanced node requires that India has not bought
The gap between a 28 nanometre fab and a leading edge one is not a matter of degree. Four requirements separate them, and none is available for purchase as a single item.
- Lithography. Below roughly 7 nanometres, patterning depends on extreme ultraviolet scanners, which are made by exactly one company in the world and shipped in very small annual quantities. Between 28 and 7 nanometres, deep ultraviolet immersion tools carry the load using multiple patterning, which trades tool cost for process complexity and cycle time. The mechanics are set out at lithography explained and the two families are compared at extreme ultraviolet against deep ultraviolet.
- Process integration experience. An advanced node is a recipe of several hundred steps whose yield is learned rather than specified. Foundries reach economic yield through years of running the same process at volume. That learning cannot be transferred as a document, which is why technology partnerships transfer mature processes first.
- An ecosystem within reach. Photomask shops, specialty gases at semiconductor grade, ultrapure chemical logistics, metrology services and a repair supply chain for the tools. India is beginning to assemble the first rungs of that around the approved projects, and the record of it is tracked at the ASML file and across this desk's supplier coverage.
- Customers who need the node. Advanced nodes exist to serve high volume, high value products: smartphone application processors, graphics and artificial intelligence accelerators, high end central processors. A fab without those customers has no path to filling capacity, and an unfilled advanced fab loses money faster than any other kind of factory.
Capital is the constraint people name, and it is real but secondary. The Dholera fab's Rs 91,526 crore approval is a very large figure by Indian industrial standards and a moderate one measured against the leading edge fab projects announced elsewhere in this decade, which routinely run to multiples of it. Money can be raised. The four requirements above take years regardless of it.
A fifth requirement sits outside the factory gate entirely. Extreme ultraviolet scanners are built by one company, in the Netherlands, from a supply chain concentrated across a handful of allied countries, and their export is licensed by governments rather than settled between buyer and seller. An advanced node fab is therefore a diplomatic transaction as well as a commercial one, and the licensing position of the destination country matters as much as its balance sheet. India's position in that system is friendly and untested, since the question has never arisen for an Indian buyer. It would arise the moment an Indian project specified a tool of that class, and the first public evidence of such a project would most likely appear in export licensing and supplier disclosure rather than in an Indian announcement.
What the ASML agreement tells you about the ceiling
The most precise public evidence about the Dholera fab's technical ceiling is the equipment agreement. On 16 May 2026, ASML and Tata Electronics both announced a memorandum of understanding through their own newsrooms, naming nodes of 28, 40, 55, 90 and 110 nanometres REPORTED. That list is entirely deep ultraviolet territory. No extreme ultraviolet tool is named, and none would be needed for any node on that list.
Reading it as a limitation misses what it is. A tool set matched to a node range is what a competently specified fab looks like. Buying extreme ultraviolet capacity for a plant whose product set is power management chips, display drivers and microcontrollers would be a capital error, not an ambition. The agreement is also a memorandum rather than a purchase order, which is why this record tiers it as reported rather than durable and continues to watch for a valued equipment order, the single most informative disclosure still outstanding on this project.
What the second phase of the mission does and does not promise
India Semiconductor Mission 2.0 was announced in the Union Budget for 2026-27 and is directed at semiconductor equipment and materials manufacturing, full stack Indian intellectual property and supply chain resilience DURABLE. Read plainly, that is a programme aimed at the layers around fabrication rather than at a more advanced fab. Equipment and materials are where India currently imports almost everything, and where a domestic base would reduce a genuine structural dependency. Nothing in the announced scope commits to a sub-28 nanometre logic fab, and this record has found no approval, no memorandum and no minister's statement that does. The programme file is at ISM 2.0.
Statements of general intent about advanced manufacturing do circulate, and they are worth separating by tier. Ambition expressed in a speech is not a target with a date, and a target with a date is not an approved project with a capacity and a rupee figure. Only the third category appears in this record's ledger. On that test, the sub-28 nanometre question currently has nothing in any of the three columns.
The realistic sequence, and the objection to it
The countries that built semiconductor industries from a standing start have generally followed the same order: packaging first, mature node fabrication second, advanced node fabrication third if at all, with design capability developing alongside. The reason is not caution. It is that each rung teaches the workforce, the supplier base and the regulators what the next rung needs, and each rung is financeable on shorter payback than the one above it. India's current portfolio maps onto that sequence precisely, with nine packaging plants, one mature logic fab and two compound or display fabs approved.
The serious objection to the sequence is timing. Mature node capacity is being added in several countries over the same window, and a fab that reaches volume in 2028 enters a market that has been preparing for it. If mature node pricing compresses, the economics that justify starting at the bottom rung weaken, and the argument for skipping ahead gets louder. The counter is that the rungs are not optional: a country that has never run a 90 nanometre line at yield cannot run a 5 nanometre one, whatever the market is doing. The demand side of that argument is examined at why mature nodes.
What would change this answer
This page would be rewritten by any of four events, in ascending order of significance. A Cabinet approval for a logic fab at a node below 28 nanometres, which would be the first hard commitment of its kind. An extreme ultraviolet tool export licence or purchase agreement involving an Indian buyer, which would be visible in supplier disclosures and in export control filings long before any building went up. A published technology transfer agreement covering a FinFET or gate all around process, which is the process architecture question underneath the node number. And a customer commitment from a company that buys advanced node wafers at volume, since no one builds an advanced fab speculatively twice.
Absent those, the answer holds. India will make chips at 28 nanometres and above, at one fab, from a date guided to mid-2028 for commercial production TARGET, and it will continue to design chips below that line and have them manufactured abroad. Anyone who reads that as failure is measuring against the wrong benchmark. Anyone who reads the approval of a mature node fab as the arrival of advanced manufacturing is measuring against no benchmark at all.