Tamil Nadu Recalibrates Data Centre Policy in Response to AI Infrastructure Demands
Tamil Nadu is undergoing a significant recalibration of its industrial policy concerning data centres, a move timed at a critical juncture. The global surge in artificial intelligence has triggered an extensive infrastructure build-out, with recent central government incentives further enhancing the appeal. However, the state, which previously incentivised the sector through tax breaks and subsidies under its 2021 data centre policy, is now allowing that policy to lapse. The government has signalled a strategic shift away from what it terms a "rat race," reflecting a broader reassessment of priorities.
Environmental and Employment Concerns Drive Policy Shift
Over the past year, Tamil Nadu's stance has evolved from celebrating substantial data centre investments to raising alarms about environmental costs and limited job creation. The core argument centres on the fact that power- and water-intensive data centres generate relatively minimal employment. AI data centres, which host graphics processing units (GPUs) for training or running large language models, demand significantly more power and water for cooling compared to traditional facilities. This issue is particularly sensitive in Chennai, where many data centre clusters are situated in areas prone to floods and water stress.
Industry Perspectives on Infrastructure Challenges
Some industry observers suggest that Tamil Nadu's new strategy might be counter-intuitive, especially as competition from states like Andhra Pradesh, Gujarat, and Maharashtra intensifies. A senior executive advising on data centre investments, who requested anonymity, highlighted that Tamil Nadu's grid infrastructure is already strained due to years of underinvestment. Despite a single-window clearance system on paper, bureaucratic delays, limited transmission capacity, and insufficient renewable energy availability remain persistent concerns.
Another analyst noted that investors are worried about the lack of firm and green power availability. AI-led racks can consume up to three times more power than traditional IT racks, with power density soaring to 30kW or more per rack for high-performance GPUs. This increased density escalates cooling requirements and infrastructure needs, reshaping how operators perceive energy infrastructure.
Economic Impact and Strategic Value Debated
Abhishek Anand, Managing Director of Insignia Policy Research, commented that the economic impact of AI data centres is still evolving. Governments can incentivise them if fiscal space permits, but such measures should not crowd out incentives for labour-intensive sectors. He advocated for easing regulatory barriers, such as allowing direct power purchases from generators and permitting private distribution players in select cities like Chennai.
In contrast, Manoj Paul, Managing Director of Equinix India, emphasised the strategic value of hosting data centres. He argued that they support industries reliant on digital infrastructure, attract anchor investments, and spur related sectors like connectivity, energy, construction, and digital services. Data centres also generate demand for equipment such as UPS systems and transformers, fostering job creation through global vendor manufacturing facilities.
Government Focus on Resource Efficiency and Innovation
State Industries Minister T R B Rajaa stated that the government is prioritising infrastructure investments that align with efficient resource use and long-term economic value. The new policy aims to nudge resource-intensive data centres towards peripheral areas with greater capacity for additional infrastructure, water, and power. Incentives will be tied to location and renewables rather than mere tax breaks. Rajaa added that Tamil Nadu is increasing its focus on innovation over storage infrastructure, citing partnerships like the one with Sarvam AI to build capability in AI research, compute, and talent.
Challenges in Decentralising Data Centres
Despite policy encouragement to develop data centres beyond Chennai, most investments remain concentrated around the city due to its strategic advantages, including proximity to end users and international cable landing stations. Infrastructure gaps elsewhere have hindered decentralisation. For instance, Larsen & Toubro's Vyoma facility in Sriperumbudur, which hosts NVIDIA GPUs and is preparing for next-generation AI processors, is expanding capacity at the same location.
Prashant Chiranjive Jain, head of the corporate centre at L&T, explained that AI training workloads are not only power-intensive but also highly variable, causing sudden swings in electricity demand that can affect grid stability. Placing data centres closer to renewable energy sources can reduce transmission losses and capital expenditure, but this must be balanced against issues like intermittent renewable power and lack of robust grid interconnectivity for contingency needs.
Manoj Paul of Equinix underscored that gigawatt-scale power supply sourced from multiple substations will be essential for future developments. This comprehensive analysis highlights Tamil Nadu's cautious yet strategic approach to balancing technological advancement with sustainable and equitable growth.
