There is a number buried in an ITIF report from February 2024 that no government press release will mention. The predicted outcome of India's $10 billion semiconductor subsidy programme, by 2029: five modest fabrication facilities, with the most sophisticated output being a 28 nanometre chip. 28nm was cutting-edge in 2011. By 2026 it is mainstream automotive technology.
Governments survive on announcements, companies thrive on subsidies, technology sovereignty has no electoral dividend and no quarterly earnings.
By 2029, when this will be India's best chip, the world will be manufacturing at 2nm. That is not a technology gap. It is a strategic vulnerability dressed up as an industrial achievement. To understand how this happened, look at where the ₹76,000 crore actually went.
The India Semiconductor Mission framework allocates 50% capital subsidy — meaning the government pays half the cost of land, buildings, and machinery for any qualifying company. Research and development received 2. 5% of the total budget. The Design Linked Incentive scheme, which could have built actual Indian intellectual property and capability, was underfunded.
Its initial application window was 45 days. "Governments survive on announcements. Companies thrive on subsidies. Technology sovereignty has no electoral dividend and no quarterly earnings.
That is why R&D gets 2. 5%." The incentive structure explains the outcome. A government's interest is in ribbon-cutting ceremonies, inauguration photographs, and milestones visible before the next election cycle.
A private company's interest is maximum subsidy, minimum risk, and quick returns. Technology sovereignty — the kind where India owns the knowledge and no foreign company can take it away — sits between these two interests and serves neither. It does not photograph well. It does not produce quarterly earnings.
So it receives 2. 5% of the budget. When a foreign company receives 70% subsidy to build a packaging plant, retains all its core intellectual property outside India, and can choose to leave tomorrow, what does India actually own? An empty building.
This is not sovereignty. It is the appearance of sovereignty, subsidised by the taxpayer. Tata's Dholera fab — which will be India's first actual wafer fabrication facility — will produce chips in the 28nm to 110nm range. This is relevant for automotive and industrial applications.
It is not AI-grade. It is not advanced consumer electronics. It is not military grade. The pathway exists, which matters.
But that pathway does not lead toward 2nm — at least not within any currently funded framework. "By 2029, India will need 3nm and 2nm chips to run an AI-driven economy. We will be dependent on Taiwan for all of them. If anything happens in the Taiwan Strait, our economy stops, our military stops, our internet could stop."
The government's defenders make a fair point: no country enters semiconductor manufacturing overnight. TSMC took two decades and billions of dollars. Intel is still struggling with its own manufacturing transitions. India is building an ecosystem, and early partnerships with Tata-PSMC and the Micron facility are genuine starting points.
CG Semi has delivered some made-in-India chips. PM Modi inaugurated KSEDC's plant in March 2026. These are real steps. But a pathway being announced is different from a pathway being funded.
The data shows that the scheme capable of building actual Indian IP is underfunded. The first round of applications from all three wafer fab applicants was rejected and had to be resubmitted. The most advanced output by 2029 will be 28nm. And by 2030, India's semiconductor market is projected to reach $100 to $110 billion — of which 90% will still be imported.
India will be a buyer in its own market. ₹76,000 crore was spent. The risk it was supposed to eliminate — complete dependence on foreign chips during a geopolitical crisis — remains exactly where it was. What was purchased instead were photographs of inauguration ceremonies and a packaging plant that wraps chips made in Taiwan.
The final question is one the transcript leaves deliberately open: is this an honest policy mistake — the natural stumbling of a country attempting a genuinely difficult industrial transition for the first time? Or is it something more deliberate — subsidy engineering that transfers public money into private facilities while using the language of sovereignty to prevent the public from asking where the technology actually went? That question belongs to every taxpayer who funded it.
