For the first time in history, technology is not one factor among many driving economic growth. It is the foundation on which all other progress is being built. India’s semiconductor and AI choices in the next three years will determine if it can become a principal player in the economy of this century.Taiwan is indispensable to the world economy because it has mastered semiconductor fabrication. South Korea’s KOSPI returned 72 per cent in 2025, driven by Samsung and SK Hynix riding the AI chip supercycle. China has deployed an estimated $150 billion towards semiconductor self-sufficiency since 2015. The US passed the CHIPS Act — a $53 billion direct subsidy, catalysing $450 billion in private investment.AdvertisementIndia has two of the three ingredients for technological greatness: Demographic scale and intellectual capital. Our engineers and scientists are at the forefront of global AI and semiconductor development.India has historically underinvested in technological sovereignty. The risk is sharpening fast. The US government’s moves restricting access to certain AI models were a clear signal that technology denial is part of the geopolitical toolkit. An India that does not build its capability will find itself on the wrong side of that divide permanently.India Semiconductor Mission, Design Linked Incentive (DLI), and IndiaAI Mission are the right first moves. I was part of the team that built them under PM Narendra Modi’s leadership. But first moves are not destinations. The question is whether India will now make the second, smarter set of moves that turn these foundations into genuine technological momentum.AdvertisementAt SEMICON India 2026, the numbers looked impressive: Twelve manufacturing units approved, Rs 1.64 lakh crore in committed investment, five facilities in production. ISM 2.0 followed with a further Rs 1.275 lakh crore. But behind the momentum sits a structural problem that must be named: Nine of those 12 approved units are conventional ATMP/OSAT facilities — wire-bond assembly, testing, marking, and packaging. They are in the lowest-margin, most substitutable, most technologically dated segment of the value chain.Understanding the structural problem is important. The electronics PLI worked on China+1 logic: Apple and Samsung needed to diversify their manufacturing bases, PLI incentives closed the cost gap, and India delivered. India now assembles 25-28 per cent of all iPhones globally. This is one of PM Modi’s most impactful industrial policy achievements.But semiconductors are not experiencing supply-chain diversification. They are experiencing an architectural revolution. Nvidia’s data centre revenue went from $3 billion in 2020 to $47 billion in 2024. The AI era is creating a new level of abstraction in compute and value. In this environment, the opportunity ranges between 6 per cent gross margins on commodity wire-bond packaging and 50-70 per cent gross margins on IP-owning chip design.Conventional wire-bond packaging is facing relentless margin compression. Advanced packaging — CoWoS integration of GPUs and memory — commands 25-35 per cent margins. ISM 2.0 must take us there.The strategy of investment is critical: Directed at design IP, research institutions, and fab capability. India has no dedicated semiconductor research institution with the required depth. A National Semiconductor Research Institute — co-invested by government and industry, mandated to develop process technology, design IP, and the talent pipeline — was part of ISM 1.0’s vision. It must not be delayed further.The AI revolution has split the semiconductor market in two. Training AI — building the models — is consolidating around Nvidia’s CUDA ecosystem and custom ASICs from hyperscalers. That market is effectively closed.Inference is wide open. The diversity of inference requirements across cloud, edge, device, defence, agriculture, and industrial applications is so large that no single architecture can dominate. Purpose-built inference chips command 50-70 per cent gross margins. India has 1,25,000 chip design engineers qualified to build them. India’s DIR-V programme on RISC-V open-source cores eliminates ARM licensing costs. IndiaAI’s sovereign compute programme, defence procurement, 5G infrastructure, and a billion-user domestic market provide ready demand anchors.What is missing is capital that funds Indian fabless companies all the way to commercial tape-out, not just prototypes. DLI must be expanded. ISM 2.0 should create a Chip Design Commercialisation Fund at Rs 1,000-crore scale — modelled on NIIF — and designate at least two sovereign AI inference chip programmes with guaranteed government offtake. Those two changes, delivered in the next budget, would do more for India’s semiconductor future than more wire-bond packaging projects.you may likeI was part of the team that built the India Semiconductor Mission, DLI, and IndiaAI because I believe that India’s economic future in the AI age depends on building this capability. But the path of least resistance in industrial policy is to approve what can be approved quickly and measure success by investment commitments rather than by the strategic position those investments create. India has the talent, domestic demand, and a PM with the vision to back these goals. What it needs now is a direct national strategy toward design and architecture.For decades, India has missed or been a distant player in major economic and industrial supercycles. This supercycle, we do not intend to miss. The countries that write the story of this century will be those that design the chips the world runs on. India has every capability to be among them.The writer, former MP, former minister of state for Electronics & Information Technology, Government of India, is BJP Kerala state president and MLA for Nemom. Views are personal