To cash in on next tech boom, India needs the right chips

Why in the News

At SEMICON India 2026, India counted 12 approved semiconductor units, five already producing. But nine of them are basic assembly and testing plants in the lowest-margin segment, and the AI boom rewards chip design instead.

What is the chip value chain, and where do India’s units sit?

1. What it is: A chip passes through design, fabrication (etching circuits onto silicon wafers) and assembly, testing and packaging, like a book written, printed, then bound.
2. ATMP/OSAT units: Assembly, Testing, Marking and Packaging (ATMP) or Outsourced Semiconductor Assembly and Test (OSAT) plants do the final step, using dated wire-bond technology and earn about 6% gross margins.
3. Higher-value segments: Advanced packaging such as CoWoS (joining graphics processors and memory in one package) earns several times more. Chip design by firms owning the intellectual property (IP) earns the most.
4. Policy so far: The India Semiconductor Mission (ISM), the Design Linked Incentive (DLI) scheme for chip design and the IndiaAI Mission were right first moves. ISM 2.0 added Rs 1.275 lakh crore.
5. The takeaway: India has won investment in the most easily replaced segment, so ISM 2.0 must climb to packaging and design.

Why will the PLI playbook not work for chips?

1. China+1 logic: The electronics Production Linked Incentive (PLI) rewards output made in India. It worked because Apple and Samsung wanted to diversify beyond China, and incentives closed the cost gap.
2. iPhone success: India now assembles 25-28% of all iPhones worldwide.
3. Architectural revolution: AI is changing chip architecture, not just where chips are made. Eg. Nvidia’s data centre revenue grew about fifteenfold in four years.
4. Training market closed: AI training chips (used to teach models) now centre on Nvidia’s CUDA software and the largest cloud firms’ custom chips.

What do other chip powers show about state backing?

1. Taiwan: It is indispensable because it has mastered semiconductor fabrication.
2. South Korea: Its main stock index, the KOSPI, returned 72% in 2025, driven by Samsung and SK Hynix in the AI chip supercycle.
3. China: It has spent an estimated $150 billion on chip self-sufficiency since 2015.
4. US: The CHIPS Act gave a $53 billion subsidy, drawing $450 billion in private investment.

Where is India’s opening in AI chips?

1. Inference is open: Inference (running trained models to answer queries) spans cloud, devices, defence, agriculture and industry, so no single architecture can dominate.
2. High-margin niche: Purpose-built inference chips command 50-70% gross margins.
3. Talent and open cores: India has 1,25,000 chip design engineers. The DIR-V programme builds processors on open-source RISC-V designs, so Indian firms avoid paying ARM licensing costs.
4. Ready demand: IndiaAI’s sovereign compute, defence procurement, 5G and a billion-user market assure buyers.

Is approving investment the same as building capability?

1. Easy approvals: The easy path judges success by investment commitments approved, not strategic position gained.
2. Missing risk capital: No capital carries fabless firms (which design but do not make chips) to commercial tape-out, the final design sent for production.
3. Technology denial: US curbs on certain AI models show technology denial is a geopolitical tool, and India has long underinvested in technological sovereignty.

Challenges

1. Imported tools: Fabs depend on imported equipment. Eg. Dutch ASML lithography machines.
2. Utility demands: Fabs need uninterrupted power and large volumes of ultrapure water.
3. Process skills gap: India has many design engineers but few with fab process experience.

Way Forward

1. National Semiconductor Research Institute: Government and industry should co-fund an institute for process technology, design IP and talent.
2. Chip Design Commercialisation Fund: ISM 2.0 should create a Rs 1,000 crore fund modelled on the National Investment and Infrastructure Fund (NIIF), alongside an expanded DLI.
3. Sovereign inference chips: The next budget should create at least two sovereign AI inference chip programmes with guaranteed government offtake.

Conclusion

India’s chip drive has built assembly capacity but not yet a place in the design-led segments where value now lies. The marker to watch is whether the next budget funds design and inference chips rather than more low-margin packaging plants.

Key numbers

1. Investment in approved units: Rs 1.64 lakh crore committed (SEMICON India 2026).
2. Gross margins by segment: advanced packaging 25-35%; IP-owning chip design 50-70%.
3. Nvidia data centre revenue: $3 billion (2020) to $47 billion (2024).
4. India’s chip market today: about $45-50 billion.

Semiconductors in India

1. ISM framework: ISM’s Rs 76,000 crore framework offers fiscal support of up to 50% for fabs and design.
2. Market size: India’s chip market is projected to cross $100 billion by 2030.

Matching Previous Year Question

“[2026] Which of the following statements about DHRUV64 is/are correct? 1. It is the third chip fabricated under the DIR-V Programme to enable creation of microprocessors for India. 2. It is India’s first homegrown 1.0 GHz, 64-bit dual-core microprocessor. (a) 1 only (b) 2 only (c) Both 1 and 2 (d) Neither 1 nor 2 Answer: C”