AI Data Centres to Drive 26.3 GW Power Demand by FY32, Government Tells Parliament

India’s AI data centre boom is set to put unprecedented pressure on the national power grid. In a written reply to Parliament on Monday, Minister of State for Power Shripad Naik disclosed that the government now expects artificial intelligence data centres to consume 26.3 gigawatts (GW) of electricity by the end of fiscal year 2031-32. This is nearly double the 13.56 GW estimate that the Ministry of Power had released just four months ago, in March 2025.

The revised projection reflects the accelerating pace of AI infrastructure buildout by both Indian conglomerates and global tech giants. Naik said the additional load would be “integrated into the grid and primarily served by renewable energy capacity.” The overall peak electricity demand is expected to cross 388 GW by March 2032, making the data centre share roughly 6.8% of the total.

Datos clave

Metric	Value
Projected AI data centre power demand by FY32	3 GW
Peak electricity demand by March 2032	388 GW
Previous estimate (March 2025)	56 GW
Primary energy source for new load	Renewable energy

Why the projection doubled

The sharp upward revision is attributed to the rapid scaling of AI workloads, which require far more computational power than traditional cloud applications. Since March, several major announcements have been made: the Adani Group partnered with Jabil Inc to build vertically integrated data centre infrastructure; Reliance Industries committed up to ₹10 lakh crore in AI over seven years; and HCLTech signed a memorandum of understanding with Sarvam AI to set up a $1.5 billion AI data centre in Odisha. Global players have also accelerated plans. Google began construction on a $15 billion AI data centre hub in Visakhapatnam, while Microsoft pledged $3 billion to expand its cloud and AI infrastructure in India.

Naik noted that the ministry is building transmission infrastructure in a phased manner and adding capacity, including renewable energy, to address the growing demand. Discussions with industry stakeholders are ongoing, according to a report by The Indian Express, covering reliable power supply, transmission embedding, and grid balancing to handle the spiky load from data centres. Another topic under discussion is locating data centres away from urban centres to avoid transmission congestion.

Challenges and local resistance

Despite the aggressive expansion, operators face significant hurdles: high power consumption, limited access to reliable renewable energy, land acquisition delays, cooling requirements, and rising capital costs. AI workloads are making data centres significantly more energy-intensive, raising questions about grid readiness and sustainability.

Local opposition is also emerging. In Maharashtra’s Thane district, residents are protesting against Amazon’s proposed 422 MW data centre over concerns about noise pollution from backup diesel generators, high electricity demand, and tree felling. The protests highlight the growing tension between India’s digital ambitions and the environmental and social costs of infrastructure expansion.

What this means for the Indian AI ecosystem

The 26.3 GW figure is a critical benchmark for startups, investors, and policy planners. Data centre power availability and cost will directly affect the viability of AI training and inference in India. If renewable energy integration falters, operators may face higher operational costs and regulatory pressure. The government’s emphasis on renewables suggests a strategic push to align AI infrastructure with India’s climate commitments, but the pace of grid upgrades and land acquisition will determine whether the target is achievable.

For Indian tech companies and digital marketers, the news underscores the importance of energy efficiency in AI operations. It also signals that the government is taking AI infrastructure seriously, which could lead to policy incentives or faster clearances – but also stricter environmental scrutiny.

Source: Inc42 – AI Data Centres To Consumer 26.3 GW Power By FY32: MoS Power