Why doesn't India build a leading-edge logic processor?

India imports 90-95% of the semiconductors it uses, and for the most demanding chips of all — the leading-edge logic that runs AI accelerators, GPUs and advanced mobile processors — it produces none.

The barrier is not ambition but physics and monopoly. A leading-edge logic node, below 7 nanometres, requires extreme ultraviolet lithography, a technology sold by a single firm, ASML. It requires design tools from Synopsys, Cadence and Siemens with no indigenous substitute, ultra-pure process chemicals, 300mm silicon wafers from Shin-Etsu and SUMCO, and a workforce of yield specialists that takes years to build. Fewer than a handful of nations cross this bar: TSMC alone holds roughly 70% of global foundry revenue as of 2025.

India's build-out is real but aimed lower. As of May 2026, twelve semiconductor projects worth about ₹1.64 lakh crore stand approved under the India Semiconductor Mission. The flagship is Tata Electronics' Dholera fab, built with Taiwan's Powerchip (PSMC), a ₹91,000 crore project backed by 50% government support, targeting first silicon in December 2026. Its planned nodes are 28 to 110 nanometres — mature, not leading-edge. Downstream, the packaging and test layer has come alive: Micron's memory ATMP plant at Sanand, inaugurated 28 February 2026, was the mission's first operational facility; Kaynes Semicon shipped India's first commercially produced chips in October 2025 and reached full production in March 2026; CG Power's OSAT line, with Renesas, launched in August 2025.

Design capability, meanwhile, is genuine. Around 20% of the world's chip design talent is based in India. IIT Madras and ISRO booted the IRIS RISC-V controller — fabricated at SCL Chandigarh — in February 2025, and the fully indigenous VIKRAM 3201 space-grade microprocessor was unveiled at Semicon India 2025. Renesas opened 3nm design centres in Noida and Bengaluru in May 2025. But these are designs and low-volume aerospace parts, not foundry-scale production.

The gap persists because leading-edge logic sits atop a stack of critical capabilities India does not yet hold: EUV lithography, advanced manufacturing equipment, EDA software, wafer substrates, specialty chemicals, and CPU instruction-set IP. Each is dominated by a small set of foreign suppliers; a fab is only the visible tip.

Closing it means building that stack, not just the fab — and doing so on a timeline that analysts place, at the earliest, beyond 2030. The near-term path India has chosen is mature nodes, packaging and design, where the foundations are more attainable and the strategic return arrives sooner.