There is a number that should be on the desk of every sustainability officer, and technology CEO in the world: 950 terawatt-hours.
That is how much electricity the world’s data centres are projected to consume annually by 2030, according to the International Energy Agency’s most comprehensive analysis to date of the relationship between artificial intelligence and global energy systems. It represents roughly double what data centres consumed in 2025 — and it is, in the IEA’s own framing, approximately equivalent to the entire electricity consumption of Japan today.
AI is not a footnote to this story. It is the story.
The Scale of the Demand Shift
The IEA’s updated projections, published in April 2026, show electricity consumption from data centres roughly doubling from 485 TWh in 2025 to 950 TWh in 2030, accounting for around 3% of global electricity demand by that date.
But within that headline figure lies a more dramatic sub-trend. Electricity demand from data centres soared by 17% in 2025 alone — far outpacing overall global electricity demand growth of 3%. And AI-focused data centres grew even faster than that.
Electricity consumption from AI-focused data centres is projected to triple between 2025 and 2030 — growing at roughly 30% annually — while conventional server electricity consumption grows at a comparatively modest 9% per year.
The driver is not mysterious. More people are using AI, and energy-intensive applications — such as AI agents — are on the rise. Power consumption per AI task is declining rapidly, with efficiency improving at a rate the IEA describes as unprecedented in energy history. But the volume of tasks is growing faster than efficiency gains can offset.
In the United States alone, data centres account for nearly half of all electricity demand growth projected between now and 2030. By the end of the decade, the country is set to consume more electricity for data centres than for the production of aluminium, steel, cement, chemicals, and all other energy-intensive goods combined.
This is not a marginal infrastructure issue. It is a structural shift in how the world’s most advanced economies consume energy — and it is happening at a pace that grid infrastructure, renewable capacity build-out, and climate commitments are all struggling to match.
The Corporate Capex Behind the Numbers
The five largest technology companies — Amazon Web Services, Google, Meta, Microsoft, and Equinix — recorded combined capital expenditure exceeding $400 billion in 2025. That figure is expected to jump by a further 75% in 2026. To put that in perspective: the collective capex of just five technology companies is now larger than global investment in oil and natural gas production.
In 2024, Big Tech companies accounted for 43% of all clean energy power purchase agreements signed globally. PPA prices rose by an average of 35% in 2024, driven largely by this surge in procurement from large AI developers.
Tech companies were responsible for signing around 40% of all corporate renewable PPAs in 2025. In parallel, the pipeline of conditional offtake agreements between data centre operators and small modular reactor nuclear projects has grown from 25 GW at the end of 2024 to 45 GW today. Amazon has committed to build a small modular reactor facility in Washington state specifically to supply carbon-free energy to its data centres.
The clean energy procurement race is, in part, a direct consequence of data centre electricity demand. But it is also exposing a structural tension: the same AI infrastructure that is driving corporate investment in renewables is simultaneously outpacing the rate at which that renewable capacity can actually be built and connected to grids.
India: The Fastest-Growing Data Centre Market in Asia
If the global picture is striking, India’s trajectory is in a category of its own.
India’s AI market is expected to reach $20–22 billion by 2027, posting a compound annual growth rate of 30%. Despite hosting nearly 20% of the world’s data, India has just 3% of the global data centre capacity. That gap is now being closed at speed — and the energy implications are already becoming visible.
As of early 2026, India’s operational data centre capacity has crossed 1.5 GW. Over half of that is clustered in and around Mumbai — a city whose total power consumption hovers around 4 GW. Mumbai alone could see its data centre capacity expand to 3.2 GW in the coming years.
India’s data centre capacity increase is projected to be fivefold between 2025 and 2030 — growing much faster than in the United States. Power demand from AI-related data centres in India could reach 50 TWh annually by 2030.
The investment numbers confirm the scale of commitment. Google is investing $15 billion for its first AI hub in India. Microsoft is putting in $17.5 billion. Amazon is targeting up to $35 billion by 2030. The Adani Group has committed $100 billion by 2035, including a 5 GW renewable-powered hyperscale data centre platform built in collaboration with Google, Flipkart, and Microsoft, with campuses in Visakhapatnam, Noida, and Hyderabad.
TCS has established HyperVault, a dedicated subsidiary focused on gigawatt-scale AI-ready infrastructure for hyperscalers and AI companies, with total planned investments of up to ₹18,000 crore, supported by TPG Capital. NTT DATA is developing a $1.2 billion AI cluster in Hyderabad, and Digital Connexion is establishing an $11 billion campus in Andhra Pradesh.
The question this raises is immediate and unresolved: where does the power come from?
India’s Grid Under Pressure
High-density AI clusters require reliable electricity supply and large parcels of industrial land — two requirements increasingly difficult to secure in major urban regions. The rapid buildout of AI data centres stands to further stress India’s power infrastructure, already identified as a key bottleneck by industry analysts.
The Digital Personal Data Protection Act of 2023 has compelled multinational companies to establish local infrastructure, increasing demand by an estimated 1,800 MW by 2027 alone. Meanwhile, India’s renewable energy additions — while accelerating — must simultaneously serve industrial decarbonisation, EV charging growth, and residential electrification alongside the surging data centre load.
Currently, coal-fired electricity accounts for the largest share of energy consumption at Indian data centres, followed by renewables, natural gas, and nuclear power. Over time, renewables are expected to become the largest source of additional energy supply for data centres, with nuclear power increasingly relevant for hyperscale facilities. Google’s planned hyperscale campus in Andhra Pradesh is already designed around a renewable energy partnership with Adani, using a combination of solar and wind to power its AI workloads.
But designing for renewables and being powered by renewables in practice are different things. India’s renewable capacity additions need to run at approximately 60 GW per year through 2030 to meet the government’s 500 GW non-fossil target — a pace significantly above the 30 GW added in FY2025. Layering gigawatt-scale data centre demand on top of that target creates a compounding challenge.
The ESG Dimension: AI’s Energy Paradox
The tension at the heart of this story is not simply about energy supply. It is about climate commitments.
Despite driving higher electricity consumption, AI’s overall emissions impact could potentially be offset if the technology enables broader emissions reductions across other sectors — in areas like grid optimisation, industrial efficiency, climate modelling, and precision agriculture. The IEA’s analysis is explicit that this offset potential is real but not automatic. It depends on how AI is deployed and, critically, on what energy sources power it.
While data centres may account for around 3–4% of global electricity demand by 2030, their share of local demand can become overwhelming: 42% in Frankfurt, nearly 80% in Dublin. In India’s case, concentration in Mumbai — a grid already under stress — creates a version of this localised intensity risk that urban energy planners are only beginning to grapple with.
For India’s corporations filing under SEBI’s mandatory BRSR framework, and for global hyperscalers with net-zero pledges, the data centre energy question is now a disclosure risk, not just an operational one. The 45 GW pipeline of small modular reactor agreements between data centre operators and nuclear developers globally reflects the recognition that conventional renewable procurement is not sufficient at this scale or speed — and that new energy sources will be required.
What Needs to Happen
The IEA’s analysis is precise about the problem. It is less prescriptive about the solution — and rightly so, because the solution is genuinely complex.
For India, it involves several simultaneous tracks: accelerating renewable energy additions well beyond current trajectories; investing in grid infrastructure that can handle concentrated high-density loads in urban clusters; ensuring that data centre policy — including the government’s classification of data centres as essential infrastructure — is matched with energy planning frameworks that account for their full power demands; and creating incentive structures that push hyperscalers toward genuine renewable energy use, not just green procurement on paper.
For global technology companies, it requires treating energy accountability as a board-level issue — not a sustainability team issue. Microsoft’s January 2026 commitment to pay full electricity costs for its data centres and refuse local property tax reductions is one indication that the industry is beginning to internalise the community and grid impact of its infrastructure decisions.
The IEA’s base case projects global data centre electricity consumption reaching 945 TWh by 2030 and climbing further to 1,200 TWh by 2035. That trajectory is not a warning. It is already priced into investment decisions made today by the largest companies in the world.
The question is whether energy systems — in India and globally — can keep pace. The answer will shape not just the future of AI, but the credibility of every net-zero commitment made over the past decade.
