India’s ₹12 trillion data centre gamble may not be enough

India’s artificial intelligence (AI) ambitions are taking the shape of a digital warehouse boom. While semiconductor chips, models, and computing platforms capture most of the value globally, domestic capital is chasing the lower-risk, lower-value layer of data centres.

But experts doubt that this boom can solve India’s deeper challenge of building high-value technology capabilities and creating enough quality jobs to compete with global peers.

There are 208 data centre projects in the pipeline, with planned investments of ₹12.3 trillion, according to data from the economic think tank Centre for Monitoring Indian Economy (CMIE). More than 60% of this investment, or about ₹7.6 trillion, was announced in FY26 alone, marking a fivefold jump in AI-linked infrastructure spending commitment between FY25 and FY26.

Maharashtra leads with outstanding projects worth ₹3.9 trillion, followed by Uttar Pradesh at ₹2.54 trillion, driven by Am AI Labs’ ₹2.27 trillion Greater Noida Hyperscale Hub announced earlier this year. It is the largest private AI investment in India, with completion targeted by December 2030.

India operates roughly 1.5GW of data centre capacity, but pipeline projects aim to add nearly 20GW by 2030. There are 40 upcoming hyperscale projects exceeding 100MW each, alongside eight projects of at least 1GW, according to CMIE data. For perspective, the world’s largest operational data centre has a capacity of roughly 650MW.

Limited manpower

Even as India Inc. rushes into data-centres, market experts remain unconvinced. Despite being capital intensive, hyperscale data centres create relatively fewer jobs per unit of capital deployed and do not really address India’s broader employment challenges, noted Sorbh Gupta, head of equity at Bajaj Finserv Asset Management.

“One reason why FPIs (Foreign Portfolio Investors) have turned cautious is the perception that AI could disrupt India’s IT services and white-collar outsourcing—the sectors that powers India’s middle-class expansion,” Gupta said. “Data centre investments alone won’t solve that concern.”

A 1GW hyperscale data-centre campus typically creates around 200–300 permanent jobs such as data-centre technicians, critical facilities engineers and security staff, alongside 700–2,000 temporary construction roles during the buildout phase, according to a November 2025 study by the Hamm Institute for American Energy and the Zage Business of Energy Initiative.

Although, at an event last week, Union minister of state for science and technology Jitendra Singh said India’s data-centre push could generate nearly one lakh engineering jobs over the next four years across AI systems, cooling technologies, smart grids, renewable-energy integration and digital infrastructure.

To be sure, countries like Singapore, China and UAE also saw rapid data-centre expansion during earlier phases of their AI buildout. But those investments were also accompanied by growth in semiconductors, cloud ecosystems, AI models and digital platforms, allowing them to move into higher-value parts of the AI chain.

Semiconductor slump

Meanwhile, domestic semiconductor project announcements fell by 94% over two years. After peaking at ₹2.4 trillion in FY23 and ₹2.3 trillion in FY24, planned investments fell to just ₹19,600 crore across nine projects in FY26, CMIE data showed.

The decline reflects the India Semiconductor Mission (ISM) ‘s changing lifecycle. Launched in 2021 with a ₹76,000 crore outlay and hefty production incentives, the scheme struggled as corporate execution lagged government ambition, noted Amit Khanna, partner at Grant Thornton Bharat.

“Indian firms are typically reluctant to commit to capital-intensive sectors with complex technological know-how, long gestation periods, and rapid obsolescence,” Khanna said. “An ecosystem requires decades to build, and when India rolled out its chip policy, Taiwan, South Korea, and China were already well ahead.”

Since ISM’s launch, 85 semiconductor projects have been announced. Only 10 are complete, 16 are under implementation, and over two-thirds remain stalled or haven’t broken ground. Instead, the sector is shifting toward less ambitious parts of the value chain—assembly, testing, packaging, and mature-node manufacturing under ISM 2.0.

The ₹91,530 crore Tata Electronics fab in Dholera focuses on making 28nm to 110nm chips for automotive and industrial uses. Micron’s Gujarat facility is an assembly and testing plant, reinforcing India’s push toward packaging and mature nodes over cutting-edge AI processors.

Shobhit Agarwal, CEO of Anarock Capital, argued the latest chip strategy is more realistic, reducing import dependence while integrating India into the global ecosystem at a limited capacity.

Data centres for the win

Unlike chip fabs, data centres involve lower technological complexity, predictable cash flows and significantly lower execution risk, said Grant Thornton Bharat’s Khanna. “India’s multilingual, data-rich ecosystem is driving demand for AI training infrastructure, especially as governments and companies push for data localization. This is accelerating investments in data centres over chip units,” he added.

That distinction defines India’s AI strategy. Countries dominating the AI rally, like Taiwan, South Korea, and the US, control the high-value layers of the ecosystem: advanced chips, frontier models, and platforms. India is rapidly building the layer hosting those systems.

“We (India) spend on land, power, water, imported GPUs (graphics processing units), and cloud stacks, while richer margins flow to global chipmakers and model owners,” said Anarock’s Agarwal.

Limits of the boom

Experts warn that data centres are capital-intensive and heavily reliant on uninterrupted power supply, even as India grapples with surging power demand. Peak demand hit a record 260.45GW last week amid severe heatwaves across North and Central India.

“A data centre is essentially a power asset with a building around it,” Agarwal said. “The real risk is whether India can provide reliable, affordable, low-carbon, 24/7 power at specific grid nodes.”

This matters because the infrastructure boom arrives as India grapples with weak industrial depth and fragile energy systems, said Nitin Bhasin, head of institutional equities at Ambit Capital. Over the past decade, India has seen unparalleled financial innovation but has failed to build equivalent strength in semiconductors, advanced manufacturing, and frontier ecosystems, he noted.

“Hence, global capital has gravitated toward economies with stronger AI competitive advantages. Data centres alone won’t make India a compelling AI investment theme for foreign investors,” he added.