Electricity demand from global data centres is projected to more than double by 2030, reaching 945 terawatt-hours (TWh)—equivalent to Japan’s total electricity consumption today—driven largely by rapid growth in artificial intelligence (AI) deployment.
This is the central finding of the International Energy Agency’s (IEA) new report Energy and AI, which delivers the most comprehensive data-driven analysis to date on the nexus between energy systems and AI technologies.
According to the IEA, data centres—many of which are being retrofitted or built specifically for AI workloads—will be responsible for more than 20% of the electricity demand growth in advanced economies by the end of the decade. In the United States alone, nearly half of the projected growth in electricity demand through 2030 will be attributable to data centres. Notably, by 2030, the US is set to consume more electricity for data processing than for manufacturing all energy-intensive industrial goods, including aluminium, steel, cement, and chemicals combined.
The scale and speed of this shift are not without precedent, but they mark a significant departure from recent trends. In many advanced economies, electricity demand had plateaued or even declined due to efficiency improvements and industrial shifts. The growth of AI infrastructure—particularly high-density data centres and training clusters for large language models—is now reversing this trajectory.
Yet the implications go beyond demand volumes. The IEA highlights several risks and strategic considerations facing energy systems: grid stability, energy sourcing, mineral supply chains, and cyber resilience.
Renewables, Natural Gas to Shoulder Load, But Flexibility is Key
To meet the mounting demand, energy systems are expected to rely increasingly on cost-competitive sources, particularly renewables and natural gas. While solar and wind will play dominant roles in markets with established capacity and favourable policy frameworks, the report suggests natural gas will continue to serve as a balancing mechanism where grid flexibility or energy storage is lacking.
However, a rapid rise in demand from concentrated AI data hubs raises questions about localized grid strain. Grid expansion and reinforcement are likely to become bottlenecks in countries with ageing infrastructure, slow permitting processes, or limited investment in flexibility measures such as demand response and storage.
Data Centres and Critical Minerals
The report also offers first-of-its-kind estimates on the demand for critical minerals linked to AI infrastructure. Key components in servers, cooling systems, and backup power units rely on high-purity metals and rare earth elements, much of which are sourced from geographically concentrated supply chains. This raises strategic concerns for countries dependent on imports, particularly as geopolitical tensions affect access to key materials like gallium, rare earth magnets, and lithium.
AI-driven systems, which depend on high-performance computing and deep neural networks, further intensify the mineral requirements compared to traditional cloud computing operations. The IEA warns that unaddressed supply vulnerabilities could undermine deployment timelines for both AI and broader digital energy systems.
Energy Security: AI as Risk and Defence Mechanism
The report notes a growing dualism in AI’s role in energy security. On one hand, cyberattacks on energy infrastructure have tripled in four years—fueled by the same AI tools that allow threat actors to identify and exploit vulnerabilities more efficiently. On the other, AI is becoming indispensable for energy operators seeking to detect anomalies, manage distributed assets, and automate defensive responses.
Adoption of AI in grid management, predictive maintenance, and optimisation of renewable generation is already underway, and the IEA points to this as a critical offset to rising emissions. While AI workloads themselves may increase emissions, the overall net impact could be mitigated or reversed through AI’s contribution to emissions reductions elsewhere—provided adoption is widespread and aligned with efficiency strategies.
Governance and Infrastructure Lag Behind Technology Curve
The IEA underscores that policy and infrastructure readiness are not keeping pace with the technology curve. Many jurisdictions lack frameworks to track, predict, or regulate electricity consumption from AI-specific workloads. In response, the IEA will launch a new Observatory on Energy, AI and Data Centres to compile real-time, granular data on AI’s electricity usage and monitor emerging applications across energy systems.
The agency is also deploying an interactive AI agent—available alongside the report—to help stakeholders digest its findings and implications. This complements ongoing policy engagement, including recent IEA contributions to the AI Action Summit co-chaired by France and India, and the Global Conference on Energy and AI hosted in December 2024.
Although data centres are rapidly evolving to be more efficient, the pace of AI expansion is outstripping these gains. To manage growth sustainably, countries must accelerate investments in generation, reinforce transmission and distribution networks, and strengthen coordination between energy regulators, the tech sector, and industrial planners. The report emphasises the need for demand-side flexibility and incentives to improve data centre load management, particularly in regions with limited peaking capacity.
Stay updated on the latest in energy! Follow us on LinkedIn, Facebook, and X for real-time news and insights. Don’t miss out on exclusive interviews and webinars—subscribe to our YouTube channel today! Join our community and be part of the conversation shaping the future of energy.
