Authored by: Varun B.R., Manager – Power Sector, Vasudha Foundation
Electrification and digitalisation are reshaping electricity demand across economic sectors. The power grid is at the centre of this transition, determining how effectively new loads can be accommodated while maintaining reliability, affordability and progress towards a clean energy future. India’s rapidly expanding digital economy, mirroring the global AI boom, is creating a new dimension to this challenge of accommodating the growing power requirements of data centres without undermining renewable energy integration.
To illustrate the scale of this growth, India’s installed data-centre power capacity increased from around 375 MW in 2020 to 1.57 GW as of August 2026, with the government projecting it to reach nearly 26.3 GW by 2031–32. This rapid expansion is accompanied by a change in the nature of electricity demand. AI-training racks can require between 100 and 150 kW, compared with approximately 5–15 kW for conventional general-purpose data-centre racks. Beyond the scale of demand, data centres require a highly reliable, continuous power supply, while renewable energy generation varies by time and location.
The announcement of a planned 1 GW hyperscale data-centre campus in Visakhapatnam, Andhra Pradesh, brings this challenge into sharper focus. The question is no longer limited to whether sufficient electricity can be procured to serve these emerging loads. It is also about how the associated generation, transmission, storage and flexibility requirements will be managed, and who will bear the costs of integrating them into the grid.
Additionality is often broached as a regulatory instrument to address this concern. In this context, additionality refers to the principle that electricity procurement by new or expanding consumers should lead to the development of new renewable energy capacity that would not otherwise have been established. However, while additionality can support the development of new clean energy capacity, a rigid mandate may not, by itself, address the wider grid-management challenges emerging in the Indian context.
Policy Support vis-à-vis AI Infrastructure
Government support signals a clear push to expand India’s AI and digital infrastructure. Data centres were accorded infrastructure status under the Union Budget 2022–23, facilitating access to long-term financing. The Union Budget FY2026–27 has further introduced a tax holiday until 2047 for eligible foreign cloud service providers undertaking global operations through data-centre infrastructure located in India. Allied initiatives, including the Semicon India Programme and the Electronics Component Manufacturing Scheme, also support the development of the broader semiconductor and electronics supply chain.
These measures are important for positioning India as a viable destination for hyperscalers and AI infrastructure investments. However, the expansion of such infrastructure will also increase the importance of power availability, grid connectivity and reliable clean electricity procurement.
This creates a policy challenge. On the one hand, India is actively incentivising investments in AI infrastructure. On the other, the power sector must ensure that the resulting electricity demand does not compromise renewable energy integration, system reliability or the affordability of electricity for other consumers. The regulatory approach must therefore account for both objectives rather than treating power procurement as an isolated commercial arrangement between data centres and electricity suppliers.
Contextualising Additionality: From Capacity Addition to Grid Integration
India’s clean energy transition has made significant progress, with installed non-fossil fuel capacity crossing 300 GW as of July 2026. However, the challenge is no longer capacity addition alone; it is increasingly about integrating this capacity into the power system. Renewable energy evacuation constraints, the pace of storage deployment and the availability of sufficient transmission capacity can limit the extent to which new renewable generation is utilised effectively.
The distinction is particularly important for AI data centres. An additionality requirement that focuses solely on procuring new renewable capacity does not necessarily guarantee that electricity will be available when the data centre needs it. For instance, a megawatt of solar capacity cannot provide a continuous megawatt of power through the night. A data centre’s round-the-clock requirements may therefore necessitate a combination of renewable generation, energy storage, other sources of firm power and appropriate grid infrastructure. Additionality can help stimulate new renewable energy investments, but it does not automatically resolve temporal mismatches between generation and consumption, transmission bottlenecks or the need for firm power. These considerations become even more important when a large load is concentrated at a particular grid location.
Further, an inflexible additionality mandate could create unintended consequences. If new renewable energy capacity and storage remain constrained, large, capital-rich data-centre operators may be better positioned to secure available clean power than smaller industrial consumers looking to electrify their processes. This could increase procurement costs or constrain access to clean power for other sectors, potentially slowing their transition away from fossil fuels. The extent of this risk will depend on how quickly new capacity is added, how grid constraints are managed and how the mandate is designed. The policy question, therefore, is not whether additionality should be pursued, but whether it should be imposed as a standalone requirement or integrated into a broader framework that accounts for the power sector ecosystem as a whole.
Glide-Path Strategies: Aligning New Loads with Clean Power Addition
International approaches provide useful lessons on how additionality can be designed to balance investment timelines with clean energy objectives. The key distinction is not simply the renewable energy share prescribed, but also the timeline for compliance, the definition of eligible generation and the treatment of grid constraints.
In Ireland, the Commission for Regulation of Utilities’ December 2025 decision requires new data centres to meet at least 80% of their annual electricity demand through additional renewable electricity projects generating in Ireland. Developers are allowed a six-year glide path to develop and commission the associated renewable generation. The framework also considers local or on-site generation and storage requirements, as well as the location and constraints of the electricity network.
In Spain, a draft decree released for public consultation in August 2026 proposed that data centres source at least 80% of their electricity consumption from newly established renewable generation, with the requirement assessed on an hourly basis. The proposal links renewable generation to the 18 months preceding the data centre’s entry into operation. This remains a proposed regulatory approach and should not be presented as an enacted requirement.
At the state level in India, Andhra Pradesh has announced a requirement for at least 70% of the energy used by data centres to come from renewable sources. While this signals an intent to align digital infrastructure growth with clean energy, the implementation framework will need to clarify the treatment of additional generation, the applicable timelines and how the requirement will interact with grid reliability.
These examples illustrate that additionality does not have to be designed as an immediate, one-size-fits-all obligation. A phased approach can provide developers with time to commission new renewable capacity while allowing system operators to plan the required network and firming infrastructure. At the same time, the experience of Ireland highlights the importance of connecting renewable energy requirements with location-specific grid conditions. For India, the objective should be to adapt these principles to its own electricity market, renewable energy profile and grid constraints, rather than transplanting a percentage-based mandate without accounting for local conditions.
An Approach for India: Governing AI Demand Beyond Additionality
A suitable regulatory framework for India should be built around three principles: supporting renewable energy integration, managing new demand in line with grid conditions and ensuring that the costs of accommodating large loads are allocated transparently. The following approaches could help operationalise these principles.
1. Link data-centre connections to grid readiness and system costs: The first step is to assess the implications of new data-centre loads at the point of connection. Large data centres should be required to provide phased demand forecasts, expected load profiles and credible power procurement plans. Distribution licensees and system operators should assess these requirements against the availability of transmission and distribution capacity, local network constraints and the need for additional grid infrastructure.
Where new investments in network capacity are required, the associated costs should be identified transparently and allocated through a clearly defined regulatory framework. Connection planning should also account for the timelines required to commission supporting infrastructure, rather than allowing data-centre capacity additions and grid development to proceed on disconnected schedules. This would help ensure that new AI loads are accommodated where the grid can support them or where the required investments can be delivered in a planned manner.
2. Move from a standalone additionality mandate to a firm clean-power requirement: Additionality should remain an important instrument for ensuring that new demand contributes to new renewable energy development. However, compliance should be linked to the ability to provide reliable clean electricity, rather than relying solely on a megawatt-for-megawatt addition of renewable capacity.
For large data centres, a phased clean-power obligation could combine additional renewable energy procurement with storage and other appropriate firming solutions. Depending on the state’s resource profile and grid conditions, these could include battery energy storage systems, pumped-storage hydropower and other sources of firm, low-carbon electricity.
The requirement should progressively account for the temporal alignment of renewable generation with consumption, supported by verifiable procurement and generation data. Storage should complement additional renewable generation by shifting energy across time, rather than being treated as a substitute for new clean energy capacity. A time-bound glide path, with clearly defined interim milestones, would allow data-centre operators to develop the required portfolio while giving utilities and regulators greater visibility into future system requirements.
3. Introduce time- and location-sensitive incentives for flexible demand: Electricity tariffs and demand-response mechanisms can help align consumption with periods of high renewable energy generation and reduce pressure on constrained parts of the grid. Rather than imposing a blanket requirement to consume solar power during daytime hours, regulators should explore mechanisms that reward data centres for shifting workloads where technically feasible.
For instance, non-urgent AI-training and other batch-processing workloads may be scheduled during periods of high renewable generation or surplus electricity, while time-critical services and latency-sensitive workloads continue to operate according to their reliability requirements. Time-varying tariffs, demand-response contracts and appropriate incentives can help make this flexibility commercially viable. Such mechanisms could improve renewable energy utilisation, including in situations where generation would otherwise be curtailed, provided that the relevant grid location and network conditions allow the additional demand to be served. They could also reduce the need for expensive peak-period power procurement and support more efficient use of existing grid infrastructure. Large data centres could additionally contribute to the cost of system strengthening and firming through transparent, cost-reflective charges where justified by the associated investments. Any such mechanism should be based on demonstrable system costs rather than an arbitrary tariff premium, with the objective of ensuring that the costs of accommodating new demand are not inadvertently shifted to other consumers.
Conclusion
As data-centre capacities ramp up, India needs a regulatory framework that accommodates AI-driven electricity demand while maintaining progress towards renewable energy integration. Additionality can play an important role, but it should not be treated as a standalone solution to challenges that extend beyond renewable capacity addition. The objective should be to ensure that AI infrastructure growth contributes to, rather than competes with, India’s wider energy transition. Regulations rooted in the local context and mindful of the power sector ecosystem as a whole will be critical to maintaining India’s attractiveness as an investment destination for hyperscalers, while ensuring that the country does not miss out on the AI boom.
Varun B R is a public policy researcher specialising in the power sector and serves as a Manager at Vasudha Foundation.
As India balances rising energy demands, climate risks, and its 2070 net-zero target, artificial intelligence offers a critical pathway for both climate adaptation and decarbonisation. The AI for CleanTech Transformation whitepaper emphasizes that successful integration requires strong public data foundations, leveraging platforms like the India Climate and Energy Dashboard (ICED) and the India Energy Stack (IES) to enable secure data access for grid modernisation and policymaking.
To minimize AI’s own environmental footprint, stakeholders must prioritize task-specific narrow AI over energy-intensive generative models. Translating these technologies into real-world impact requires user-centric deployment in operational areas like battery energy storage and electric logistics. Ultimately, AI must empower local communities, such as delivering personalised agro-meteorological advisories to smallholder farmers, while actively bridging the digital divide to ensure an inclusive, resilient clean energy transition.
