​​​​​​​India’s semiconductor strategy needs to move beyond chip packaging, says Rajeev Chandrasekhar

India needs to shift its semiconductor strategy from conventional chip assembly and packaging towards chip design, advanced packaging, fabrication and research, BJP leader and former minister of state for Electronics & Information Technology Rajeev Chandrasekhar has argued.

In an article, Chandrasekhar said the next phase of India’s technology ambitions would depend on building higher-value capabilities as artificial intelligence reshapes the global semiconductor industry.

Chandrasekhar said India’s semiconductor programme has made significant progress, pointing to 12 approved manufacturing units, ₹1.64 lakh crore in committed investment and five facilities in production. He also cited the subsequent India Semiconductor Mission 2.0, which he said involved another ₹1.275 lakh crore.

However, he argued that the composition of the investments matters as much as their headline value. “Nine of those 12 approved units are conventional ATMP/OSAT facilities — wire-bond assembly, testing, marking, and packaging,” he wrote.

According to Chandrasekhar, conventional packaging sits towards the lower-margin end of the semiconductor value chain and faces increasing competition. He argued that India should instead build capabilities in chip design, intellectual property, advanced packaging and fabrication.

AI is changing the opportunity

Chandrasekhar contrasted India’s electronics manufacturing success with the structural changes underway in semiconductors.

He said the electronics production-linked incentive programme benefited from the “China+1” strategy as companies diversified manufacturing. India now assembles 25-28% of iPhones globally, according to his article.

But he argued that semiconductors are undergoing a different transformation, driven by AI computing. “Semiconductors are not experiencing supply-chain diversification. They are experiencing an architectural revolution,” Chandrasekhar wrote.

He pointed to the sharp increase in Nvidia’s data-centre revenue as an indication of the scale of the AI chip opportunity. In his view, this shift makes semiconductor design and intellectual property increasingly important to the value captured by companies. Inference chips as an opening

Chandrasekhar identified AI inference — the process of running trained AI models — as an area where he believes India could develop a stronger position.

He argued that while AI training is increasingly concentrated around Nvidia’s CUDA ecosystem and custom chips developed by hyperscalers, inference requirements are more diverse across cloud computing, edge devices, defence, agriculture and industrial applications. “Inference is wide open,” he wrote.

Chandrasekhar said India has 1.25 lakh chip-design engineers and could use open-source RISC-V technology through the DIR-V programme to develop specialised chips without relying on ARM licensing.

Calls for research and commercial capital

He also called for a dedicated National Semiconductor Research Institute focused on process technology, design IP and talent development.

Another proposal is a ₹1,000-crore Chip Design Commercialisation Fund, modelled on the National Investment and Infrastructure Fund, to help Indian fabless companies finance development through commercial tape-out.

Chandrasekhar also proposed at least two sovereign AI inference-chip programmes with guaranteed government offtake.

“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,” he wrote.

He argued that India’s next semiconductor phase should therefore focus on building technological capability rather than simply expanding the number of approved facilities.