India Is Building Its Own Chips. The Ceiling Is in Washington.

On February 16, 2026, India launched Semiconductor Mission 2.0, a program framed around "sovereign capabilities" and a strategic shift from chip design to full-scale manufacturing ^[1]. Three days later, on February 19, India became the tenth member of Pax Silica, the U.S.-led alliance designed to secure the entire silicon stack — from critical minerals to fabrication to AI deployment — and reduce global dependence on China ^[2]. The sovereignty mission and the dependency bloc did not arrive in sequence, one giving way to the other. They arrived in the same week, as a simultaneous bet. The contradiction was present at the moment of accession. At the design layer, Indian sovereignty is real and accelerating. On May 28, C2i Semiconductors taped out an AI power chip "designed end-to-end in India," backed by the government's Design Linked Incentive scheme, which now supports 24 projects across 105 companies ^[3]. The same week, NITI Aayog released a report calling for a $120–150 billion domestic semiconductor value chain by 2035 ^[4]. Vice Chairman Lahiri made the premise explicit.

And that sovereignty must begin at the infrastructure layer. — Ashok Kumar Lahiri

In late July, Semicon 2.0 launched with a Rs 1,27,500 crore outlay explicitly targeting Indian-owned intellectual property. IT Secretary Maya Sundarakrishnan said the goal was products "designed, owned, and commercialised from India," and the India Cellular and Electronics Association chairman called for "creating Indian-owned semiconductor products, IP and globally competitive technology companies" ^[5]. The sovereignty push did not end with Pax Silica. It intensified after it. At the packaging layer, the picture is more modest but still real. On February 26, a week after accession, Modi inaugurated Micron's $2.75 billion ATMP plant in Sanand, Gujarat — a facility for assembly, testing, marking, and packaging, not wafer fabrication — and linked it directly to Pax Silica ^[6]. U.S. Ambassador Gor called it India's entry "into the global semiconductor supply chain as a manufacturing nation" ^[6]. Assembly and packaging employs people and builds industrial muscle, but it is also the lowest-value segment of the chain. India was making things. It was not yet making the thing that determines how advanced those things can be. That thing is the lithography machine, and here the contradiction sharpens. On May 16, three months after joining Pax Silica, Tata Electronics signed an $11 billion memorandum of understanding with ASML to build India's first commercial 300mm fab at Dholera ^[7]. The deal delivers ASML lithography tools — the machines that etch circuits onto silicon wafers — but the technology transfer comes via Taiwan's Powerchip at what officials described as "mature process nodes," not cutting-edge extreme ultraviolet lithography ^[7]. EUV is the gate: without it, a fab cannot print circuits at the most advanced densities. India's own roadmap, announced by Minister Vaishnaw in January, targets 7nm by 2030 and 3nm by 2032 ^[8]. That 3nm node requires EUV. There is no Russian EUV, no Japanese EUV — ASML is the sole maker of the advanced machines, and no Indian program has produced a domestic alternative. And ASML's EUV sales are governed by the same U.S. export-control apparatus now being weaponized against China. In June, Commerce Secretary Lutnick accused ASML of illegally shipping EUV technology to China, and Congress began weighing a bill to subject ASML to U.S.-style export restrictions ^[9]. The mechanism is straightforward: export controls condition ASML's EUV sales, EUV gates the process node, and the process node gates India's most advanced ambition. India's semiconductor ceiling is set in Washington and Eindhoven, not in Dholera. India appears to know this. The hedging is methodical. On May 12, India was in advanced talks with Russia for a critical minerals pact covering lithium and rare earths ^[10]. Two weeks later, on May 26, it signed a bilateral critical minerals pact with the United States, explicitly framed as reducing "strategic dependence on China-dominated sources" ^[11]. The sequencing was not accidental. India has also signed minerals deals with Argentina, Australia, and Japan, and is in talks with Peru and Chile ^[10]. In April, India and Japan launched an AI Strategic Dialogue covering the "entire AI stack" including hardware, merging Indian talent with Japanese "hardware precision and industrial expertise" ^[12]. And in June, India formally rejected USTR's Section 301 findings on industrial overcapacity. DGTR Amit Singhal was blunt.

Overcapacity is a country’s perspective. We don’t think we have overcapacity in anything. — Amit Singhal

India is spreading its bets across multiple technology partners, pushing back against U.S. trade pressure even from inside the bloc, and positioning itself — in the words of experts at an April summit — as a "trusted neutral balancer" and "global anchor for stability" ^[13]. The contradiction is not that India surrendered sovereignty. It did not. The C2i chip was designed in India. The Dholera fab will make chips in India. The NITI Aayog plan envisions a $150 billion domestic value chain. The contradiction is that every one of these ambitions, at its most advanced edge, runs through a gate someone else controls. The 3nm roadmap requires a machine sold by a Dutch company whose most advanced tools are subject to American permission. India is building a semiconductor ecosystem that is real at the design and packaging layers, bounded at the equipment layer, and hedged at the mineral layer — because India can see the ceiling, even if it does not say so.