Data centres are projected to consume four times their current electricity usage by 2035, driven primarily by artificial intelligence infrastructure deployment, according to new industry forecasts reported by TechCrunch AI. The anticipated power demand would equal India’s total national electricity consumption, presenting unprecedented challenges for energy providers and regulators.
The projections arrive as hyperscale operators expand AI training facilities and inference infrastructure to meet surging demand for large language models and machine learning applications. Current data centre electricity consumption already accounts for approximately 1-2% of global power usage, a figure set to rise sharply as AI workloads require substantially more computational resources than traditional cloud services.
The energy intensity of AI operations stems from the computational requirements of training and running neural networks. A single large language model training run can consume as much electricity as hundreds of homes use annually, whilst inference operations—serving AI responses to users—demand continuous power at scale. As enterprises integrate AI capabilities across operations, the cumulative effect on grid infrastructure becomes material.
Market Implications and Winners
The power surge creates distinct market opportunities and pressures. Utility companies serving major data centre markets—including Virginia’s Loudoun County, Dublin, and Singapore—face immediate pressure to expand generation capacity and grid infrastructure. Energy infrastructure firms stand to benefit from multi-billion-pound investments in substations, transmission lines, and backup power systems.
Conversely, AI companies face mounting operational costs and potential capacity constraints. Electricity expenses, already a significant portion of data centre operating budgets, will intensify margin pressure on cloud providers and AI-as-a-service platforms. Operators unable to secure reliable power access may face geographical constraints on expansion, potentially reshaping the competitive landscape.
The projections also accelerate the business case for on-site generation and alternative energy sources. Tech giants including Microsoft, Google, and Amazon have already committed to renewable energy procurement, but the scale of projected demand exceeds current renewable capacity in key markets. Nuclear power, including small modular reactors, is receiving renewed attention from data centre operators seeking carbon-neutral baseload power.
Regulatory Response Takes Shape
Energy regulators in major markets are beginning to impose stricter requirements on new data centre developments. Ireland’s grid operator has limited new data centre connections in the Dublin area, whilst Singapore has maintained a moratorium on new facilities pending infrastructure upgrades. Similar constraints are emerging in other capacity-constrained markets.
The European Union’s energy efficiency directive now requires data centres to report power usage effectiveness metrics and waste heat recovery plans. These regulatory frameworks will likely expand as governments balance economic benefits from data centre investment against grid stability and climate commitments.
The power demand also intersects with national security considerations. Countries are increasingly viewing data centre capacity as strategic infrastructure, with implications for permitting, foreign investment screening, and energy allocation during shortages.
What to Monitor
Key indicators include utility companies’ capital expenditure announcements in data centre markets, which will signal whether energy infrastructure can keep pace with demand. Power purchase agreements between tech companies and energy providers will reveal pricing trends and supply constraints. Additionally, advances in chip efficiency and cooling technology could moderate consumption growth, making semiconductor roadmaps relevant to energy forecasting.
The trajectory of AI adoption will ultimately determine whether these projections materialise or prove conservative. If AI integration accelerates beyond current expectations, even a fourfold increase may underestimate future requirements, forcing a fundamental reassessment of how societies allocate electrical generation capacity between traditional uses and digital infrastructure.
