October 3, 2026: Rajeev Chandrasekhar has argued in his latest column has that India has made a start in building a domestic semiconductor industry, but the next phase will have to move beyond factories and conventional chip packaging if the country wants to capture a larger share of the value created by the AI boom.
Writing in The Indian Express on October 2, in his column #SignalToNoise column for the @IndianExpress Chandrasekhar, who was part of the team that built the India Semiconductor Mission (ISM), Design Linked Incentive (DLI) scheme and IndiaAI Mission, has called for a stronger policy push towards chip design, semiconductor intellectual property, advanced packaging and AI inference hardware.
For India’s startup ecosystem, the argument is particularly relevant. The country has built a sizeable chip-design talent pool, but many of its semiconductor startups still face a difficult financing gap between developing a prototype and getting a commercial chip into production.
Rajeev Chandrasekhar argues that this is where government policy needs to change. “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,” he wrote.
His starting point is India’s existing strengths. According to Rajeev Chandrasekhar, the country has “two of the three ingredients for technological greatness: Demographic scale and intellectual capital”. Indian engineers and scientists, he says, are already playing significant roles in global AI and semiconductor development.
The third ingredient, in his argument, is technological capability that can be converted into globally competitive products and companies.
The government has made significant commitments to build that capability. At SEMICON India 2026, Chandrasekhar points to 12 approved semiconductor manufacturing units and ₹1.64 lakh crore in committed investment, with five facilities in production. The second phase of the semiconductor programme has subsequently added another ₹1.275 lakh crore of government support.
India needs to fund the next wave of chip startups: Rajeev Chandrasekhar
But Rajeev Chandrasekhar sees a problem in the composition of those investments. “Nine of those 12 approved units are conventional ATMP/OSAT facilities — wire-bond assembly, testing, marking, and packaging,” he wrote. “They are in the lowest-margin, most substitutable, most technologically dated segment of the value chain.”
The distinction matters for investors and founders watching India’s semiconductor sector. Assembly and testing are essential parts of the supply chain, but the biggest technology businesses tend to emerge where companies control proprietary designs, architectures, manufacturing processes or other forms of intellectual property.
Rajeev Chandrasekhar points to India’s electronics manufacturing success as a useful comparison. The country’s production-linked incentive programmes helped attract global manufacturers looking to diversify beyond China. Apple and Samsung expanded their manufacturing presence, while India became a major base for iPhone assembly.
“India now assembles 25-28 per cent of all iPhones globally,” he wrote, describing the expansion as “one of PM Modi’s most impactful industrial policy achievements.”
Semiconductors present a different challenge, he argues.“Semiconductors are not experiencing supply-chain diversification. They are experiencing an architectural revolution.”
That shift is visible in the economics of AI hardware. Rajeev Chandrasekhar cites Nvidia’s data-centre revenue rising from $3 billion in 2020 to $47 billion in 2024 and argues that the AI boom is moving value towards specialised compute and chip design.
His focus is particularly on AI inference, the computing required to run AI models after they have been trained.
The market for training the largest AI models has become heavily concentrated around Nvidia’s CUDA ecosystem and custom chips developed by major cloud companies, Chandrasekhar argues. Inference has a broader range of applications, from cloud computing and smartphones to defence, agriculture, industrial systems and edge devices.
“Inference is wide open,” he wrote. “The diversity of inference requirements across cloud, edge, device, defence, agriculture, and industrial applications is so large that no single architecture can dominate.”
That is potentially a significant opening for Indian fabless semiconductor startups. Rajeev Chandrasekhar says India has 1,25,000 chip-design engineers capable of working on such products. He also points to the country’s RISC-V programme, India’s sovereign compute initiatives, defence procurement, 5G infrastructure and its large domestic market as potential sources of demand.
The problem, in his view, is getting companies from engineering capability to commercial products. “What is missing is capital that funds Indian fabless companies all the way to commercial tape-out, not just prototypes,” he wrote.
A chip startup can spend years and substantial capital moving from architecture and verification to tape-out, fabrication, packaging, testing and customer qualification. Conventional venture funding can struggle with that timeline, particularly when revenue arrives much later than it does for a typical software startup.
Rajeev Chandrasekhar proposes expanding the DLI programme and creating a ₹1,000-crore Chip Design Commercialisation Fund, modelled on the National Investment and Infrastructure Fund.
He also wants ISM 2.0 to support at least two sovereign AI inference-chip programmes with government offtake. “Those two changes, delivered in the next budget, would do more for India’s semiconductor future than more wire-bond packaging projects,” he wrote.
The other gap he identifies is semiconductor research. Chandrasekhar argues that India still lacks a dedicated institution with the depth needed to work on semiconductor process technology, chip-design IP and talent development.
A National Semiconductor Research Institute, jointly backed by the government and industry, was part of the original ISM vision, he says, and should now be pursued.
For India’s deep-tech ecosystem, the policy debate is moving into a more difficult phase. Building factories is capital intensive, but building companies around proprietary semiconductor technology brings a different set of challenges: patient capital, access to fabrication, specialised research talent, anchor customers and the ability to survive several years before commercial scale.
India’s chip strategy will therefore be watched not only through the number of fabs and packaging facilities announced, but also through what happens to the country’s fabless startups and semiconductor IP companies.
How many Indian chip designs make it to commercial tape-out? How many find customers outside India? How much semiconductor IP is owned by Indian companies? And how many startups can secure the capital required to cross the gap between a working prototype and a mass-produced chip?
Those questions will become increasingly important as the global AI hardware market expands. Rajeev Chandrasekhar’s argument is that India has already established the foundations and now needs to make a more targeted bet on design and architecture.
“The countries that write the story of this century will be those that design the chips the world runs on,” he wrote. For India’s startup industry, that is the part of the semiconductor story worth watching closely.
